Clinical Toxicology Downloaded from informahealthcare.com by Cornell University on 12/28/14 For personal use only.
CLINICAL TOXICOLOGY 9( l), pp. 75-87 (1976)
Cadmium, Lead, and Copper Blood Levels in Normal Children
THOMAS J. SMITH, Ph.D. Department of Family and Community Medicine University of Utah Medical Center Salt Lake City, Utah ANTHONY R. TEMPLE, M.D. Department of Pediatrics University of Utah Medical Center Salt Lake City, Utah JAMES C. READING, Ph.D. Department of Family and Community Medicine University of Utah Medical Center Salt Lake City, Utah
T h e r e h a s been only one recently published study of concentrations of cadmium in pediatric blood samples using advanced analytical techniques [ 3 ] . In this study a comparison w a s made between the blood l e v e l s of cadmium and s e v e r a l other t r a c e m e t a l s in B r i t i s h children hospitalized for reasons unrelated to heavy metal toxicity and in another group of children hospitalized with a possible diagnosis of lead poisoning. However, there is a l s o available a n unpublished study 75 Copyright 0 1976 hy Marcel Dekker, Inc All Rights Reserved Nelther this work nor any part may he reproduced or transmitted in any form or by any means, electronic or mechanical, Including photocopying, microfilmmg, and recording, or by any information storage and retrieval system, without permission in writing from the publisher
SMITH, TEMPLE, AND READING
76
Clinical Toxicology Downloaded from informahealthcare.com by Cornell University on 12/28/14 For personal use only.
by the Texas State Board of Health, who examined blood levels of selected metals in a group of school children in Amarillo, Texas with known o r suspected exposure to airborne cadmium from a nearby zinc smelter [9], Because of the need for good normal values from which to make necessary clinical and environmental decisions, our study was undertaken to measure the "normal'' levels of cadmium, lead, and copper in a population of well children with no known expos u r e to cadmium, lead, o r copper. M E T H O D S AND M A T E R I A L S SamDle Collection Duplicate 5-ml blood samples were taken from a population of 60 children, 2 months to 13 y e a r s of age, hospitalized for elective surgery in the Latter-Day Saints P r i m a r y Children's Hospital in Salt Lake City, Utah. All were white, middle and upper middle c l a s s children. The population was divided into two sets of 30 children. The second set was sampled approximately one month after the first. After collection, the samples, including the duplicates, were randomized and sent to the laboratory of the Department of Environmental Sciences, American Smelting and Refining Company, for analysis. The results were reported by sample number, eliminating the possibility of comparing duplicate results. In preparation for this study, analytical determinations were made of cadmium, lead, copper, and zinc contamination in standard vacuumized blood collection tubes. Substantial amounts of zinc and cadmium contamination were found. After testing several brands, Venoject by Jintin-Terumo blood tubes were chosen for their low level of cadmium contamination. None of the blood tubes tested were without substantial zinc contamination. A study is currently in progress to evaluate this problem in m o r e breadth and detail. SamPle Analvsis Each blood tube was weighed to the nearest milligram before and after the sample was removed to obtain a sample size. The whole blood was digested with nitric and perchloric acids down to a volume of 2 to 3 ml of a nearly colorless solution, the residue taken up in deionized distilled water and adjusted to a final volume of 25 ml. The final solution contained approximately 1%nitric acid. Lead and copper
77
Clinical Toxicology Downloaded from informahealthcare.com by Cornell University on 12/28/14 For personal use only.
TRACE METAL BLOOD LEVELS
analyses were performed on this solution. The cadmium w a s extracted from a 20-ml aliquot of this solution with dithizone in chloroform. Before extraction, 5 ml of ammonium citrate buffer was added and the pH was adjusted to 8.5 with ammonium hydroxide. The dithizone- chloroform phase containing the cadmium was digested with nitric acid to destroy the organic complex, and the final solution made to 10 m l with deionized distilled water. The final solution contained approximately 1%nitric acid. The concentrations of cadmium, lead, and copper in solution were determined with a specially designed, long path, atomic absorption spectrometer. This instrument has an electrically heated 25-cm ceramic tube through which the sample is aspirated in a hydrogen flame. The light path of the hollow cathode lamp p a s s e s through the center of this tube, The 1%absorbance concentrations for this instrument were 0.01 pg Pb/ml, 0,001 pg Cd/ml, and 0.1 pg Cu/ml, where a l l of these elements were in solutions of 1%HN03. These sensitivities correspond to detection limits of 0.2 pg Cd/100 gm, 2 pg Pb/ 100 gm, and 20 pg Cu/lOO gm of blood, respectively, f o r each of the elements. Corrections for salt scatter effects were made by observing the solution absorbance on a wavelength close to the trace element wavelength, but where the element does not absorb. The elemental concentrations were determined from standard curves. These curves were prepared f o r each batch of samples and were prepared from two s e t s of standards: (a) standard solutions of cadmium, lead, and copper in 1%nitric acid, and (b) increments of standard cadmium solution added to separatory funnels and extracted with dithizone in the same manner as the samples of blood were extracted. The standard curves provided a check on the recoveries of added elements and also a check on the possible contamination in routine analyses. Recovery studies using whole blood with added cadmium, lead, and copper were conducted to ensure that the added elements were not substantially lost in either the digestion step o r in extraction. This does not rule out the possibility that chemical species containing the element in the blood may be lost during the digestion. Data Analysis A random effects one-way analysis of variance was performed, in which each child was considered a treatment, in o r d e r to calculate the within and between subject variability [ 101. In this analysis the expected value for the mean square for e r r o r is u (within subject
e
variance) and the expected value of the mean square between children
SMITH, TEMPLE, AND READING
Clinical Toxicology Downloaded from informahealthcare.com by Cornell University on 12/28/14 For personal use only.
78
i s u + Zu (where u is the variance between children). The anlye a a sis of variance will provide estimates of both u and ua2. e Correlation coefficients were calculated to determine the strength of relationships between the metal levels and between each metal level and age. The average of each subject's duplicate blood metal concentrations were used in the correlation coefficient calculations. In order to represent the boundaries of normal values found in these data, tolerance intervals were formed. A tolerance interval is an interval which covers a specified proportion of the population (P) with a given probability (1 - (u). For example, if P = .95 and 1 - Q = .95, then the tolerance interval will contain 95% of the population (in this case blood values) with probability .95. In practical terms, this means that if a blood sample were taken from a member of a normal population, then it is highly certain that the blood value would have only one chance in 20 of lying outside the tolerance limits. A tolerance interval analysis was chosen instead of the more common confidence interval analysis because a confidence interval only gives information about the mean of a population of values; it tells nothing about the individual values. Since it is assumed that the children tested were normal, we a r e interested in information about the blood values of the normal population as a whole, not just the mean. This approach also has utility in deciding if a given value is part of the normal population o r is abnormal. The tolerance intervals given in this paper are for two values of P (95% and 99% of the population of blood values) and two values of probability 1 - (Y (.95 and .99). Hence, four tolerance intervals were calculated for each metal, so the reader may choose the one which he feels is most appropriate. A tolerance interval is of the form * Ks, where K is determined from a table [4] and s is an estimate of the appropriate standard deviation. The standard deviation used in this case was an estimate of the standard deviation of an individual blood measurement (u;),
x
An estimate of u;
is obtained from the sum of the between and within
subject variances calculated in the analysis of variance, i.e., 6:
=
+ 6 ', Then the square root of this estimate gives an estimate e a of the standard deviation for an individual observation. This estimate was used in the calculation of the tolerance intervals, i.e., s = bI Tolerance limits are only valid for normally distributed data. 1 2
u
.
D'Agostino's test of normality was used to verify normality [ 21. Several transformations were used on the cadmium data to try and produce normality ["I.
TRACE METAL BLOOD LEVELS
79
RESULTS
Clinical Toxicology Downloaded from informahealthcare.com by Cornell University on 12/28/14 For personal use only.
Analytical Method The reproducibility and metal recovery for metals in whole blood is shown in Table 1. All of the metals except cadmium were recovered essentially 100% from spiked blood samples. The recovery of cadmium was significantly l e s s (t-test on set means, p < .01)than loo%, averaging 7970. These results demonstrate the suitability of the method for measuring cadmium, lead, and copper in the normal blood ranges, The reproducibility was quite good on replicate blood samples. It w a s found that the variability in duplicate blood samples from the subjects was greater than that observed in the laboratory on replicate samples from a pooled blood source. Table 2 shows the results of a n analysis of variance for the children's blood values for each of the three metals tested [ l o ] . Even though consecutive tubes of blood were drawn from each subject, more variation was seen in the test subjects' data (within subject variation) than in the multiple replicates in laboratory tests of the analytical procedure (Table 1). Some difficulty was experienced with the cadmium extractions during the processing of the f i r s t set of samples, resulting in unsatisfactory analyses. As a result of this, no values are reported for the Set I cadmium analyses. These problems were corrected before Set II was analyzed. Blood L e v e l s in Children The age distribution of the test subjects is shown in Fig. 1. T h e children ranged in age from 2 months to 13 years, with an average age of 4.9 years. The distribution was skewed to the upper age bracket. The data on the metal levels in whole blood for these children are summarized in Table 3. Cadmium levels averaged 0.66 pg/lOO gm of whole blood, with a standard deviation of 0.25 pg/lOO gm. No significant relationships were found between cadmium and age or the other metal concentrations (Table 4). A graph of the distribution of cadmium blood values is shown in Fig. 2. Lead levels averaged 15.7 pg/lOO gm, with a standard deviation of 6.33. The distribution of lead levels was slightly skewed to the l a r g e r values with a maximum level of 37 p g / l O O gm. The distribution of blood lead values is shown in Fig. 3. Blood lead values were not significantly correlated with any of the other metals o r age.
82 156
pg/lOO gm added
21 64
pooled blood
pooled blood pg/lOO gm added
0.57 1.36
Mean, pg/lOO gm
7.1
2.9
2.4 4.4
0.069 0.181
Standard deviation (S.D.)
aSignificantly l e s s than 100% recovery at the p < .01 level.
80
Copper
Lead 40
Cadmium pooled blood 1.0 */lo0 gm added
Metal
4.6
3.5
11 6.9
12 13
Relative S.D., %
92.5
3
107.3
-
79.0a
-
Per cent recovery
3
8 11
6 11
Number replicates
TABLE 1. Precision and Recovery for Metals in Whole Blood
Clinical Toxicology Downloaded from informahealthcare.com by Cornell University on 12/28/14 For personal use only.
M
r
Ew
c3
3:
cn
3
0
43
81
TRACE METAL BLOOD LEVELS TABLE 2. Analysis of Variance for Paired Blood Samples Metal
Number tested
Lead
26
59
Copper 59
0.66
15.7
Standard deviation within subjects ( 6 )
0.198
5.2
30.7
Standard deviation among subjects (6 )
0.147
3.63
17.9
Mean Clinical Toxicology Downloaded from informahealthcare.com by Cornell University on 12/28/14 For personal use only.
Cadmium (Set 11)
e
a
L
0
n
5
z
li
123
4
2
1
2
3
4
Age (years)
FIG. 1. Age distribution of subjects.
TABLE 3. Concentrations of Metals in Whole Blood Mean, pg/lOO gm Cadmium Lead Copper
0.66 15.7 123
Range, pg/lOO gm
Standard deviation
Number samples
0.21-2.64
0.25
52
3.0-37.0
6.33
116
37-215
35.5
116
SMITH, TEMPLE, AND READING
82
TABLE 4. Correlation Coefficients f o r Age, Cadmium, Lead, and Copper against Each Other
Clinical Toxicology Downloaded from informahealthcare.com by Cornell University on 12/28/14 For personal use only.
Age
Cadmium
Lead
Copper
Age
1.0000
Cadmium
0.0058
1.0000
Lead
-0.2319
0.1825
1.0000
Copper
-0.2213
-0.0694
- 0.09 67
.2
4
.6
1.2
1.0
.8
1.4
1.0000
1.6
pq Cd 1100 grn whole blood
FIG, 2. Distribution of whole blood cadmium values.
401
10
20
h 30
90
50
pg Pb/100 gm whole blood
FIG. 3. Distribution of whole blood lead values.
60
TRACE METAL BLOOD LEVELS
83
Copper levels averaged 123 pg/lOO gm of whole blood, with a standard deviation of 35.5 and a range of 37-215 pg/lOO gm. No significant relationships were found between copper and any of the other variables. A graph showing the distribution for blood copper is shown in Fig. 4.
Clinical Toxicology Downloaded from informahealthcare.com by Cornell University on 12/28/14 For personal use only.
Tolerance Intervals Before the tolerance intervals were calculated, the data were tested for normality. Both the copper and lead data were determined to be very close to normally distributed; however, the cadmium data were badly nonnormal. Hence, it was necessary to transform the cadmium data. A log transformation was attempted and, although it improved things considerably, D' Agostino's test still found statistically significant nonnormality. Next, various transformations were attempted, as outlined by Johnson [ 7 ] . Even though transforinations were found which gave good normality, these transformations were not 1 to 1 and, as a result, it was impossible to take the calculated tolerance intervals on the transformed data and transform them back to the original units. For this reason, and also because there seemed to be some b a s i s for assuming a log normal distribution for the cadmium data, it was determined that the tolerance interval analysis would be done on the log transformed cadmium data, even though the normality assumption did not appear to be entirely fulfilled. The tolerance intervals a r e shown in Table 5. The intervals containing 95% of normal values with a probability of .95 were 0.22-1.70
401 r
0 L
30i 20
4
n
01 13
5
z
.
. 20
40
60
80
100
120
140
160
p9 C u / 1 0 0 9 m whole blood
FIG. 4. Distribution of whole blood copper values.
180
200
220
SMITH, TEMPLE, AND READING
84
Clinical Toxicology Downloaded from informahealthcare.com by Cornell University on 12/28/14 For personal use only.
TABLE 5. Tolerance Intervals for Cadmium, Lead, and Copper (1 - a )
P
Cadmium (Set 11)
Lead
Copper
.95
.95
(0.22,1.70)
(0.87,30.5)
(40.0,206)
.95
.99
(0.16,2.35)
(0,35.2)a
(13.9,232)
.99
.95
(0.19,1.94)
(0,31.6)a
(33.8,212)
.99
.99
(0.13,2.78)
(0,36.6)a
(5.75,240)
aZeros inserted for computed negative values. pg Cd/100 gm, 0.87-30.5 pg Pb/100 gm, and 40.0-206 pg Cu/lOO gm. It should be noted that the upper tolerance limits for cadmium are probably lower than they would be if an adequate transformation could have been found.
DISCUSSION
The primary metal under study in this survey was cadmium. Environmental cadmium has increasingly come under study as a suspected cause of a variety of disease states, including cancer, hypertension, and renal malfunction [5]. It is important to have reliable data on the level of this trace metal in normal populations in o r d e r to verify the differences observed in disease states and to provide background data needed to evaluate the impact of environmental contamination. There are very few data in the scientific literature regarding whole blood concentrations of cadmium and other trace metals in normal children. The recently published study by Delves et al. [3] was done on in-patients in a hospital. The data reported in this paper were obtained from well children that were voluntarily hospitalized for elective surgery, such as removal of tonsils and adenoids. Subjects were selected so that their blood levels were not suspected of being altered as a result of major disease processes. Previous studies of adults show a wide variation in average blood levels of cadmium, ranging from 0.05 pg/lOO gm to 2.0 pg/lOO gm of whole blood. Friberg et al. [6] attribute much of this variation to the use of different analytical techniques, analytical difficulties with low
Clinical Toxicology Downloaded from informahealthcare.com by Cornell University on 12/28/14 For personal use only.
TRACE METAL BLOOD LEVELS
85
cadmium concentrations, and from probable contamination of samples. We found that standard vacuumized blood collection tubes have appreciable cadmium contamination, largely from cadmium in silicone rubber stoppers, a finding which highlights the difficulty in obtaining reliable blood data. In light of past difficulties, great c a r e was exercised in evaluating the analytical method and in collecting samples. The reproducibility of the method was evaluated on replicate samples from a pooled blood source and by collecting duplicate blood samples from the subjects. We found more variability in the measurements from a given subject than within the sample population (between subjects), indicating the sample collection process and the characteristics of the analytical method are all sources of the within-subject variability. Not a l l of the cadmium was recovered with our analytical method. It was found that, on the average, '79% of a known amount of added cadmium w a s carried through to the final analysis in our control samples. The extraction step is the most likely source of this loss. The mean level of cadmium in this population, 0.66 pg/lOO gm, was higher than that reported by Delves et al. [3] for both their control subjects, 0.46 pg/lOO gm, and their possible lead poisoning cases, 0.53 pg/lOO gm. One reason that our r e s u l t s may be higher is that, even though the blood tube contamination level was v e r y low, it would still elevate the observed blood levels slightly. Second, their analytical method differs from the combined extraction-atomic absorption technique used here. They used a flash vaporization of a l l cadmium in a microsample that h a s been extracted after a digestion procedure. T h i s technique may give low results if either the digestion o r the extraction is incomplete and some of the cadmium remains in the partially digested sample material. Spiked blood samples can indicate complete recovery in this situation because the added cadmium is not bound to the red cells or other blood proteins in the same fashion a s endogenous cadmium. The technique used in the present study h a s a robust digestion procedure which will oxidize all organic matter prior to extraction. Cadmium has been shown to have a strong tendency to accumulate in the kidneys and liver with age [5]. Depending on the amount of exchange between the body s t o r e s of cadmium and the blood, this accumulation may be reflected in an increase in blood cadmium concentration with age. No correlation between increasing blood cadmium levels and age was found in our subjects. The second metal of interest was lead. The average blood.lead level found in these children was lower than those reported f o r normal child r e n in other studies [l]. T h e r e was good agreement between the data from this survey and the control group in the Delves et al. [3]
Clinical Toxicology Downloaded from informahealthcare.com by Cornell University on 12/28/14 For personal use only.
86
SMITH, TEMPLE,
AND READING
study. No c a s e s of an excessive lead burden were found. The highest value, 37 pg/lOO gm, was l e s s than the upper limit considered to be "normal," 40 pg/lOO gm [ E l . The upper tolerance limits found in these children a r e also in good agreement with the limit of 30 pg/lOO m l reported by Waldron [ 111. The children in this study were from middle to upper class families, without evidence of excessive environmental lead exposure. The authors' experience, information from the Intermountain Regional Poison Control Center, and other unpublished public health surveys of blood lead levels in children in Salt Lake City, Utah, are in agreement with o u r finding of no abnormal lead levels in these children. The third metal studied was copper. A relationship between copper and age was found by Delves et al. [3]. While there appears to be a decline in copper levels with age (Table 4), the correlation coefficient was not significant a t the 5% level. A regression line fitted to our data has nearly the same slope as that of Delves' control group data, but is approximately 15 pg/lOO gm higher. The qualitative relationship for copper levels declining with increasing age appears to be present. SUMMARY Cadmium, lead, and copper levels were measured in duplicate Whole blood samples from 60 apparently normal children, ranging in age from 2 months to 13 years, who were hospitalized for elective surgery. Metals were measured by atomic absorption spectroscopy after the samples were chemically oxidized and digested. Cadmium was concentrated by dithizone extraction before analysis. Blood cadmium averaged 0.66 pg/lOO gm, with a standard deviation of 0.25. No samples had cadmium concentrations l e s s than could be detectable. The average concentration of lead was 15.7 pg/lOO gm with a standard deviation of 6.33. Copper averaged 123 pg/lOO gm with a standard deviation of 35.5. Tolerance intervals were calculated for each metal in order to estimate the bounds of whole blood metal values in normal children. The intervals containing 95% of normal values with a probability of .95 were 0.22-1.70 pg Cd/100 gm, 0.87-30.5 pg P b / l O O gm, and 40.0-206 pg Cu/100 gm. ACKNOWLEDGMENTS The authors wish to thank Kenneth W. Nelson, Michael 0. Varner, Sandra Nackowski, and the chemists a t American Smelting and Refining
TRACE METAL BLOOD LEVELS
87
Company's Department of Environmental Sciences for their advice and assistance.
Clinical Toxicology Downloaded from informahealthcare.com by Cornell University on 12/28/14 For personal use only.
REFERENCES J. J. Chisolm, Jr., Chronic lead intoxication in children, Devel. Med. Child Neurol., 7, 529 (1965). R. B. D'Agostino, An-omnibus t e s t of normality f o r moderate and l a r g e sample size, Biometrika, 58, 341 (1971). H. T . Delves, B. E. Clayton, and J. Kcknell, Concentration of t r a c e m e t a l s in the blood of children, Brit. J. Prev. SOC. Med., 27, 100 (1973). W. J. Dixon and F. J. Massey, Jr., Introduction to Statistical Analysis, 3rd ed., McGraw-Hill, New York, 1969. D. F. Flick, H. F. Kraybill, and J. M. Dimitroff, Toxic effects of cadmium: a review, Environ. Res., 4, 71 (1971). L. Friberg, M. Piscator, and G. Nordberg, Cadmium in the Environment, Chemical Rubber Company Press, Cleveland, Ohio, 1971. N. L. Johnson, Systems of frequency c u r v e s generated by methods of translation, Biometrika, 36, 149 (1949). J. S. Lin-Fu, Lead Poisoning in C h i l z e n , Children's Bureau Publication No. 452- 1967, United States Dept. Health, Education, and Welfare, Social and Rehabilitation Service, U.S. Govt. Printing Office, Washington, D.C. (1965). Report, "An environmental impact study of the American Smelting and Refining Company's zinc s m e l t e r , Amarillo, Texas," Air Pollution Control Services, T e x a s State Department of Health, 1972. G. W. Snedecor and W. G. Cochran, Statistical Methods, 6th ed., Iowa State Univ. Press, 1969. H. A. Waldron, The blood lead threshold, Arch. Environ. Health, - 29, 271 (1974).
CLINICAL TOXICOLOGY 9( l), pp. 75-87 (1976)
Cadmium, Lead, and Copper Blood Levels in Normal Children
THOMAS J. SMITH, Ph.D. Department of Family and Community Medicine University of Utah Medical Center Salt Lake City, Utah ANTHONY R. TEMPLE, M.D. Department of Pediatrics University of Utah Medical Center Salt Lake City, Utah JAMES C. READING, Ph.D. Department of Family and Community Medicine University of Utah Medical Center Salt Lake City, Utah
T h e r e h a s been only one recently published study of concentrations of cadmium in pediatric blood samples using advanced analytical techniques [ 3 ] . In this study a comparison w a s made between the blood l e v e l s of cadmium and s e v e r a l other t r a c e m e t a l s in B r i t i s h children hospitalized for reasons unrelated to heavy metal toxicity and in another group of children hospitalized with a possible diagnosis of lead poisoning. However, there is a l s o available a n unpublished study 75 Copyright 0 1976 hy Marcel Dekker, Inc All Rights Reserved Nelther this work nor any part may he reproduced or transmitted in any form or by any means, electronic or mechanical, Including photocopying, microfilmmg, and recording, or by any information storage and retrieval system, without permission in writing from the publisher
SMITH, TEMPLE, AND READING
76
Clinical Toxicology Downloaded from informahealthcare.com by Cornell University on 12/28/14 For personal use only.
by the Texas State Board of Health, who examined blood levels of selected metals in a group of school children in Amarillo, Texas with known o r suspected exposure to airborne cadmium from a nearby zinc smelter [9], Because of the need for good normal values from which to make necessary clinical and environmental decisions, our study was undertaken to measure the "normal'' levels of cadmium, lead, and copper in a population of well children with no known expos u r e to cadmium, lead, o r copper. M E T H O D S AND M A T E R I A L S SamDle Collection Duplicate 5-ml blood samples were taken from a population of 60 children, 2 months to 13 y e a r s of age, hospitalized for elective surgery in the Latter-Day Saints P r i m a r y Children's Hospital in Salt Lake City, Utah. All were white, middle and upper middle c l a s s children. The population was divided into two sets of 30 children. The second set was sampled approximately one month after the first. After collection, the samples, including the duplicates, were randomized and sent to the laboratory of the Department of Environmental Sciences, American Smelting and Refining Company, for analysis. The results were reported by sample number, eliminating the possibility of comparing duplicate results. In preparation for this study, analytical determinations were made of cadmium, lead, copper, and zinc contamination in standard vacuumized blood collection tubes. Substantial amounts of zinc and cadmium contamination were found. After testing several brands, Venoject by Jintin-Terumo blood tubes were chosen for their low level of cadmium contamination. None of the blood tubes tested were without substantial zinc contamination. A study is currently in progress to evaluate this problem in m o r e breadth and detail. SamPle Analvsis Each blood tube was weighed to the nearest milligram before and after the sample was removed to obtain a sample size. The whole blood was digested with nitric and perchloric acids down to a volume of 2 to 3 ml of a nearly colorless solution, the residue taken up in deionized distilled water and adjusted to a final volume of 25 ml. The final solution contained approximately 1%nitric acid. Lead and copper
77
Clinical Toxicology Downloaded from informahealthcare.com by Cornell University on 12/28/14 For personal use only.
TRACE METAL BLOOD LEVELS
analyses were performed on this solution. The cadmium w a s extracted from a 20-ml aliquot of this solution with dithizone in chloroform. Before extraction, 5 ml of ammonium citrate buffer was added and the pH was adjusted to 8.5 with ammonium hydroxide. The dithizone- chloroform phase containing the cadmium was digested with nitric acid to destroy the organic complex, and the final solution made to 10 m l with deionized distilled water. The final solution contained approximately 1%nitric acid. The concentrations of cadmium, lead, and copper in solution were determined with a specially designed, long path, atomic absorption spectrometer. This instrument has an electrically heated 25-cm ceramic tube through which the sample is aspirated in a hydrogen flame. The light path of the hollow cathode lamp p a s s e s through the center of this tube, The 1%absorbance concentrations for this instrument were 0.01 pg Pb/ml, 0,001 pg Cd/ml, and 0.1 pg Cu/ml, where a l l of these elements were in solutions of 1%HN03. These sensitivities correspond to detection limits of 0.2 pg Cd/100 gm, 2 pg Pb/ 100 gm, and 20 pg Cu/lOO gm of blood, respectively, f o r each of the elements. Corrections for salt scatter effects were made by observing the solution absorbance on a wavelength close to the trace element wavelength, but where the element does not absorb. The elemental concentrations were determined from standard curves. These curves were prepared f o r each batch of samples and were prepared from two s e t s of standards: (a) standard solutions of cadmium, lead, and copper in 1%nitric acid, and (b) increments of standard cadmium solution added to separatory funnels and extracted with dithizone in the same manner as the samples of blood were extracted. The standard curves provided a check on the recoveries of added elements and also a check on the possible contamination in routine analyses. Recovery studies using whole blood with added cadmium, lead, and copper were conducted to ensure that the added elements were not substantially lost in either the digestion step o r in extraction. This does not rule out the possibility that chemical species containing the element in the blood may be lost during the digestion. Data Analysis A random effects one-way analysis of variance was performed, in which each child was considered a treatment, in o r d e r to calculate the within and between subject variability [ 101. In this analysis the expected value for the mean square for e r r o r is u (within subject
e
variance) and the expected value of the mean square between children
SMITH, TEMPLE, AND READING
Clinical Toxicology Downloaded from informahealthcare.com by Cornell University on 12/28/14 For personal use only.
78
i s u + Zu (where u is the variance between children). The anlye a a sis of variance will provide estimates of both u and ua2. e Correlation coefficients were calculated to determine the strength of relationships between the metal levels and between each metal level and age. The average of each subject's duplicate blood metal concentrations were used in the correlation coefficient calculations. In order to represent the boundaries of normal values found in these data, tolerance intervals were formed. A tolerance interval is an interval which covers a specified proportion of the population (P) with a given probability (1 - (u). For example, if P = .95 and 1 - Q = .95, then the tolerance interval will contain 95% of the population (in this case blood values) with probability .95. In practical terms, this means that if a blood sample were taken from a member of a normal population, then it is highly certain that the blood value would have only one chance in 20 of lying outside the tolerance limits. A tolerance interval analysis was chosen instead of the more common confidence interval analysis because a confidence interval only gives information about the mean of a population of values; it tells nothing about the individual values. Since it is assumed that the children tested were normal, we a r e interested in information about the blood values of the normal population as a whole, not just the mean. This approach also has utility in deciding if a given value is part of the normal population o r is abnormal. The tolerance intervals given in this paper are for two values of P (95% and 99% of the population of blood values) and two values of probability 1 - (Y (.95 and .99). Hence, four tolerance intervals were calculated for each metal, so the reader may choose the one which he feels is most appropriate. A tolerance interval is of the form * Ks, where K is determined from a table [4] and s is an estimate of the appropriate standard deviation. The standard deviation used in this case was an estimate of the standard deviation of an individual blood measurement (u;),
x
An estimate of u;
is obtained from the sum of the between and within
subject variances calculated in the analysis of variance, i.e., 6:
=
+ 6 ', Then the square root of this estimate gives an estimate e a of the standard deviation for an individual observation. This estimate was used in the calculation of the tolerance intervals, i.e., s = bI Tolerance limits are only valid for normally distributed data. 1 2
u
.
D'Agostino's test of normality was used to verify normality [ 21. Several transformations were used on the cadmium data to try and produce normality ["I.
TRACE METAL BLOOD LEVELS
79
RESULTS
Clinical Toxicology Downloaded from informahealthcare.com by Cornell University on 12/28/14 For personal use only.
Analytical Method The reproducibility and metal recovery for metals in whole blood is shown in Table 1. All of the metals except cadmium were recovered essentially 100% from spiked blood samples. The recovery of cadmium was significantly l e s s (t-test on set means, p < .01)than loo%, averaging 7970. These results demonstrate the suitability of the method for measuring cadmium, lead, and copper in the normal blood ranges, The reproducibility was quite good on replicate blood samples. It w a s found that the variability in duplicate blood samples from the subjects was greater than that observed in the laboratory on replicate samples from a pooled blood source. Table 2 shows the results of a n analysis of variance for the children's blood values for each of the three metals tested [ l o ] . Even though consecutive tubes of blood were drawn from each subject, more variation was seen in the test subjects' data (within subject variation) than in the multiple replicates in laboratory tests of the analytical procedure (Table 1). Some difficulty was experienced with the cadmium extractions during the processing of the f i r s t set of samples, resulting in unsatisfactory analyses. As a result of this, no values are reported for the Set I cadmium analyses. These problems were corrected before Set II was analyzed. Blood L e v e l s in Children The age distribution of the test subjects is shown in Fig. 1. T h e children ranged in age from 2 months to 13 years, with an average age of 4.9 years. The distribution was skewed to the upper age bracket. The data on the metal levels in whole blood for these children are summarized in Table 3. Cadmium levels averaged 0.66 pg/lOO gm of whole blood, with a standard deviation of 0.25 pg/lOO gm. No significant relationships were found between cadmium and age or the other metal concentrations (Table 4). A graph of the distribution of cadmium blood values is shown in Fig. 2. Lead levels averaged 15.7 pg/lOO gm, with a standard deviation of 6.33. The distribution of lead levels was slightly skewed to the l a r g e r values with a maximum level of 37 p g / l O O gm. The distribution of blood lead values is shown in Fig. 3. Blood lead values were not significantly correlated with any of the other metals o r age.
82 156
pg/lOO gm added
21 64
pooled blood
pooled blood pg/lOO gm added
0.57 1.36
Mean, pg/lOO gm
7.1
2.9
2.4 4.4
0.069 0.181
Standard deviation (S.D.)
aSignificantly l e s s than 100% recovery at the p < .01 level.
80
Copper
Lead 40
Cadmium pooled blood 1.0 */lo0 gm added
Metal
4.6
3.5
11 6.9
12 13
Relative S.D., %
92.5
3
107.3
-
79.0a
-
Per cent recovery
3
8 11
6 11
Number replicates
TABLE 1. Precision and Recovery for Metals in Whole Blood
Clinical Toxicology Downloaded from informahealthcare.com by Cornell University on 12/28/14 For personal use only.
M
r
Ew
c3
3:
cn
3
0
43
81
TRACE METAL BLOOD LEVELS TABLE 2. Analysis of Variance for Paired Blood Samples Metal
Number tested
Lead
26
59
Copper 59
0.66
15.7
Standard deviation within subjects ( 6 )
0.198
5.2
30.7
Standard deviation among subjects (6 )
0.147
3.63
17.9
Mean Clinical Toxicology Downloaded from informahealthcare.com by Cornell University on 12/28/14 For personal use only.
Cadmium (Set 11)
e
a
L
0
n
5
z
li
123
4
2
1
2
3
4
Age (years)
FIG. 1. Age distribution of subjects.
TABLE 3. Concentrations of Metals in Whole Blood Mean, pg/lOO gm Cadmium Lead Copper
0.66 15.7 123
Range, pg/lOO gm
Standard deviation
Number samples
0.21-2.64
0.25
52
3.0-37.0
6.33
116
37-215
35.5
116
SMITH, TEMPLE, AND READING
82
TABLE 4. Correlation Coefficients f o r Age, Cadmium, Lead, and Copper against Each Other
Clinical Toxicology Downloaded from informahealthcare.com by Cornell University on 12/28/14 For personal use only.
Age
Cadmium
Lead
Copper
Age
1.0000
Cadmium
0.0058
1.0000
Lead
-0.2319
0.1825
1.0000
Copper
-0.2213
-0.0694
- 0.09 67
.2
4
.6
1.2
1.0
.8
1.4
1.0000
1.6
pq Cd 1100 grn whole blood
FIG, 2. Distribution of whole blood cadmium values.
401
10
20
h 30
90
50
pg Pb/100 gm whole blood
FIG. 3. Distribution of whole blood lead values.
60
TRACE METAL BLOOD LEVELS
83
Copper levels averaged 123 pg/lOO gm of whole blood, with a standard deviation of 35.5 and a range of 37-215 pg/lOO gm. No significant relationships were found between copper and any of the other variables. A graph showing the distribution for blood copper is shown in Fig. 4.
Clinical Toxicology Downloaded from informahealthcare.com by Cornell University on 12/28/14 For personal use only.
Tolerance Intervals Before the tolerance intervals were calculated, the data were tested for normality. Both the copper and lead data were determined to be very close to normally distributed; however, the cadmium data were badly nonnormal. Hence, it was necessary to transform the cadmium data. A log transformation was attempted and, although it improved things considerably, D' Agostino's test still found statistically significant nonnormality. Next, various transformations were attempted, as outlined by Johnson [ 7 ] . Even though transforinations were found which gave good normality, these transformations were not 1 to 1 and, as a result, it was impossible to take the calculated tolerance intervals on the transformed data and transform them back to the original units. For this reason, and also because there seemed to be some b a s i s for assuming a log normal distribution for the cadmium data, it was determined that the tolerance interval analysis would be done on the log transformed cadmium data, even though the normality assumption did not appear to be entirely fulfilled. The tolerance intervals a r e shown in Table 5. The intervals containing 95% of normal values with a probability of .95 were 0.22-1.70
401 r
0 L
30i 20
4
n
01 13
5
z
.
. 20
40
60
80
100
120
140
160
p9 C u / 1 0 0 9 m whole blood
FIG. 4. Distribution of whole blood copper values.
180
200
220
SMITH, TEMPLE, AND READING
84
Clinical Toxicology Downloaded from informahealthcare.com by Cornell University on 12/28/14 For personal use only.
TABLE 5. Tolerance Intervals for Cadmium, Lead, and Copper (1 - a )
P
Cadmium (Set 11)
Lead
Copper
.95
.95
(0.22,1.70)
(0.87,30.5)
(40.0,206)
.95
.99
(0.16,2.35)
(0,35.2)a
(13.9,232)
.99
.95
(0.19,1.94)
(0,31.6)a
(33.8,212)
.99
.99
(0.13,2.78)
(0,36.6)a
(5.75,240)
aZeros inserted for computed negative values. pg Cd/100 gm, 0.87-30.5 pg Pb/100 gm, and 40.0-206 pg Cu/lOO gm. It should be noted that the upper tolerance limits for cadmium are probably lower than they would be if an adequate transformation could have been found.
DISCUSSION
The primary metal under study in this survey was cadmium. Environmental cadmium has increasingly come under study as a suspected cause of a variety of disease states, including cancer, hypertension, and renal malfunction [5]. It is important to have reliable data on the level of this trace metal in normal populations in o r d e r to verify the differences observed in disease states and to provide background data needed to evaluate the impact of environmental contamination. There are very few data in the scientific literature regarding whole blood concentrations of cadmium and other trace metals in normal children. The recently published study by Delves et al. [3] was done on in-patients in a hospital. The data reported in this paper were obtained from well children that were voluntarily hospitalized for elective surgery, such as removal of tonsils and adenoids. Subjects were selected so that their blood levels were not suspected of being altered as a result of major disease processes. Previous studies of adults show a wide variation in average blood levels of cadmium, ranging from 0.05 pg/lOO gm to 2.0 pg/lOO gm of whole blood. Friberg et al. [6] attribute much of this variation to the use of different analytical techniques, analytical difficulties with low
Clinical Toxicology Downloaded from informahealthcare.com by Cornell University on 12/28/14 For personal use only.
TRACE METAL BLOOD LEVELS
85
cadmium concentrations, and from probable contamination of samples. We found that standard vacuumized blood collection tubes have appreciable cadmium contamination, largely from cadmium in silicone rubber stoppers, a finding which highlights the difficulty in obtaining reliable blood data. In light of past difficulties, great c a r e was exercised in evaluating the analytical method and in collecting samples. The reproducibility of the method was evaluated on replicate samples from a pooled blood source and by collecting duplicate blood samples from the subjects. We found more variability in the measurements from a given subject than within the sample population (between subjects), indicating the sample collection process and the characteristics of the analytical method are all sources of the within-subject variability. Not a l l of the cadmium was recovered with our analytical method. It was found that, on the average, '79% of a known amount of added cadmium w a s carried through to the final analysis in our control samples. The extraction step is the most likely source of this loss. The mean level of cadmium in this population, 0.66 pg/lOO gm, was higher than that reported by Delves et al. [3] for both their control subjects, 0.46 pg/lOO gm, and their possible lead poisoning cases, 0.53 pg/lOO gm. One reason that our r e s u l t s may be higher is that, even though the blood tube contamination level was v e r y low, it would still elevate the observed blood levels slightly. Second, their analytical method differs from the combined extraction-atomic absorption technique used here. They used a flash vaporization of a l l cadmium in a microsample that h a s been extracted after a digestion procedure. T h i s technique may give low results if either the digestion o r the extraction is incomplete and some of the cadmium remains in the partially digested sample material. Spiked blood samples can indicate complete recovery in this situation because the added cadmium is not bound to the red cells or other blood proteins in the same fashion a s endogenous cadmium. The technique used in the present study h a s a robust digestion procedure which will oxidize all organic matter prior to extraction. Cadmium has been shown to have a strong tendency to accumulate in the kidneys and liver with age [5]. Depending on the amount of exchange between the body s t o r e s of cadmium and the blood, this accumulation may be reflected in an increase in blood cadmium concentration with age. No correlation between increasing blood cadmium levels and age was found in our subjects. The second metal of interest was lead. The average blood.lead level found in these children was lower than those reported f o r normal child r e n in other studies [l]. T h e r e was good agreement between the data from this survey and the control group in the Delves et al. [3]
Clinical Toxicology Downloaded from informahealthcare.com by Cornell University on 12/28/14 For personal use only.
86
SMITH, TEMPLE,
AND READING
study. No c a s e s of an excessive lead burden were found. The highest value, 37 pg/lOO gm, was l e s s than the upper limit considered to be "normal," 40 pg/lOO gm [ E l . The upper tolerance limits found in these children a r e also in good agreement with the limit of 30 pg/lOO m l reported by Waldron [ 111. The children in this study were from middle to upper class families, without evidence of excessive environmental lead exposure. The authors' experience, information from the Intermountain Regional Poison Control Center, and other unpublished public health surveys of blood lead levels in children in Salt Lake City, Utah, are in agreement with o u r finding of no abnormal lead levels in these children. The third metal studied was copper. A relationship between copper and age was found by Delves et al. [3]. While there appears to be a decline in copper levels with age (Table 4), the correlation coefficient was not significant a t the 5% level. A regression line fitted to our data has nearly the same slope as that of Delves' control group data, but is approximately 15 pg/lOO gm higher. The qualitative relationship for copper levels declining with increasing age appears to be present. SUMMARY Cadmium, lead, and copper levels were measured in duplicate Whole blood samples from 60 apparently normal children, ranging in age from 2 months to 13 years, who were hospitalized for elective surgery. Metals were measured by atomic absorption spectroscopy after the samples were chemically oxidized and digested. Cadmium was concentrated by dithizone extraction before analysis. Blood cadmium averaged 0.66 pg/lOO gm, with a standard deviation of 0.25. No samples had cadmium concentrations l e s s than could be detectable. The average concentration of lead was 15.7 pg/lOO gm with a standard deviation of 6.33. Copper averaged 123 pg/lOO gm with a standard deviation of 35.5. Tolerance intervals were calculated for each metal in order to estimate the bounds of whole blood metal values in normal children. The intervals containing 95% of normal values with a probability of .95 were 0.22-1.70 pg Cd/100 gm, 0.87-30.5 pg P b / l O O gm, and 40.0-206 pg Cu/100 gm. ACKNOWLEDGMENTS The authors wish to thank Kenneth W. Nelson, Michael 0. Varner, Sandra Nackowski, and the chemists a t American Smelting and Refining
TRACE METAL BLOOD LEVELS
87
Company's Department of Environmental Sciences for their advice and assistance.
Clinical Toxicology Downloaded from informahealthcare.com by Cornell University on 12/28/14 For personal use only.
REFERENCES J. J. Chisolm, Jr., Chronic lead intoxication in children, Devel. Med. Child Neurol., 7, 529 (1965). R. B. D'Agostino, An-omnibus t e s t of normality f o r moderate and l a r g e sample size, Biometrika, 58, 341 (1971). H. T . Delves, B. E. Clayton, and J. Kcknell, Concentration of t r a c e m e t a l s in the blood of children, Brit. J. Prev. SOC. Med., 27, 100 (1973). W. J. Dixon and F. J. Massey, Jr., Introduction to Statistical Analysis, 3rd ed., McGraw-Hill, New York, 1969. D. F. Flick, H. F. Kraybill, and J. M. Dimitroff, Toxic effects of cadmium: a review, Environ. Res., 4, 71 (1971). L. Friberg, M. Piscator, and G. Nordberg, Cadmium in the Environment, Chemical Rubber Company Press, Cleveland, Ohio, 1971. N. L. Johnson, Systems of frequency c u r v e s generated by methods of translation, Biometrika, 36, 149 (1949). J. S. Lin-Fu, Lead Poisoning in C h i l z e n , Children's Bureau Publication No. 452- 1967, United States Dept. Health, Education, and Welfare, Social and Rehabilitation Service, U.S. Govt. Printing Office, Washington, D.C. (1965). Report, "An environmental impact study of the American Smelting and Refining Company's zinc s m e l t e r , Amarillo, Texas," Air Pollution Control Services, T e x a s State Department of Health, 1972. G. W. Snedecor and W. G. Cochran, Statistical Methods, 6th ed., Iowa State Univ. Press, 1969. H. A. Waldron, The blood lead threshold, Arch. Environ. Health, - 29, 271 (1974).
