| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Articles |
1
Wisconsin State Laboratory of Hygiene, Toxicology Section, 2601 Agriculture Dr., P.O. Box 7996, Madison, WI 53707-7996.
2
Centers for Disease Control and Prevention, National
Center for Environ-mental Health, Environmental Health Lab
Services, 4770 Buford Hwy., Atlanta, GA 30341.
3
Based in part on the data presented here, CDC has revised its position regarding the use of FP-based methods by CDC childhood lead poisoning grantees. This new position is described in a letter to those grantees dated February 25, 1999.
a Author for correspondence. Fax 608-224-6259; e-mail nvstox{at}mail.slh.wisc.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Methods: We describe a suitable PT scheme and evaluate FP laboratory performance based on program results. Monthly testing events consisting of five FP specimens were provided to six participating laboratories. Results were evaluated against target values determined by referee laboratories.
Results: Preliminary FP laboratory results showed poor agreement with specimen target values, exhibiting a mean absolute bias of 0.29 µmol/L (5.9 µg/dL). Five of six participating laboratories demonstrated significant improvement in later testing events, with bias decreasing to 0.12 µmol/L (2.5 µg/dL). Performance varied widely between the participating laboratories and appeared to be method dependent. When evaluated using CLIA blood lead acceptability criteria, the proportion of acceptable individual specimen results (n = 35) ranged from 54% to 100%. On a testing event basis (n = 7), the proportion of acceptable events ranged from 29% to 100%.
Conclusions: A suitable FP PT program now exists to capably assist and monitor FP laboratories. Based on overt PT results, properly utilized FP testing methods can accurately measure blood lead concentration.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The use of blood spots dried on filter paper (FP)1 has been investigated as an alternative to liquid blood for lead measurement. Possible advantages of FP sampling include greater ease of collection, simpler packaging, lower shipment cost, and long-term specimen stability. Methods utilizing FP samples were first described in the 1970s as a modification of the Delves cup flame atomic absorption (AA) method for liquid blood (3)(4)(5)(6) but were never widely accepted because of concerns about contamination and other technical issues (7)(8)(9).
Recent publications have described several new FPsample testing methods that use a variety of analytic approaches. Methods using electrothermal atomization (graphite furnace) AA measurement after sample extraction with solutions of HNO3 (10)(11) or (NH4)2HPO4 (12), inductively coupled plasma mass spectrometry measurement after HNO3 extraction with yttrium internal standard (13), and modifications of the original Delves cup AA procedure incorporating a preashing step (14)(15) have all been reported since 1991. These recent publications largely conclude that FP collection is a viable alternative for blood lead screening purposes, and a few laboratories now routinely perform FP blood lead testing on pediatric specimens.
The reliability of FP methods has been questioned (16)(17)(18)(19), and continues to generate discussion (20)(21)(22).4 One major concern has been the absence of FP-based proficiency testing (PT) as required under the CLIA ‘88 laboratory regulations (23) and the associated external method validation data available through PT participation. Enlisting the cooperation of laboratories with FP testing expertise, we sought to establish a PT program that would be viable for all FP methods in current use. This report describes the characteristics of this pilot PT program and examines the results from the participating FP laboratories.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
blood donors and collection
Fresh human blood with physiologically bound lead was used for the
FP specimen matrix. Blood that was hemolyzed or to which a lead
solution had been added were reported to be unsuitable because of FP
blood spot characteristics that differ from fresh human blood
(10). Nonhuman blood was considered unsuitable for the same
reason. All procedures were approved by the University of
Wisconsin-Madison Center for Health Sciences Human Subjects Committee
and the CDC Internal Review Board, and are in accordance with the
Helsinki Declaration. Volunteer donors were recruited from a cohort of
mixed sex and race industrial workers with occupational lead exposure.
The cohort was monitored independently for their lead exposure in
compliance with Occupational Safety and Health Administration
guidelines. Six nonexposed adults were also recruited to provide
specimens with background lead concentrations.
After giving informed consent, each donor underwent a standard venipuncture by trained medical personnel, using equipment obtained from Becton Dickinson. Three 10-mL blood specimens were collected in tubes containing liquid EDTA anticoagulant. All blood collection containers used in the program were lot tested for the presence of lead by a WSLH adaptation of published protocols (20). Each volunteer donation was a discrete FP PT specimen. Blood from 45 donors was used in the study. The donors were questioned about medication use in the 24-h period before collection, and a "Yes" or "No" response was recorded. The blood specimens were then transported and stored at refrigerated temperatures.
specimen preparation
To minimize the possibility of environmental contamination,
specimen preparation steps were performed in a trace metal clean room
conservatively rated as Class 1000 (M4.5). The following steps were
performed within 30 h of blood collection. Contents of the three
tubes from an individual donor were combined and gently mixed in a 50-
to 100-mL beaker by a magnetic stirring device. Blood was then spotted
onto Schleicher & Schuell no. 903 filter paper sheets, which were
suspended horizontally in air by clips. Five of the six participating
laboratories received FP sheets containing three 50-µL spots. These
sheets were spotted using an Eppendorf positive displacement repeating
pipette with disposable tip assemblies. Laboratory A requested sheets
with five, and later six, spots of 30 µL each to comply with their
unique FP specimen submission requirements. This was done with an air
displacement 30-µL Eppendorf pipette with disposable tips. The
accuracy of pipettes used for FP spotting was verified before use. The
remainder of the blood was dispensed in 2-mL aliquots into 3-mL
additive-free evacuated blood tubes, using an Eppendorf positive
displacement repeating pipette. These aliquots were then provided to
referee laboratories for value assignment purposes.
After spotting, the FP sheets remained suspended horizontally to allow the blood spots to air dry. The sheets were dried for an average of 75 min (range, 58–103 min) and then returned to zip-lock plastic bags for storage at room temperature. Subsequent distribution of the FP specimens to participants took place within 60 days. The aliquots of liquid blood were stored under refrigeration until distribution to reference laboratories for value assignment. These specimens were distributed within 1 week.
The FP sheets had designated handling and spotting areas marked on them. Other than the zip-lock bags the sheets were stored in, the spotting areas of the sheets did not come in contact with any surfaces during the spotting or drying process. Specimens from the first three testing events (October through December 1997) were spotted on blank FP sheets that had been provided by the individual participating laboratories. Although the sheets varied considerably in size and appearance, they shared the common characteristics of Schleicher & Schuell no. 903 composition, designated handling and spotting areas, and prior suitability testing by the participants. All other specimens were spotted on custom Schleicher & Schuell no. 903 sheets obtained for the program and designed to specifications suitable for the participants. Before use for PT purposes, these sheets were also evaluated for suitability by all of the participating laboratories.
blood specimen characterization
The blood specimens were visually inspected for hemolysis and
lipemia immediately before spotting on the FP sheets. Specimens
exhibiting hemolysis were rejected for study purposes. The presence of
lipemia was noted but was not a criterion for specimen exclusion. The
hemoglobin concentration was measured for all specimens, using an
Instrumentation Laboratory Model 682 Co-oximeter. The homogeneity of
blood during the spotting procedure was verified by measurement of
hemoglobin concentrations on serial aliquots of a representative blood
specimen. Aliquots dispensed at 15-s intervals for 6 min gave the
following results: mean ± SD, 2279 ± 25 µmol/L (14.7 ± 0.16
g/dL); range, 2233–2310 µmol/L (14.4–14.9 g/dL). No time-dependent
trends were observed.
Blood lead concentrations were determined using the protocol of the established PT program (24). Briefly, 10 referee laboratories were provided 2-mL aliquots of liquid blood for each specimen, which were analyzed using their routine procedures. Methods included electrothermal atomization AA and anodic stripping voltammetry techniques. Target values were then determined by the mean of the referee laboratory results, after deletion of outliers. Results falling outside the range, target value ± 0.19 µmol/L (4 µg/dL) or 10%, were excluded as outliers.
reporting and evaluation
Monthly FP PT events commenced in October 1997. Each monthly
testing event consisted of five specimens. The FP specimens, along with
a report form, were shipped by surface mail in
Tyvek® envelopes. Participants then reported FP
results to WSLH for evaluation. Participants analyzed 45 discrete FP PT
specimens. Some specimens had identical lead concentrations. The 10
initial specimens from the October and November 1997 testing events
were preliminary challenges and were used to assess the suitability of
the FP PT scheme. Duplicate FP sheets were provided for these initial
events, and results were not evaluated using CLIA criteria, although
referee laboratory target value information was provided to
participants. The remaining 35 specimens from events spanning December
1997 through June 1998 were evaluated for proficiency, using the
standard CLIA acceptability criteria for blood lead. Single FP sheets
were provided for all other events to more closely approximate actual
patient specimen submissions. Additional FP sheets were made available
upon request after reporting deadlines.
For the evaluated FP PT events, the CLIA acceptable range for
individual results included the target value ± 0.19 µmol/L (±4
µg/dL) or 10% (whichever is greater). Results were also examined on
an event and cumulative basis, using the CLIA criteria. An event score
of 
80% (four of five) acceptable specimen results constituted
satisfactory performance. Satisfactory scores on at least two of three
consecutive events constituted successful performance. Target values
were rounded to the nearest whole number, consistent with the
established PT program. Participant results are also rounded in the
established program, but FP results reported to the 0.005 µmol/L (0.1
µg/dL) level were not altered in this way. No FP participant
results were evaluated as unacceptable because of target value
rounding.
Evaluation reports were distributed to the participating laboratories within 2 weeks of the reporting deadline. The individual laboratories were identified by code letters A-F to maintain confidentiality. Reports included tables listing the referee laboratory mean, the SD, and individual results, as well as the participant laboratory data. Results outside the CLIA acceptable range were indicated with an asterisk following the reported value. Although the CLIA regulatory criteria were utilized for evaluation, the PT program was not regulatory.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
participant group results
The individual participant results for both preliminary and
evaluated specimens, plotted as a function of deviation from the target
value, are shown in Fig. 1
. Results on the preliminary challenges, to the left of the
vertical line in Fig. 1
, exhibit wide scatter. Absolute deviation from
the target value for this group of specimens (mean ± SD), was
0.29 ± 0.13 µmol/L (5.9 ± 2.7 µg/dL) and ranged from
+0.77 µmol/L (+16 µg/dL) to -0.92 µmol/L (-19 µg/dL).
Although the observed data scatter for the group as a whole was
bidirectional, the bias observed in results from individual
laboratories, when present, was unidirectional.
|
Following the preliminary events, all specimens were evaluated using
the CLIA acceptability criteria. Shown to the right of the vertical
line in Fig. 1
, a substantial improvement in participant accuracy
coincided with the initiation of PT evaluation. The mean absolute
deviation from target values decreased by >50%, to 0.12 ± 0.05
µmol/L (2.5 ± 1.0 µg/dL), with a range from +0.53 µmol/L
(+11 µg/dL) to -0.72 µmol/L (-15 µg/dL). Five of the six
individual laboratories exhibited statistically significant
improvement; laboratory A, P = 0.012; laboratories B,
C, and D, P <0.001; and laboratory E. P =
0.004. The performance of laboratory F did not change significantly
between the preliminary and evaluated specimens. The improvements can
be attributed to changes made by participants, although the specific
nature of these changes is not known. No changes to program procedures
were made.
The 35 FP PT challenges evaluated by the six participants produced 210 discrete results. Of these, 176 (84%) fell within the CLIA acceptable range. Of the 34 results falling outside the acceptable range, 29 (85%) fell below the minimum threshold, whereas only 5 (15%) were unacceptably high.
method-specific performance
The six participating FP laboratories used three distinct
methodologies. Laboratories A, E, and F extracted blood spots with an
(NH4)2HPO4-Triton
X-100 solution, followed by electrothermal atomization AA measurement.
The methods use blood-based calibrators and thus are similar to the
method described by Wang and Demshar (12). The extraction
solution used by laboratory A also contained
HNO3. Laboratory B used a Delves cup flame AA
procedure with a preashing step and matrix-matched calibrators, similar
to the method described by Verebey et al. (15). Laboratories
C and D extracted blood spots with a HNO3-Triton
X-100 solution followed by electrothermal atomization AA measurement.
These laboratories utilized aqueous calibrators, based on the method
described by Yee and Holtrop (10). Grouped by these
methodologies, regression of the individual laboratory results against
target values is shown in Fig. 2
. Data from the participating laboratories correlated well with
target values. Laboratories A, E, and F demonstrated the highest
correlation, r 0.98. Regression slopes for laboratories E
and F were not statistically different from unity, whereas laboratory A
showed a small but significant (P <0.05) negative bias.
Laboratories B, C, and D exhibited substantial negative bias in the
slope of their regressions, approaching 25% in the case of laboratory
C. The bias was highly significant statistically (P <0.001)
for each of these laboratories. In addition, the
y-intercepts of laboratories B and D differed significantly
from zero (P <0.01 for both), exceeding 0.13 µmol/L (2.7
µg/dL) in both laboratories.
|
predicted results at cdc decision levels
FP results at significant blood lead decision level concentrations
were predicted using the participant regression equations. Predicted FP
results at CDC decision level lead values of 0.48, 0.97, and 2.17
µmol/L (10, 20, and 45 µg/dL) (2) can be seen in Table 1
. Because the highest PT specimen concentration was 1.83
µmol/L (38 µg/dL), the predicted results at the 2.17 µmol/L (45
µg/dL) decision level represent an extrapolation of the actual data.
Predicted results fell within ± 0.05 µmol/L (1.0 µg/dL) for
five of the participants at the 0.48 µmol/L (10 µg/dL) decision
level and for three of the participants at the 0.97 µmol/L (20
µg/dL) decision level. The largest predicted differences at these
decision levels were -0.11 and -0.23 µmol/L (-2.2 and -4.7
µg/dL), respectively. At the 2.17 µmol/L (45 µg/dL) decision
level, predicted results from three laboratories fell within ±
0.14 µmol/L (2.8 µg/dL), but results from the other three
laboratories had a predicted negative bias exceeding 0.24 µmol/L (5
µg/dL).
|
An evaluation of participant performance under the CLIA ‘88
acceptability criteria for blood lead is shown in Table 2
. The CLIA regulations established performance criteria at
several levels (23). Acceptable results for individual PT
specimens are defined as the target value ± 0.19 µmol/L (4
µg/dL) or 10%, whichever is greater. CLIA mandates that testing
events consist of five PT specimens. An event score of 80% (four of
five) acceptable specimen results constitutes satisfactory event
performance. Event scores, in turn, are evaluated on a cumulative
basis. Satisfactory results on 67% (two of three) consecutive
testing events constitute successful performance. The 35 evaluated PT
specimens were provided as 7 five-specimen testing events, conforming
to the CLIA configuration. The events took place on a monthly basis to
accelerate data accumulation, rather than three times per
year as specified under CLIA. Performance varied considerably among the
participants. On an individual specimen basis, acceptable results
ranged from 100% (35 of 35) attained by laboratories A and F to 54%
(19 of 35) scored by laboratory C.
|
This scoring included seven alphanumeric results. Laboratory A did not
quantify results exceeding 0.97 µmol/L (20 µg/dL) in the first
evaluated testing event, reporting results as >0.97 µmol/L (>20
µg/dL) for three specimens with target values of 1.06, 1.16, and 1.83
µmol/L (22, 24, and 38 µg/dL). Laboratory D reported a result of
>1.69 µmol/L (>35 µg/dL) for the specimen with a target value of
1.83 µmol/L (38 µg/dL). Laboratory A also reported results of
<0.10 µmol/L (2 µg/dL) on specimens with target values of 0.05 and
0.10 µmol/L (1 and 2 µg/dL). Laboratory E reported a result of
<0.05 µmol/L (<1 µg/dL) on a specimen with a target value of 0.14
µmol/L (3 µg/dL). Identical data points and the use of alphanumeric
results led to fewer than 35 distinct data points on the plots in Fig. 2
. All seven of the alphanumeric results were scored as acceptable for
evaluation purposes.
At the testing event level, three of the six participants, laboratories A, E, and F, achieved satisfactory performance on 100% (seven of seven) of the testing events. The poorest performance was observed in laboratory C, which achieved satisfactory performance in 29% (two of seven) of the events. When the cumulative scoring criteria were used, laboratories A, E, and F also demonstrated successful performance throughout the seven testing events. With the exception of laboratory C, all of the laboratories were demonstrating successful performance at the completion of the final testing event.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Some statements about individual laboratory performance can be made. Performance varied widely between individual laboratories. The best performance was exhibited by laboratories A, E, and F, all of which use an extraction reagent containing (NH4)2HPO4 and Triton X-100 as well as blood-based calibrators. In contrast, the poorest performance, present as a negative bias, was demonstrated by laboratories C and D, both of which use a HNO3/Triton X-100 extraction reagent and aqueous calibrators. The source of this negative bias is not well understood. Possible causes include matrix effects related to the use of aqueous calibrators or adverse effects of EDTA on the efficiency of the HNO3 extraction. The EDTA present in the PT specimen blood spots would not be present in patient specimens submitted to these laboratories for routine analysis. In November 1997, laboratory D submitted a small (four specimens) experimental data set utilizing an alternative calibration of EDTA-anticoagulated blood rather than the aqueous calibration the laboratory routinely uses. This data set showed much closer agreement to target values, which is consistent with either of the proposed explanations. Unfortunately, laboratories C and D discontinued participation after the first year of the pilot study; therefore, the ability of the FP PT program to resolve this question is limited.
Negative bias was not confined to laboratories C and D; laboratory B
also exhibited a substantial negative bias. In all three laboratories,
the bias was more pronounced at higher blood lead concentrations.
However, because laboratories B and D also displayed large positive
y-intercepts, the bias is not apparent at lower blood lead
concentrations. This is illustrated by the predicted results seen in
Table 1
. Regression slopes significantly below unity are consistent
with observations made by Yee and Holtrop (10) and Verebey
et al. (14), who reported slopes of 0.83 and 0.88,
respectively, when comparing paired venous and FP samples in their
studies.
There are two noteworthy limitations of the FP Pilot PT Program. One is the lack of FP PT specimens exceeding 1.93 µmol/L (40 µg/dL). Although the National Blood Lead PT Program routinely distributes specimens at these higher concentrations, the cohort of lead-exposed workers from which the FP specimens are drawn is routinely monitored by the Occupational Safety and Health Administration, which diminishes the probability of obtaining specimens exceeding 1.93 µmol/L (40 µg/dL). Alternative blood sources will be sought to overcome this limitation. In addition, questions relating to the potential contamination of the FP sheets during production and the potential for contamination of FP specimens during blood collection and handling were beyond the scope of this study. Other areas for additional study remain. The FP PT Program is investigating the suitability of a nonhuman blood source, the long-term stability (>60 days) of spotted FP sheets, the effect of medication use by donors, and the possible influence of hematocrit and lipemia on FP testing results. These data will be reported at a later date. Consistent with all PT, the results reported here were obtained on known evaluation specimens. Blind specimen submission would provide additional valuable data on the reliability of these methods.
Based on the results of this study, we have obtained preliminary Health Care Financing Administration certification for the FP PT Program as a CLIA provider. This will provide FP laboratories with service comparable to that available for laboratories testing liquid blood and will also require them to conform, at a minimum, to CLIA standards required for other blood lead testing laboratories. It is notable that only five laboratories currently are enrolled in the FP PT Program. Although only laboratories A, B, and E currently offer FP testing commercially, this number is likely to grow.
In conclusion, an effective FP PT program has been developed to capably assist and monitor FP laboratories. Based on overt proficiency testing results, properly utilized FP testing methods can accurately measure blood lead concentration.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  K. Verebey, T. P. Moyer, D. E. Nixon, and O. K. Ash Filter Paper-collected Blood Lead Testing in Children The authors of the editorial cited above respond: Clin. Chem., July 1, 2000; 46(7): 1024 - 1028. [Full Text] [PDF]
|
|
|
|
|
|
N. Stanton, J. Maney, and R. Jones More on Filter Paper Lead Testing Clin. Chem., July 1, 2000; 46(7): 1028 - 1029. [Full Text] [PDF]
|
|
|
|
|
|
T. P. Moyer, D. N. Nixon, and K. O. Ash Filter Paper Lead Testing Clin. Chem., December 1, 1999; 45(12): 2055 - 2056. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
