Accuracy of Oral Fluid Lead TestingLynn Gardner, MD, Emory University Atlanta Georgia; Robert Geller, MD, Emory University Atlanta Georgia;Robyn Hannigan, Ph.D. University of Massachusetts; Yu Sun, MD, MPH Ga. Department of Community Health; Anil Mangla, M.S.,Ph.D., MPH. FRIPH, Ga. Department of Community Health

Disclaimer

Drs. Gardner, Geller, and Mangla are independent evaluators of this lab method and have no conflicts of interest with this lab methodology or with any vendor of testing services.

Background

Incidence of lead poisoning decreasing in U.S.

Estimated 250,000 cases in children 1-5 yrs in 2008

Background

Lead poisoned children are at higher risk for development of :

Neurodevelopmental deficits Behavioral abnormalities Learning Problems

Background

Problems associated with current blood lead screening practices:

Painful for patients

Difficulty in obtaining blood samples in children

Amount of time consumed in office setting

Objective◦Given that there is a tight correlation

between oral fluid leads and venous blood lead in vitro, would oral fluid be a suitable screening test for elevated blood lead in a clinical setting?

Limitations◦The prevalence of lead poisoning in the

study cohort was less than the national average of 1.5%

◦Due to the small number of elevated blood lead levels, the performance characteristics of this test at higher blood lead levels was not established.

This study compares oral fluid lead levels and blood lead levels in a clinical setting.

Study Site

Children’s Healthcare of Atlanta at Hughes Spalding

• Primary Care Clinic Site

98% Medicaid Population

• Continuity Clinic site for pediatric attendings and residents at Emory University School of Medicine

Methods

Oral fluid samples collected on 500 children aged 6 months old to 5 years old

Children having venous blood lead levels drawn for routine screening purposes were eligible.

Methods

The blood lead levels were measured using standard methodology at a CLIA certified laboratory.

The oral fluid lead levels were measured using an ICP-MS (DRC II, PerkinElmer Sciex).

Oral fluid samples from 50 children were gathered twice to provide internal controls, but were only counted once.

MethodsThe mean absolute difference between

the sample groups was determined to test the hypothesis that group means are equal.

Results

Five hundred patients agreed to enroll. No eligible patient declined to

participate.474 patients had both blood and oral

fluid samples available for analysis.26 did not have blood available.

Results

Of the 474 patients with both blood and oral fluid samples available:

# with oral fluid lead level <4mcg/dL = 455 # with blood lead level <4mcg/dL = 455 % correlation = 92% # with oral fluid lead level ≥4mcg/dL = 19 # with blood lead level ≥4mcg/dL = 19 % correlation = 92%

All study results

y = 0.9183x

R2 = 0.5083

0

2

4

6

8

10

12

14

16

0 2 4 6 8 10 12 14 16

Blood Lead Level

Sa

liva

Le

ad

Le

va

l

Data Results9/474 patients were identified as false

positives (1.9%)0/474 patients were missed as false

negatives (0%)

Determining Cut-Off valuesA sensitive test is positive when a disease is

present A specific test is usually negative when a

patient is disease freeHigher sensitivity of a test is associated with

fewer false negativesTo establish a cut-off value, determine the

test characteristics to assess highest sensitivity with highest specificity

Determining Cut-Off valuesCut – off points Sensitivity Specificity

2 mg/dL 0.84 0.96

3 mg/dL 0.92 0.99

4 mg/dL 0.96 0.99

5 mg/dL 0.93 0.99

6 mg/dL 0.88 1

7 mg/dL 0.85 1

Conclusions

Oral fluid may be a reliable medium to use when screening children for lead exposure for levels < 4mcg/dL.

Oral fluid lead levels ≥ 4mcg/dL should be confirmed by a venous blood sample.

The sample size in the ≥ 4mcg/dL group (n=19) was inadequate to draw conclusions on accuracy at higher lead levels.

Further studies are being conducted.

Conclusions

The use of oral fluid for lead screening should improve screening success by:

Reducing parental refusal rates. Eliminating inability to obtain an

adequate sample. Decreasing sample collection times. Allowing for large groups of children

to be screened more quickly and easily than conventional methods.

Conclusions

Based on these results, we believe that oral fluid can be used as a screening tool instead of blood to accurately screen for lead poisoning in children.

Preliminary Laboratory Validation

Sample Preparation & Analysis• Samples are cut to a

standard length • Samples are weighed

(triplicate masses taken)• Samples are acid

digested (ultrapure HNO3) by microwave digestion

• Samples, once digested, are brought to volume and spiked with internal standard (Indium, final concentration 20 ppb)

• Samples are analyzed by ICP-MS (PerkinElmer DRCII)

• Follow EPA 6020 with modification

• Pb isotopes monitored (204, 206, 207, 208)

• Concentrations based on external calibration (208Pb) and internal standardization (115In)

• Calibration every 10 samples– Required R2 > 0.999 to proceed

• Check standards run every 5 samples– Required 95% of known check to

proceed

Sample Preparation & Analysis• 100 blank swabs analyzed

per manufacturer batch, Pb concentrations used as “blank”

• Analytical concentrations corrected for preparation dilution and blank concentration

• “Oral Fluid” concentration reported as mg/dL

• Method Detection Limit performed for each analytical run

• Within and Between-Day variation assessed using spiked swab samples

• Percent Recovery studies using spiked swabs performed to assess digestion efficiency

Linearity

Inter-Run Precision (CV) and AccuracyKnown Concentration (mg/dL)

1.0 mg/dL 10 mg/dL

Day 1

1.1.981.021.050.971.011.02

10.129.8710.129.9910.099.9510.05

Day 2

0.881.010.971.031.011.000.98

9.4310.019.749.969.9610.039.84

Day 3

1.051.0410.30.961.041.021.00

9.739.939.699.7810.2310.169.69

MeanS.DCVAccuracy

1.010.044.3999.19

9.920.191.9699.22

Oral Fluid Lead Proficiency Correlation

0

2

4

6

8

10

12

0 5 10 15 20 25 30

N

ug

/dL

Venous

Saliva

27 pairs of samples were analyzed by an ICP-MS for saliva and blood lead for each individual patient.

Calibration data using external calibration standards (n=6) and blank, Internal standard (20 ppb In), instrument configuration (EPA 6020, DRC II ICP-MS, no cell gas).

Method Detection LimitPerformed at the start ofeach daily analytical run

7 individual standards comprised of a “spiked” swabcontaining 0.30 ug/dL Pb

Standard Deviation of 7measurements * Student-t for n=6 at 99% confidence

15 MDL determinationsMDL = 0.006 +/- 0.002 ug/dL Pb(0.06 ng Pb per mL fluid)

Concentration of Pb in standard (ug/dL)

0 2 4 6 8 10 12

Co

un

ts p

er s

eco

nd

Pb

/In

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

mean data n = 50 independent calibrations95% Confidence Band 95% Prediction Band

R2 = 0.999; y = 1.059e-2+/- 0.00499 + 9.68e-2+/- 0.00115* (x)Constant Variance Test:Passed(P = 0.181)

METHODS Standard Operating Procedure – EPA 6020 for ICP-MS with the following modificationsIndium as single internal standard, 6 calibration standards (0.1 to 10 ug/dL Pb), 2 QA/QC standards (0.2, 7.0 ug/dL), run allowed to proceed if R2 > 0.999, QA/QC stds within 95% of known.

Concentration of standards (2% ultrapure HNO3) entered in software as ug/dL to allow for direct reporting of results in appropriate units, blank subtracted prior to quantification, all Pb counts per second normalized to In counts per second prior to quantification by the software.

Matrix suppression study revealed logarithmic response of detector to higher matrix concentrations, correction factor applied for dilutions during sample preparation and matrix effect (total dissolved solids) = 1.639log(counts per second Pb/In)

Digestion blanks also analyzed every 10 samples (see outline of methods) = below method detection limit

All data processing, calibrations, and QA/QC handled by PerkinElmer software (EPA 6020 methods adapted for use in this project)

Assessment of blank swab Pb concentration

Assessment of % yieldthrough digestion

Assessment of matrixeffect

Assessment of pre-weightneed

Blank swab conc of Pbbelow detection, cost-timeinefficient

Samples logged in labSwab tips cut to std length(ceramic scissors)

Tips weighed in triplicateAcid digested (HNO3)diluted, spiked with In,Analyzed by ICP-MS intriplicate

Analytical run - calibrationevery 10 samples, QA/QCwithin run of 10, MDL at start of each day100+ swabs cut, weighed in triplicate

acid digested (ultrapure HNO3)Quantified by ICP-MSAverage blank below detection limit of 0.006 ug/dL

Artificial saliva (1% diastase) spiked with 50 ug/dL Pb50 replicates% yield +/- 92.1%

Artificial saliva (1% diastase) spiked with 1 ug/dL to 500 ug/dL Pb100 samplesmatrix suppression higher at lower Pb concentrationsCorrection factor for counts per second (cps) data accounting for sample prep dilutions and matrix suppression1.639log(cps)

First 100 samples - preweighed in labwhen arrive weigh again, differenceassumed to be “oral fluid” volume

Std OpProc