PBPK Model for Lead: Uncertainties and Parameter Estimation

1
PBPK Model for Lead Uncertainties and Parameter
Estimation

• by
• Sangam Uma Reddy

Thesis Supervisor Dr. Mukesh Sharma
2
Overview

• Introduction
• Objective of the study
• Literature review
• Methodology
• Results and discussion
• Conclusions

3
Lead

• Versatile heavy metal
• Extensively used
• Cheap, useful, easy to mine, physical properties
  - ubiquitous in air, food, water and soil
• Cumulative Neurotoxin, no known biological
  function
• one of most hazardous substances (ATSDR)

4
Usage of Lead

• Batteries
• Pigments
• Rolled/Extrusions
• Ammunitions
• Cable Sheathing
• Petrol Additives

Source ILZSG, 1997
5
Effects of Lead

• Damage Central Nervous System
• Causes reduction in IQ and attention span
• Affects mental and physical development
• Reading and learning disabilities, hyperactivity
  and other behavioral problems
• Impairs formation of Hemoglobin, thus Anemia
• Irreversible brain damage
• Even death at higher concentration

6
However

• Lead continues to be in environment after
  several years of unleaded gasoline (Morisawa et
  al. 2001) Why?

• After phase out of lead from gasoline
• Immediate drop in air
• Exposure continues
• Food
• Water
• Soil
• Air ???

7
Objective of the study

• To estimate the parameter values (KELI and KEKI)
  of PBPK model using the observed blood and urine
  lead levels

8
Exposure Mechanism

• Routes of humans exposure

Absorption - 50 in children - 10 in
adults

• Ingestion

Absorption - 50 in children and adults

• Respiration

• Dermal

Absorption - Insignificant
9
Distribution

• Blood
• Soft tissue
• Bone

Liver, Kidneys, Brain and Muscle
95 of the Pb body burden in bones (OFlaherty,
1993)
Excretion

• Urine
• Bile
• Sweat
• Nails
• Hair

10
Blood lead level

• Lead health effects are many
• indicated by blood lead levels (PbB)

PbB an important biomarker
Acceptable levels of PbB 10 ?g/dL
Source CDC, 1991
11
PBPK Model

• PBPK - Physiologically Based Pharmacokinetic
  Model
• mathematical description of uptake and
  disposition of substances to quantitatively
  describe relationship among critical biological
  process
• Requires chemical substance-specific
  physicochemical parameters and species-specific
  physiological and biological parameters
• Numerical estimates of parameters are
  incorporated with set of differential and
  algebraic equations that describes the
  pharmacokinetic process

12
PBPK model for a chemical substance

• Model Representation
• Model Parameterization
• Model Simulation
• Model Validation

Source Krishna and Anderson, 1994
13
PBPK Model for Lead
Source Morisawa et al., 2001
14
PBPK Model for Lead
Liver
Kidney
Rapid Perfused Tissues
Slow Perfused Tissues
Bone
15
PBPK Model for Lead
Partitioning between tissue and plasma
Conc. in venous blood of each organ
16
Model Parameters

• Absorption through Inhalation Exposure (ALU)
• 30 50 (adults)
• Absorption through Gastrointestinal Tract (AGI)
• 8 -11 (adults)
• 40 50 (children)
• Metabolic Constants (KELI and KEKI)
• 30 (liver)
• 70 (kidney)

17
Uncertainty and Variability in PBPK models

• Model errors and data gaps
• Uncertainty in extrapolating animal data to the
  case of humans (especially metabolic parameters)
• Measurement errors and analytical uncertainties
• Uncertainty in exposure levels and parameter
  values
• Inter-or-intra species variability in kinetics
  may be due to differences in
• Physiology (body weight, body fat, Organ sizes,
  shapes)
• Variation (e.g. genetic) in metabolism and
  biochemistry
• Co-exposure to other chemicals
• Disease states

18
Methodology of the Study
Blood samples
Urine samples
Water Samples
Air samples
Filtration
Parameter Estimation and Risk characterization
19
Sampling Location
20
Sample Collection

• Air Sample Collection
• Food Sample Collection
• Blood Sample Collection
• Urine Sample Collection

21
Collection of Food Samples
Collected using Market Basket method
Groups Food Items
Non-Leafy Vegetables Potato, Brinjal, Tomato, Ladyfinger, Pumpkin, Beans, Cauliflower, Cucumber, Onion, Gourd, Cabbage, Carrot, Radish, Bottle Gourd
Leafy Vegetables Spinach, Fenugreek, Coriander
Fruits Banana, Orange, Papaya, Grapes, Apple, Guava
Cereals Wheat, Rice
Pulses Moong, Masoor, Arhar, Urad (Green), Urad (Black), Chana, Rajma, Chole
Milk Cow Milk, Buffalo Milk
22
Collection of Food Samples
No. of Food Samples Collected for Study
Site Food Group Singhpur Bhitoor Fields
Non-Leafy Vegetables 28 26 12
Leafy Vegetables 6 6 3
Fruits 12 12 2
Cereals 6 6 -
Pulses 24 24 -
23
Food Sample Collection
Duplicate Diet Survey
24
Blood Sample collection
25
Sample Analysis

• Air Sample Analysis
• Food Sample Analysis

• Filter Paper Conditioning
• Sample Extraction
• Instrumentation and Analysis

• Sample Processing
• Sample Extraction
• Instrumentation and Analysis

26
Sample Analysis

• Blood Sample Analysis
• Urine Sample Analysis

• Sample Extraction
• Instrumentation and Analysis

• Sample Extraction
• Instrumentation and Analysis

27
Sample Extraction
Extraction of Pb
Microwave Digestion System (Ethos Ez Labsatation,
Milestone, Italy)
28
Sample Analysis
Sample Analysis GFAAS (GBC Avanta Sigma)
Calibration Working Standards Wavelength 283.3
nm Volume injected 20 ?L
Graphite Furnace Program
Final Temperature Ramp Time Hold Time Gas Type
40 2.0 1.0 Inert
90 5.0 5.0 Inert
120 10.0 5.0 Inert
400 10.0 5.0 Inert
400 1.0 1.0 None
2100 1.5 2.0 None
2300 1.0 1.0 Inert
MDL 0.8 ppb Recovery Food Samples
94-95 Blood Samples 89
29
Risk Characterization
Examine Probability Distribution of lead levels
of Food Items and Quantity of Food Consumed
Non-Leafy Vegetables
30
Risk Characterization
Quantity of Food Consumed
Pb Levels in Food Items
Dietary Lead Intake
31
Results and Discussion
Average lead levels in food items
Food Group Present Study at Pratap Pur Hari (1) Urban area, Kanpur (2) US cities (3) Bombay (4) China (5) Britain (6) Basque (7) Madrid (8) IEPHM (9)
Cereals 106.38?80.12 (25.93207.01) n 6 119.99?82.13 (28.97223.90) n 8 (2136) 18.2 n 15 56.4100.0 (4616) n 59 20 33 (1065) n 12 3041?132 170
Pulses 220.80?116.47 (43.60405.36) n 48 283.28?118.42 (65.23415.98) n 21 253.3 n 13 33.025.6 (4143) n 34 10 (lt530) 22.4?3.0 350
Leafy vegetables 317.68?61.80 (191.68437.53) n15 325.60?74.06 (181.56541.61) n 32 100.4 n 11 10 430
Non-leafy vegetables 101.97?52.80 (36.17243.70) n 66 121.91?58.29 (24.84279.29) n 114 (5649) 4.1 n 32 20 23 (545) n 12 182?6 360
Fruits 5.65?1.77 (1.0517.35) n 18 7.32?6.1 (2.2217.60) n 10 (5223) 7.4 n 7 lt10 15 (lt525) n 10 181?19
Milk 0.46?0.19 (ND0.65) n 6 4.08?2.78 (0.47.79) n 8 (383) 1.6 n 4 lt10 9 (lt520) n 4 34.9?1.8 50
Water 3.96?0.86 (3.165.25) n 6 8.43?3.99 (4.516.04) n 11 lt 5 1.2 n 13 111

• Conclusions
• Pb concentrations in food items in Kanpur city
  are high compared to other cities.
• High in leafy vegetables.
• Concentration in food items from rural area is
  somewhat less to urban samples.

2Sharma et al. (2005) 3-ATSDR (1999) 4-Tripathi
et al. (1997) 5-Zhang et al. (1998) 6-Ysart et
al. (1999) 7-Urieta et al. (1996) 8-Cuadrado
et al. (1995), only data of Madrid are taken.
9-Krishnamurti and Vishwanathan (1991), only
data of Uttar Pradesh are taken.
32
Probability Distribution Plots Pb Levels in
Food Items

• Conclusion
• Except for non leafy vegetables and pulses Pb
  levels in all food items are normally distributed
  at 95 confidence.

33
Probability Distribution Plots Food Intake
Conclusion Food intake for all food items are
normally distributed at 95 confidence
34
Lead intake

• Lead intake through Inhalation
• Rural 0.28 µg/m3
• Urban 0.66 µg/m3 (Maloo, 2003)

• Lead intake through Ingestion

Diet consumption pattern
Food Item Food Intake (g/day) Food Intake (g/day) Food Intake (g/day)
Adult Veg dieta Adult veg dietb
Cereals 500 439?67
Pulses 57 54?8
Leafy Vegetable 21 46?8
Non-Leafy Vegetable 113 118?18
Fruits 18 31?8
Milk 163 228?5
Water 2 L 2?0.5
a Source Planning Commission, India (2002) b
Source Survey Conducted at Pratap Pur Hari
35
Pictorial Depiction of Probability exposure
assessment

x

Exposure
I2
C2
I1
C1
Exposure using Planning Commission data
x
x


I1
I2
C1
C2
Exposure
Exposure using Field data
I Food intake C Concentration of Pb in food
items
36
Comparison of Dietary Intake Values obtained
using Field data and Planning Commission data

• Conclusion
• To address the variability/uncertainty actual
  measurements of dietary Intake should be taken
  rather than going by fixed food consumption
  pattern

37
Blood and Urine Pb Levels
Mean8.34 SD1.94
Mean8.37 SD2.02
38
Comparison of Present Study PbB Levels with Other
Studies
Area No. of Samples Geometric mean concentration (µg/dL)
Deonar (suburban Bombay)a 28 8.9 (2.9-31.2)
Parel (central Bombay)a 60 11.5 (2.9-47.7)
Byculla (central Bombay)a 94 11.9 (1.1-35.3)
Greater Bombaya 77 14.4 (2.9-41.2)
Kanpurb 24 18 (ND-140)
Present study 68 8.34 (4.56-13.59)

• Conclusion
• PbB observed in present study are comparable to
  that of Deonar Study
• Study by Seth (2000) shows higher levels as data
  reported is for year 1996 when leaded gasoline
  was used

Source a R. N. Khandekar et at, (1987)
Source b Seth (2000)
39
PbB and PbU Levels
Area Survey Site Number of Subjects Age PbB (µg/L) PbU (µg/L) Correlation Coefficients References
Bangkok 52 19-57 32.3 (1.37) 2.35 (1.70) 0.31a Zhang etal._1998a.
Kuala Lumpur 47 21-47 65.4 (1.4) 4.74 (1.79) 0.43b Moon et al._1996.
Manila 45 21-64 37 (1.36) 3.64(1.82) 0.08 Zhang et al._1998b.
Tainan 51 22-66 33.9 (1.26) 1.54 (1.99) 0.12 Ikeda et al._1996.
Beijing 50 20-62 43.4(1.38) 5.73(1.69) 0.31a Zhang et al._1997.
Jinan 50 21-55 35.3(1.44) 2.16(1.55) 0.27 Ibid.
Nanning 50 23-57 54.5(1.42) 1.57(1.99) 0.38b Ibid.
Shanghai 50 23-58 55.4(1.47) 1.81(1.8) 0.45b Ibid.
Xian 50 24-58 43.4(1.32) 3.34(2.14) 0.02 Kae Higashikawa et al., 2000
Tokyo Kyoto 61 40-68 37.7(1.7) 1.74(2.63) 0.63b Shimbo et al._1999.
Seoul Pusan 55 31-49 47.2(1.27) 3.11(2.02) 0.11 Moon et al._1995.
--- 84 --- 300 21.33 0.9a Gross, 1979
Pratap Pur Hari 35 20-45 82.3(16.1) 5.58(1.34) 0.82a The present study

• Conclusions
• High correlation between PbB and PbU was
  observed in present study, Study by Gross (1979)
  and Shimbo et al. (1999).
• If PbB levels are high, kidney enhances its
  performance in terms of getting toxic metals out
  of system

Numbers in the parentheses show standard
deviation a. P lt 0.05 b. P lt 0.01
40
Relationship of Urinary Lead Excretion Rate and
PbB
Present Study
Gross, 1979

• Conclusions
• Trend is comparable with that of Gross, 1979
• KEKI may be variable from one person to another

41
Validation of PBPK Model
Morisawa et al. (2001) examined for reliability
of PBPK model by comparing simulated results with
experimental data Same exercise performed in
present study on same data using Mathematica
program for confidence on model performance
Experiment I (Rabinowitz et al. (1976))
Subject Dose pattern Body Weight (kg) Dose Rate (?g/day) Dose period (days)
A Dietary 70 204 104
42
Validation of PBPK Model
Conclusion The output from Mathematica program
matches with experimental data
Dots represent experimental data Solid line
output by Mathematica program
43
Performance of PBPK Model

• Conclusion
• In no case PbB level for an individual was
  outside the model computed range of his/her
  interval estimate of PbB.

44
Parameters (KELI, KEKI) Estimation
Steady state PBPK Model
(1)
.(2)
Cart Cven, RA
(3)
Cart Cven, SL
(4)
Cart Cven, BO
(5)
45
Cont
.(6)
.(7)
Cart Cven
(8)
..(9)
Let us take, MUout KEKICKIVKI
(mass/day) MUout Mass of lead excreted in
urine (mass/day)
46
Cont
MUout PbU x Urine discharge
From eq. (2),
.(10)
Recall Cart Measured PbB from subjects, Cven,KI
for all subjects calculated from eq (10)
Rewriting eq (9) to obtain Cven,LI
(11)
47
Cont
RHS of eq (11) is known
Corresponding CPi for Liver and kidney
Calculate Cven,LI and Cven,KI
CPi
Ci
Concentration of Pb in organ/tissue is known
Recall, MUout KEKICKIVKI(12)
In eq (12) all variables known, KEKI can be
estimated In eq (1) all variables known, KELI
can be estimated
48
KELI
Mean 0.16 SD 0.03
49
KEKI
Mean 0.66 SD 0.11

• Conclusion
• Metabolic parameters (KELI and KEKI) show
  substantial variation and one should take these
  parameters as random variables in model to fully
  reflect the uncertainties caused due to
  variability in KELI and KEKI

50
Lead in air, blood and urine (Azar et al., 1975)
Group Philadelphia Cab-drivers Starke FL Barksdale WI Los Angeles Cab-drivers Los Angles Office workers Present Study
PbA (µg/m3) 2.62 0.81 1.01 6.10 3.06 0.28
PbB (µg/dl) 21.6 15 12.9 23.7 18.9 8.23
PbU(µg/day) 22.7 15.2 18.2 26.4 20 8.37
Renal Clearance (kg/day) 0.104 0.095 0.135 0.11 0.103 0.101 0.015

• Conclusions
• Average value of clearance is close to the value
  reported in other studies through renal
  clearance.
• This study additionally provides information on
  associated uncertainties in renal clearance.

51
Parameter Sensitivity
Conclusions Varied in range 0.43 0.54 0.66
0.87 Unit change in KEKI PbB inc/dec
by 6 µg/dL In
plausible range of KEKI error in
PbB estimate can be 15
Conclusions Varied in range 0.09 0.12 0.16
0.19 Unit change in KELI PbB inc/dec
by 34 µg/dL In
plausible range of KELI error in
PbB estimate can be 23
52
Probabilistic Risk Characterization Improvements
Case1 Variability only in Pb concentration in
food intake

PbB Parameters (KELI, KEKI) (fixed) Food Intake (fixed) Pb Concentration in Food (Variable)
Case2 Variability in food consumption and
concentration

PbB Parameters (KELI, KEKI) (fixed) Food Intake (variable) Pb Concentration in Food (Variable)
Case3 Variability in food consumption,
concentration and parameters (KELI, KEKI)

PbB Parameters (KELI, KEKI) (variable) Food Intake (variable) Pb Concentration in Food (Variable)
53
Risk Estimation
Case 1
Case 1
8.9x10-5
8.9x10-5
Case 2
Case 2

• Conclusions
• The results suggest that by not considering the
  uncertainties, the error in risk characterization
  will be underestimated and risk engineers will
  err on side of false protection.
• Therefore it is important to address/include the
  uncertainties in risk Characterization.

5.4x10-4
Case 3
9.3x10-3
54
Conclusions

• The PBPK model parameters (food intake, Pb
  concentration food items, KEKI and KELI) vary
  from person to person to a large extent and thus
  they should be considered as random variables.
• Parameter values (KEKI and KELI) were found
  sensitive to model output (PbB). In the plausible
  range of KEKI and KELI, the error in PbB
  estimates can be 15 and 23 respectively.
• Overall risk characterization was done by
  considering these parameters as variables.

55
Conclusions

• The results suggest that by not considering the
  uncertainties, the error in risk characterization
  will be underestimated as given below

Variability only in Pb concentration in food intake 8.98 x 10-5
Variability in food consumption and concentration 5.43 x 10-4
Variability in food consumption, concentration and parameters (KEKI and KELI) 9.34 x 10-3
56
Conclusions

• It can be concluded that by not considering the
  uncertainties, the error in risk characterization
  will be underestimated. While wishing to remain
  conservative in the quantification of risk to err
  on the side of protection of humans and the
  environment, an underestimated uncertainty (e.g.,
  food intake, Pb concentration food items, KELI
  and KEKI) may eclipse safety and may result in a
  false sense of protection.

57
THANK YOU