L04-Dose-response relationships KvG 2024.pdf

04/06/2024 Dose-response relationships 1 The dose-response-relationship  the basis of toxicology Survival, growth or other measure of performance...

04/06/2024 Dose-response relationships 1 The dose-response-relationship  the basis of toxicology Survival, growth or other measure of performance NOEC LC50 or EC50 (log) concentration 2 1  04/06/2024 Example of a dose-response relationship alcohol consumption (g) plasma peak concentration (g/l) 3.5 150 3.0 coma 120 2.5 100 2.0 1.5 intoxication 60 1.0 30 20 0.5 reduced reaction 10 speed 0 (Figure provided by Dr HJM Verhaar) 3 Toxicological endpoints derived from dose-response relationships LD50/LC50: median lethal dose/concentration ED50/EC50: dose/concentration causing 50% effect ED10/EC10: dose/concentration causing 10% effect NOEC/LOEC: no/lowest observed effect concentration Units differ:  LD50, ED50 = dose in mg/kg body weight (often oral/topical dosing)  LC50, EC50, EC10, NOEC, LOEC = concentration in medium  Air: mg/m3  Water: mg/L  Soil, sediment, food: mg/kg 4 2  04/06/2024 Toxicological endpoints derived from dose-response relationships LD50/LC50: median lethal dose/concentration ED50/EC50: dose/concentration causing 50% effect ED10/EC10: dose/concentration causing 10% effect NOEC/LOEC: no/lowest observed effect concentration These endpoints provide measure of toxicity  Enable comparison of toxic potency of chemicals  how toxic is a chemical?  which chemical is most toxic?  The lower the endpoint value, the more toxic the chemical 5 Chemical LD50 Acute oral toxicity of selected botulin-toxin A 0.0005 diphtheria-toxin 0.3 chemicals expressed by their 2,3,7,8-TCDD 1 LD50 (lethal dose 50%) tetrodotoxin 15 in µg per kg body weight strychnin 500 aflatoxin 600 aldicarb 900 nicotin 1 000 Made by nature Methyl mercury 1 000 Made by man parathion 3 000 (Xenobiotics) HCN 10 000 Food additive thallium 10 000 DDT 113 000 Salt (NaCl) 4 000 000 Water (H2O) 160 000 000 6 3  04/06/2024 Dose-response curve characteristics Dichotomous/quantal response: % survival (animals) Graded (continuous) response: % enzyme activity 7 Dose-response curve characteristics 110 Dichotomous/quantal response: % mortality (animals) 100 Graded (continuous) response: % enzyme inhibition 90 80 70 Response 60 50 40 30 20 10 0 0,001 0,01 0,1 1 10 Dose or Concentration 8 4  04/06/2024 Potency and effectivity Three characteristics describing a dose-respons curve  Location on x-axis (ED50 or EC50, potency)  Maximal response on y-axis (effectivity)  Steepness (slope) of the curve 9 Toxicity test data Resonse Test concentration (mg/kg) 10 5  04/06/2024 Toxicological endpoints  LC50 – EC50 – EC10 – LOEC – NOEC LC50 / EC50 / EC10 determined by fitting dose-response model LOEC / NOEC determined by statistical test 11 NOEC / LOEC 12 6  04/06/2024 Definitions of NOEC and LOEC NOEC highest concentration tested with no significant difference in response with control LOEC lowest concentration tested with significant difference in response compared with control Comparison treatment – control 1. Simple: Student t test,  comparing concentrations one-by-one with control 2. Advanced: ANOVA combined with posthoc test  e.g. Williams, Dunnett’s etc. 13 NOEC : e.g. William’s test Assumption: response Yj at concentration cj normally distributed; Expectation Mj, with M0 > M1 > M2.. > Mk and variance σ2 NOEC Resonse LOEC Test concentration (mg/l) 14 7  04/06/2024 Disadvantages/problems with NOEC/LOEC - Obtained by statistical test (hypothesis testing) - Equal to one of the test concentrations (cj's) - Sensitive for the number of replicates - Sensitive for variation in response - Depends on the statistical test chosen, and on σ - No confidence intervals 15 Disadvantages/problems with NOEC - Inefficient use of test data (most data points ignored) - Level of effect at NOEC regularly > 20% - Poor testing leads to high (unprotective) NOECs Nevertheless, NOEC still in use 16 8  04/06/2024 Alternative approach: curve fitting 17 Dose-response curves 18 9  04/06/2024 EC50 (or LC50) estimates applying a logistic model Ymax 𝑌𝑚𝑎𝑥 b = slope 𝑌 𝑐 = 𝑐 1+ 𝐸𝐶50 Note: curve described by three parameters EC50 characterizing dose-response curve: concentration 1. Maximum response (effectiveness) 2. Steepness or slope 3. Location on X-axis (EC50, potency) 19 EC50, ECx 𝑌𝑚𝑎𝑥 Three parameters: Ymax, b and EC50 𝑌 𝑐 = 𝑐 Estimated by fitting equation to data, using 1 + 𝐸𝐶 sum-of-squares method; 50 Statistical programme (e.g. SPSS or R) needed to give 95% confidence intervals for Ymax, b and EC50 𝑌𝑚𝑎𝑥 Estimation of ECx 𝑌 𝑐 = 𝑥 𝑐 (e.g. EC10 or EC5) 1 + 100 − 𝑥 ∗ 𝐸𝐶𝑥 20 10  04/06/2024 ECx values: advantages - Obtained via estimation (regression based) - Not restricted to one of the test concentrations (cj’s) - Uses all information from test - Depend on the model chosen - Confidence intervals - Makes use of all available data from the test 21 Example of EC50 calculation Ymax = 9.58 Resonse b = 3.61 EC50 = 2.55 (2.39-2.71) mg/kg Test concentration (mg/kg) 22 11  04/06/2024 Special cases 1. Very steep dose-response curves, often for survival data  Logistic model not able to properly fit curve  Alternative: trimmed Spearman Karber method 2. Logistic model not fitting well  other (non-symmetric) models, e.g. Weibull-type curves 3. Hormesis 4. Dose-response curve with lower limit 23 Steepness of survival curves: depends on homogeneity of responses Probability curves 24 12  04/06/2024 Weibull models: non-symmetric curves 25 Special case: hormesis Stimulated response at low exposure levels van Ewijk & Hoekstra, 1993 26 13  04/06/2024 Hormesis 𝑌𝑚𝑎𝑥 ∗ (1 + 𝑓 ∗ 𝑐) 𝑌 𝑐 = 𝑐 1 + 2 ∗ 𝑓 ∗ 𝐸𝐶50 + 1 ∗ 𝐸𝐶 50 f = hormetic parameter Acc. to van Ewijk & Hoekstra, 1993 27 Logistic with lower limit For instance for plant growth starting from small plants 𝑌𝑚𝑎𝑥 − 𝑌𝑚𝑖𝑛 𝑌 𝑐 = 𝑌𝑚𝑖𝑛 + 𝑐 1 + 𝐸𝐶 50 Need estimating one additional parameter, Ymin 28 14  04/06/2024 Use of dose-response curves Forward use  predict effect caused by toxicant concentration Backward use  estimate exposure concentration that triggered certain response 29 Dosis-respons curve: forward use  to predict effect size 110 100 90 80 70 Response 60 50 40 30 20 10 0 0,001 0,01 0,1 1 10 Dose or Concentration 30 15  04/06/2024 Dosis-respons curve: backward use 110 100  to derive chemical properties 90 80 70 Response 60 50 40 30 20 10 0 0,001 0,01 0,1 1 10 Dose or ED Concentration 50 31 Dosis-respons curve: backward use 110 100  to derive chemical properties 90 and trigger values 80 70 Response 60 50 40 30 20Safe Unsafe 10 0 0,001 0,01 0,1 1 10 NOAEL LOAEL Dose or ED Concentration 50 NOAEL = No Observed Adverse Effect Level; LOAEL = Lowest Observed Adverse Effect Level 32 16  04/06/2024 Dosis-respons curve: backward use 110 100  to derive: 90 - critical effect dose (CED) or 80 - benchmark dose (BMD) 70 at critical effect size (CES) Response 60 50 40 Safe Unsafe 30 20 CES = 10 0 0,001 0,01 0,1 1 10 CED Dose or Concentration BMD 33 BMDL Safe Unsafe CED BMDL BMD  CES: Critical effect size BMD 95% CI  CED: Critical effect dose  BMD: Benchmark dose  CI: confidence interval  BMDL: benchmark dose lower confidence bound 34 17  04/06/2024 Dosis-respons curve: backward use 110  Response to unknown mixture 100 90 80 70 Response 60 50 40 30 20 10 0 0,001 0,01 0,1 1 10 Equivalent Dose or dose Concentration 35 Learning goals After this lecture you will be able to: Understand the concept of dose-response relationships Mention the three parameters characterizing dose-response curves Describe the disadvantages of the NOEC/LOEC approach and the advantages of the ECx approach Explain the difference between backward and forward use of dose- response curves Mention different types of dose-response relationships 36 18