Toxicity Testing cont’d

• Toxicology is Easy - Discussion

• Shape of Dose-Response Curve
– Linear vs Sigmoid
– Steep vs Flat
• Why LC50?
• Acute Toxicity Test Design
• Probit Analysis
Shape of the Dose-Response Relationship
 Why LC50?
100%
Response

0%

Low High
Concentration (or Dose)
 Acute Toxicity Test Design
1. Test Material (toxicant)
• Pure
• Commercial formulation
• Mixtures of known concentration
• Carriers/solvents
• Unknown mixtures (eg. sediment,effluent)
2. Test Organism
• Most sensitive Ex: Daphnia
• Most representative Fathead minnows
• Wild species Rats/Mice
• Rear in lab Animal Cell cultures
• Known physiology Algal cell cultures

• Bred for uniformity Duckweed

Mealworms
• Certified disease free
Earthworms
• Known susceptible
Frogs/tadpoles
strain
3. Exposure Method/Apparatus

• Oral dose/gavage
• Diet
• Intraperitoneal injection
• Inhalant
• Dermal
• Dry vial
• Static vs Flow-thru aquaria
4. Experimental Design

• Sample size
• Unbiased allocation of subjects
• Test environment (temp, 02, pH, light cycle, food, etc)
• Negative controls (untreated, solvent/carrier)
• Positive controls (toxin with well known effect)
• Baseline measurements (size, test envt, etc)
5. Range-Finding Test

• 10X progression of toxin concentration

• 3-5 individuals per concentration
• 5 concentrations plus control(s)
6. Definitive Test

• Expand on meaningful conc. From range test

• 5+ conc. Plus control(s)
• Geometric progression of conc. (2X or higher)
• 1 conc. Kills < 35%; 1 conc. Kills > 65%
• 10+ individuals per concentration
• Replicates?
• 96h, no food
7. Endpoint (what response to measure?)

• Death (LC50, LD50)

• Paralysis, loss of equilibrium (EC50)
• Other sub-lethal endpoints (EC50)

Pop. growth rate Bone formation

Indiv. growth rate Protein production
Foraging behavior Enzyme activity
Escape behavior Chromosome breakage
Learning/cognitive RNA synthesis
 8. Calculations
•Plot %mortality vs log conc. (or dose)
•Do not include control data in curve fit
•If control mortality exceeds 10%, correct
w/ Abbotts formula
•Do probit analysis for accurate LC50
•Calc. 95% confidence intervals
(Lindane)

Diff. toxicity to diff. spp.

Diff. Toxicity via diff routes of exposure – why?
 Yikes! – those are
my tax dollars!

What about
transgenerational
effects??
Scale of Toxicological Endpoints

Increasing reality
Increasing cost
Increasing uncertainty
 Probit Analysis
• Turn a curve into a line
Can connect dots more accurately
Allows accurate “inverse prediction”
Allows statistical analysis using regression/linear models

• Probit = probability unit

% of population responding as a function of standard deviation units from the
mean
 No. of Individuals Tolerance

No. of Individuals
Concentration Log of Concentration

Dose-Response
Cumulative Percent

Cumulative Percent

Concentration Log of Concentration

Result = ‘nearly’ straight line
 50%
Mean
Normal Distribution Std Dev Units

1. What % of observations
fall with each SD unit?

2. Express % from above

as cumulative percent.

3. Assign probits to
cumulative %.
 50%
Mean
Normal Distribution Std Dev Units

1. What % of observations
fall with each SD unit?

2.5 13.3 34.2 34.2 13.3 2.5

2. Express % from above

as cumulative percent.

3. Assign probits to
cumulative %.
 50%
Mean
Normal Distribution Std Dev Units

1. What % of observations
fall with each SD unit?

2.5 13.3 34.2 34.2 13.3 2.5

2. Express % from above

as cumulative percent.
2.5 15.8 84.2 97.5

3. Assign probits to
cumulative %.
 50%
Mean
Normal Distribution Std Dev Units

1. What % of observations
fall with each SD unit?

2.5 13.3 34.2 34.2 13.3 2.5

2. Express % from above

as cumulative percent.
2.5
2.5 15.8 84.2 97.5

3. Assign probits to
2.5 15.8 84.2 97.5
cumulative %.
Log Probit
1.009 6.18

6.18

1.009
“Inverse Prediction”

Log LC84= 0.93; LC84= 8.51

Log LC16 = 0.43; LC16 = 2.69
Log LC50 = 0.68; LC50 = 4.79

Why calc. 95% C.L.? How?

 Calculating 95% Confidence Limits of LC50's
(source: F. Matsumura. 1985. Toxicology of Insecticides, 2nd Ed., Plenum, pp.14-
16)
Example Worksheet - Rotenone toxicity to Macrosphoniella sanborni

1. Use inverse prediction from the graph to estimate the Log 10 of the LC84, LC16,
and LC50 then “un-log” the values and express as mg/L.
Log10 Conc. Conc.(mg/L)
eg: LC84 =
LC16 =
LC50 =

2. Calculate S and Log10 (S) (use the 'un-logged' dose/conc. values).

LC84 LC50
S = LC50 __ LC16
2
S= Log10 (S) =
 Calculating 95% Confidence Limits of LC50's
(source: F. Matsumura. 1985. Toxicology of Insecticides, 2nd Ed., Plenum, pp.14-
16)
Example Worksheet - Rotenone toxicity to Macrosphoniella sanborni

1. Use inverse prediction from the graph to estimate the Log 10 of the LC84, LC16,
and LC50 then “un-log” the values and express as mg/L.
Log10 Conc. Conc. (mg/L)
eg: LC84 = .93 8.51
LC16 = .43 2.69
LC50 = .68 4.79

2. Calculate S and Log10 (S) (use the 'un-logged' dose/conc. values).

+ +
LC84 LC50 8.51 4.79
S = LC50 __ LC16 4.79 2.69
2 2
S= 1.78 Log10 (S) = 0.250
3. Determine N
N = the total number of individuals tested between the range of dosages that
correspond to the LC16 to the LC84.

N=

4. Calculate Log10(f) and f.

2.77
Log10(f) = √N x Log10 (S)

Log10(f) = f=

5. Calculate Upper and Lower 95% Confidence Limits (multiply or divide 'unlogged'
conc/dose values by f):

Upper Limit = LC50 x f =

Lower Limit = LC50 / f =

3. Determine N
N = the total number of individuals tested between the range of dosages that
correspond to the LC16 to the LC84.

N = 49 + 46 + 48 = 143

4. Calculate Log10(f) and f.

2.77
Log10(f) = √ N x Log10 (S) =

Log10(f) = f=

5. Calculate Upper and Lower 95% Confidence Limits (multiply or divide 'unlogged'
conc/dose values by f):

Upper Limit = LC50 x f =

Lower Limit = LC50 / f =

3. Determine N
N = the total number of individuals tested between the range of dosages that
correspond to the LC16 to the LC84.

N = 49 + 46 + 48 = 143

4. Calculate Log10(f) and f.

2.77 2.77
=
Log10(f) = √ N x Log10 (S) √ 143 x 0.250

Log10(f) = 0.058 f = 1.143

5. Calculate Upper and Lower 95% Confidence Limits (multiply or divide 'unlogged'
conc/dose values by f):

Upper Limit = LC50 x f =

Lower Limit = LC50 / f =

3. Determine N
N = the total number of individuals tested between the range of dosages that
correspond to the LC16 to the LC84.

N = 49 + 46 + 48 = 143

4. Calculate Log10(f) and f.

2.77 2.77
=
Log10(f) = √ N x Log10 (S) √ 143 x 0.250

Log10(f) = 0.058 f = 1.143

5. Calculate Upper and Lower 95% Confidence Limits (multiply or divide 'unlogged'
conc/dose values by f):

Upper Limit = LC50 x f = 4.79 x 1.143 = 5.47

Lower Limit = LC50 / f = 4.79 / 1.143 = 4.19

LC50 = 4.79 mg/L (4.19 – 5.47; 95% C.L.)

Log scale