ERF of dioxin

This page is a knowledge crystal of subtype variable. The page identifier is Op_en5823
Moderator:Jouni (see all)
Citation of this page: Jouni T. Tuomisto, Päivi Meriläinen: Dioxin exposure-response function on population level. Opasnet 2009. [3]. Accessed 31 Oct 2024.
Upload data {{#opasnet_base_link:Op_en5823}}

ERF of dioxin describes quantitative relationships between exposure to polychlorinated dibenzo-p-dioxins, polychlorinated dibenzofurans (PCDD/Fs), and polychlorinated biphenyls (PCBs) and several health effects such as cancer, developmental defects and others. It assumed that effects are mediated via the Ah receptor and that toxic equivalencies (TEFs)) apply.

Question

What are quantitative relationships between exposure to dioxin (a common name for polychlorinated dibenzo-p-dioxins, polychlorinated dibenzofurans (PCDD/Fs), and polychlorinated biphenyls (PCBs)) and several health impacts including

• remaining lifetime cancer risk and
• developmental defects in molar teethR↻

Answer

+ Show code- Hide code library(OpasnetUtils) objects.latest("Op_en5823", code_name = "initiate") # [[ERF of dioxin]] ovariables ERF, threshold. oprint(summary(EvalOutput(ERF)))

Rationale

Dioxins are a group of polychlorinated dibenzo-p-dioxins (PCDDs) and dibenzofurans (PCDFs). They are persistent environmental contaminants that accumulate in the human body. Their elimination half-life is quite high (~7 years). 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) is the most toxic PCDD/Fs congener, and it is classified as a known human carcinogen by the International Agency for Research on Cancer (IARC).

• Health effects related to long-term exposure
  • impairment of the immune system
  • impairment of the developing nervous system
  • impairment of the endocrine system
  • impairment of reproductive functions
  • increased cancer risk

Data

Human health effects caused by dioxins

Dioxins are persistent environmental pollutants and they accumulate in the food chain. Dioxins cause a large variety of effects in laboratory animals. They are carcinogenic at large doses, and they also cause developmental defects. The evidence of human effects has been more limited, because the exposure levels have been much lower than in animal tests. However, an increased cancer risk has been observed after high industrial occupational exposures. In addition, mild tooth mineralisation defects have been observed in children in Finland, even after typical exposures of the 1980's. Children are exposed to dioxins mostly via mother's milk. The dioxin levels have been decreasing since then, and no tooth defects have been observed at the current exposure levels.

ERF of dioxin: Difference between revisions(-)

Obs	Exposure agent	Trait	Response metric	Exposure route	Exposure metric	Exposure unit	ERF parameter	Scaling	Threshold	ERF	Description
1	TEQ	Cancer, total	Lifetime probability	Ingestion	Intake	pg /kg /d	CSF	BW	0	0.000156	US EPA. From ERF of dioxin: 156000 (mg/kg/d)^-1
2	logTEQ	Developmental dental defects incl. agenesis	Yes/No according to "Developmental Defects of Enamel Index"	Ingestion etc. (as it was in Seveso)	log(TCDD serum concentration+1)	ng/kg in fat	ERS	None	0	0.26 +- 0.12	Alaluusua et al. 2004; PL Gradowska PhD thesis 2013. From ERF of TCDD. Resulting distribution based on one simulation. Weibull(alfa=0.2925,beta=2.192)
3	logTEQ	Tooth defect	Yes/No according to "Developmental Defects of Enamel Index"	Placenta and mother's milk	log(TCDD serum concentration+1)	log(pg/g in fat)	ERS	None	0	0:0.06:0.12	Alaluusua et al. 2004 data with PL Gradowska PhD thesis 2013 approach but we used the response function y = k x + b (see below)
4	TEQ	Dental defect	Yes/No according to "Developmental Defects of Enamel Index"	Placenta and mother's milk	Ingestion	pg/d	ERS	None	0	0.001382	Alaluusua et al. 1996
5	TEQ	Cancer	Morbidity	Ingestion	Intake	pg /kg /d	UR	BW	0	0.000032; 0.000035; 0.00016	From ERFs of environmental pollutants
6	TEQ	Dioxin recommendation	Tolerable daily intake	Ingestion	Intake	pg /kg /d	TDI	BW	0	1	Lower limit of 1-4 pg/kg/d

ERF of dioxin on cancer is indexed by age. It applies to adult age categories, > 18 years (gender combined).

Cancer

The U.S. Environmental Protection Agency (US EPA) calculated an oral cancer slope factor (CSF) for 2,3,7,8 - TCDD (the most toxic dioxin compound). This CFS equals to 156000 (mg/kg bw-d)^-1 and represents the upper 95^th percentile estimate of the probability of developing cancer per unit intake of chemical over a lifetime.

Evidence concerning cancer risk is mainly from animal studies, and dioxins are probably quite weak carcinogens in humans. Hormesis type of dose-response is suspected. Evidence concerning other health effects is inconsistent.

In this specific case (op_fi:Hämeenkyrön jätteenpolttolaitos):

• MSWI is likely to increase background dioxin exposure (additional low exposure)
• the risk of accidental exposure is low (dioxin emissions will increase only if burning process is working improperly)
• health effects of long-term exposure are relevant
• effects on development and endocrine functions are more relevant than the risk of cancer
• The health effects of low doses should be modelled from animal and human data. Eg. Alaluusua et al. (1996) have studied tooth development. In a study by Miettinen et al. (2005), exposure to 0.5 μg TCDD/kg body weight on GD 15 resulted in maternal adipose tissue concentration of 2185 pg/g fat. In that study, linear extrapolation of the data predicts a maternal adipose tissue concentration of 100-120 pg/g fat after exposure to 0.03 μg TCDD/kg body weight. This estimated maternal adipose tissue concentration is sufficient to induce developmental dental defects in rat offspring, and is similar to the highest values measured in the Finnish average population (PCDD/F 145.5 pg WHO-TEQ/g fat (Kiviranta et al. 2005).

Sensitive subgroups: foetuses, newborns, young females (women below or at childbearing age), individuals with high fish consumption (e.g. fishermen), individuals working in incineration plants etc.

Tolerable daily intake (TDI): 1-4 pg/kg body weight

U.S.EPA estimates the dioxin and other cancer potencies by using cancer slope factors (CSFs). CSF for dioxins is 156000 kg*d/mg. This estimate has often been considered as overestimate, as the aim is to produce conservative assessments (where false positive is better than a false negative). It has also been suggested that dioxin cancer response is secondary to toxicity and has a threshold, below which cancer risk does not exist. World health organisation has approached dioxin risks from another endpoint, namely developmental defects. WHO has assessed that there is a threshold below which developmental defect risk is negligible. A tolerable daily intake is 1-4 pg/kg/d, and the average Finnish average is close to this or slightly below. In this draft assessment we will use both approaches. The U.S.EPA approach gives a higher value for risks and thus produces the plausible upper end of the confidence interval.

Effective dose resulting in a 0.01 increase in lifetime risk of cancer mortality (ED₀₁): 45 ng/kg body weight (95% CI 21-324 ng/kg body weight).

= 0.01/(45 pg/(kg body weight * 200 g body fat/kg body weight))

= 0.044 /(pg/g body fat)

Dental defects

Some of the content was previously in Heande.

Exposure-response functions for tooth defects caused by TCDD (study-specific) describes study-specific exposure-response functions for either enamel defects in molars or missing or smaller molars.

What is the quantitative relationship between exposure to TCDD during infancy and childhood and the risk (probability) of developmental dental defects described as defects of tooth enamel? Exposure is expressed in terms of concentration of TCDD in serum lipid.

Seveso children

Serum dioxin concentration vs. dental defects (Alaluusua et al. 2004, Seveso children study)

• ERS = 0.26 (± 0.12)

Transformation between serum concentration and intake:

Failed to parse (SVG (MathML can be enabled via browser plugin): Invalid response ("Math extension cannot connect to Restbase.") from server "https://wikimedia.org/api/rest_v1/":): {\displaystyle C_s = \frac{I * t_{1/2} * f}{a * ln2 * BF%},}

where
Cs = serum concentration of dioxin in pg/g fat
I = average daily intake of dioxin in pg/kg/day
t1/2 = the half-life of dioxin (2737.5 d = 7.5 a)
f = fraction of ingested dioxin actually absorbing from the gut (0.80)
BF% = body fat percentage
a percentage of total daily dietary intake of dioxins that come from fish (0.86).

The previous equation applies in a single individual. In the case of dental aberrations, the main exposure comes from the mother during pregnancy and breast feeding. For this, we use

Failed to parse (SVG (MathML can be enabled via browser plugin): Invalid response ("Math extension cannot connect to Restbase.") from server "https://wikimedia.org/api/rest_v1/":): {\displaystyle C_{s,i} = \frac{I_{a,m} * t_{1/2,m} * f_m * FE}{ln2 * BF_i},}

where
Cs,i = serum concentration of dioxin in the infant in pg/g fat
Ia,m = average daily intake of dioxin of the mother in absolute amounts pg/day
t1/2,m = the half-life of dioxin in the mother (2737.5 d = 7.5 a)
fm = fraction of ingested dioxin actually absorbing from the gut in the mother (0.80)
FE = fraction of mother's dioxin load that is transported to the infant during breast feeding (0.25) Vartiainen et al. REF
BF = body fat amount in the infant (into which the dioxin is evenly distributed) during the period when tooth are sensitive 
    to defects and the exposure at its highest (ca. six months of age) (1 kg)

ERF of dioxin on dental aberrations is a continuous random variable indexed by age. It applies to the first two age groups of the Beneris population (0-2 and 2-18 years, gender combined). It has been agreed that this ERF can be applied to WHO-TEQ concentration of dioxin (PCDD/F) and dioxin-like PCB in body fat.

Probability distribution of ERF of dioxin on dental aberrations was created based data on dioxin accident in Seveso in 1976 extracted from study by Alaluusua et al. This data is summarized in a table below.

Table: Developmental defects of enamel in individuals who were children (< 5 years of age) at the time of the Seveso accident by exposure group.

Exposure group	Number of exposed individuals	Number of enamel defect cases after 25 years since Seveso accident	Serum TCDD concentration range (pq/g lipid)	Mean serum TCDD concentration (pq/g lipid)	Risk (%)
non-ABR zone	39	10		40.5	26
Exposed group 1	10	1	31-226	128.5	10
Exposed group 2	11	5	238-592	415	45
Exposed group 3	15	9	700-26000	3000	60

It has been assumed that the TCDD exposure in children from the non-ABR zone follows lognormal distribution with mean 40.5 (ng/kg in fat) and geometric standard deviation 4 while the exposure in the remaining groups is log-uniformly distributed over the range of serum levels reported in the table. The log-logistic model (with constant term) was chosen to model the relation between the log-transformed and scaled serum TCDD level and the probability of developmental defects of enamel. The independent variable used was ln(serum TCDD level+1). Probability distribution of the coefficient for the independent variable was constructed using the following approach. Let ni denote number of children in exposure group i and ri be the number of observed enamel defect cases in group i, i= 1,...,4.

1. Sample ni exposures from distribution of TCDD serum level in group i. Denote these exposures as xj, j=1,...,75.
2. Assign ri responses randomly to ni people.
3. Fit the dose-response model to simulated data, call it model p0.
4. Compute predicted probability for every person in the study, i.e. compute p0(xj).
5. Re-sample the response of each person assuming that the probability that person j responds is p0(xj), j=1,...,75.
6. Re-fit the dose-response model.
7. Iterate steps 5-6 100 times.
8. Repeat steps 1 - 7 1500 times.
9. Create the density histogram of simulated estimates of regression coefficient (only positive values are kept).
10. Fit parametric probability density function to the histogram.

The units used:

• (ng/kg in fat)^-1
• (ln(ng/kg fat))^-1

The approach described above was used to produce two different functions. First, Gradowska 2013 used logistic regression with exposure transformation log(concentration + 1). See the trait Developmental dental defects incl. agenesis in the data table. Second, a linear function P(y) = intercept + beta * x was used; P(y) is the probability of tooth defect, x is exposure (with the same transformation) and beta is the slope coefficient. See the trait Tooth defect. In this case it was assumed that there is a non-dioxin-related background that does not affect the magnitude of dioxin effect.

Show details

cases <- c(10, 1, 5, 9) # Number of children with tooth defects in different populations children <- c(39, 10, 11, 15) # Total number of children out <- data.frame() for(k in 1:1500) { r <- c( rbinom(children[1], 1, cases[1] / children[1]), rbinom(children[2], 1, cases[2] / children[2]), rbinom(children[3], 1, cases[3] / children[3]), rbinom(children[4], 1, cases[4] / children[4]) ) x <- c( # serum dioxin concentrations distributions in different groups exp(rnorm(children[1], log(40.5) - 0.5 * log(4)^2, log(4))), exp(runif(children[2], log(31), log(226))), exp(runif(children[3], log(238), log(592))), exp(runif(children[4], log(700), log(26000))) ) # for(L in 1:100) { # Does not converge nicely so this is skipped fit <- lm(r ~ log(x + 1)) # p <- pmax(0, fit$coefficients[1] + fit$coefficients[2] * log(x + 1)) # r <- rbinom(length(r), 1, p) # } out <- rbind(out, data.frame(Intercept = fit$coefficients[1], Slope = fit$coefficients[2])) } rownames(out) <- 1:nrow(out) out <- out[out$Slope >= 0 , ] xi <- exp(runif(nrow(out), 0, log(1000))) plott <- data.frame(x = xi, y = out$Slope * log(xi + 1)) ggplot(plott, aes(x = x, y = y))+geom_point()+scale_x_log10()+geom_smooth() ggplot(plott, aes(x = x, y = y))+geom_point()+geom_smooth() hist(out$Slope) # A fairly good fit for slope is Triangular(0, 0.06, 0.12) using log(TEQ pg/g in serum fat + 1) as the exposure parameter and yes/no dental defects as response)

• A previous attempt to model dental defects is here. Note that the hidden box only shows well in the edit mode.

Finnish children

Alaluusua and coworkers studied children from Finnish general population born in 1987. They estimated dioxin exposure by using area under curve:

Failed to parse (SVG (MathML can be enabled via browser plugin): Invalid response ("Math extension cannot connect to Restbase.") from server "https://wikimedia.org/api/rest_v1/":): {\displaystyle AUC = \frac{C (1 - e^{-k_e t})}{k_e},}

where
AUC = area under the curve (pg a /g)
C = concentration in mother's milk (pg /g)
ke = mother's elimination rate for dioxin during lactation (0.2877 /a)
t = time of nursing (a)

Mineralization defects of the permanent first molars.

Outcome	Number of children with exposure (pg*year/g milk fat)
	Low exposure (<8.0)	Moderate exposure (8.0-16)	High exposure (>16)
Normal	22	41	22
Mild defect in only one tooth	1	5	2
Moderate defect or mild defect in more than one tooth	0	3	4
Severe defect	0	0	2
All	23	49	30

We need to convert the AUC to mother's dioxin daily intake.

Failed to parse (SVG (MathML can be enabled via browser plugin): Invalid response ("Math extension cannot connect to Restbase.") from server "https://wikimedia.org/api/rest_v1/":): {\displaystyle C_{s,m} = \frac{I_{a,m} * t_{1/2,m} * f_m}{ln2 * BF_i},}

where
Cs,m = dioxin concentration in serum (or fat or milk) in the mother (pg/g fat)
Ia,m = average daily intake of dioxin of the mother in absolute amounts pg/day
t1/2,m = the half-life of dioxin in the mother when not nursing (2737.5 d = 7.5 a)
fm = fraction of ingested dioxin actually absorbing from the gut in the mother (0.80)
BFm = body fat amount in the mother (into which the dioxin is evenly distributed)

When C_s,m from this equation is put into the previous equation, we can solve I_a,m:

Failed to parse (SVG (MathML can be enabled via browser plugin): Invalid response ("Math extension cannot connect to Restbase.") from server "https://wikimedia.org/api/rest_v1/":): {\displaystyle I_{a,m} = \frac{ln2 * BF * AUC * k_e}{t_{1/2,m} * f_m (1 - e^{-k_e t)}},}

where we assume an average value of 0.5 a for t because we don't have data about the actual length of nursing.

Using this equation the estimated AUCs for Finnish children (4, 12, and 20 pg*a/g for groups <8.0, 8.0-16, and >16) result in long-term intakes of 38, 114, and 190 pg/d, respectively. Therefore, we can use these values in a regression analysis to find a dose-response between mother's long-term daily intake of dioxin and probability of tooth defect. The linear slope from the highest and lowest group is (0.25 - 0.04)/(190-38) = 0.001382 (pg/d)^-1.

PCB and cancer

	Upper bound slope factor	Central-estimate slope factor
High risk and persistence	2.0	1.0
Low risk and persistence	0.4	0.3
Lowest risk and persistence	0.07	0.04

In Beneris slope factor of 2 is used.

ERF of PCB on cancer indexed by variable age. It applies to adults, i.e. > 18 years old (gender combined).

The U.S. Environmental Protection Agency (US EPA) recommends using cancer slope factors (CSFs) when evaluating potential cancer risks of PCB mixtures. There are three tiers of CSFs for environmental PCBs that depend on the exposure pathway. These are: high risk and persistence, low risk and persistence, lowest risk and persistence. In each of these tiers EPA reports central and upper bound estimate of CSF. In general, central estimate slope factors are used to estimate a typical individual’s risk while upper-bound slope assure that this risk is not likely to be underestimated if the underlying model is correct.

According to the US EPA exposures via food chain are associated with the highest risk and persistence. Therefore CSFs from the first tier are recommended to be used when estimating cancer risks from food chain pathways.

(mg/kg bw/d)^-1

Calculations

+ Show code- Hide code library(OpasnetUtils) d <- opbase.data("Op_en5823") #d[["Exposure agent"]] <- "TEQ" d$Obs <- NULL colnames(d) <- gsub(" ", "_", colnames(d)) d$Result <- ifelse(d$Result == "", "0", as.character(d$Result)) ERF <- Ovariable("ERF", data = d[d$Observation == "ERF", colnames(d) != "Observation"]) threshold <- Ovariable("Threshold", data = d[d$Observation == "Threshold", colnames(d) != "Observation"]) #test1 <- EvalOutput(ERF) #oprint(test1) #test2 <- EvalOutput(threshold) #oprint(test2) #cat(colnames(test1@output)[test1@marginal]) #cat("\n") #cat(colnames(test2@output)[test2@marginal]) #cat("\n") objects.store(ERF, threshold) cat("Ovariables ERF, threshold stored.\n")

See also

• Dioxin

References

• Is the fear of dioxin cancer more harmful than dioxin? Jouko Tuomisto, Jouni T. Tuomisto: Toxicology Letters 2012.
• Alaluusua et al ETAP 1996: Polychlorinated dibenzo-p-dioxins and dibenzofurans via mother's milk may cause developmental defects in the child's teeth.
• Lancet. 1999 Jan 16;353(9148):206.

Developing teeth as biomarker of dioxin exposure. Alaluusua S, Lukinmaa PL, Torppa J, Tuomisto J, Vartiainen T.

• Mocarelli et al EHP 2008: Dioxin Exposure, from Infancy through Puberty, Produces Endocrine Disruption and Affects Human Semen Quality
• Alaluusua et al 2004: Developmental dental aberrations after the dioxin accident in Seveso.
• Crump et al. 2003. Meta-analysis of dioxin-cancer dose-response for three occupational cohorts. Environmental Health Perspectives 111 (5), 681-687.
• Kiviranta et al. Chemosphere. 2005 Aug;60(7):854-69.
• Kogevinas 2001. Human health effects of dioxins: cancer, reproductive and endocrine system effects. Human Reproduction Update 7 (3), 331-339.
• Tuomisto JT et al. Int J Cancer. 2004 Mar 1;108(6):893-900.
• Tuomisto et al. 1999. Synopsis on dioxins and PCBs. Publications of the National Public Health Institute B17/1999.
• van Leeuwen FX et.al. Chemosphere. 2000 May-Jun;40(9-11):1095-101.