Critique of Pinget et al. (2019)
In May 2019 a study was released proposing a link between titanium dioxide (TiO2) particles and inflammatory bowel diseases and bowel cancer.
FSANZ has reviewed the study and determined it does not change our previous assessment of titanium dioxide. The study's deficiencies in design and reporting limit its utility for regulatory purposes. The study's limitations mean that no conclusion can be drawn from it about titanium dioxide and infammatory bowel diseases and bowel cancer.
TiO2 has been tested in long-term carcinogenicity studies in rats and mice, in which TiO2 was fed in the diet at concentrations sufficient to cause white faeces. These studies found no evidence of inflammatory changes or induction of tumours. Read more about nanotechnology and titanium dioxide.
This study does not change FSANZ's previous assessment that food grade titanium dioxide is safe to use as a food additive as listed in the Australia New Zealand Food Standards Code. Our technical observations on the study are listed below:
- TiO2 was administered in drinking water which means the results are of uncertain relevance to consumption of TiO2 in a food matrix.
- The number of animals used in the study and for determination of various endpoints was low compared to standard toxicological testing.
- Intended dose rates of TiO2 were 0, 2, 10 and 50 mg/kg bw/day however no data on water consumption were presented, and the authors do not comment on whether the mice consumed water at the expected rate. Furthermore, homogeneity of dispersal of TiO2 in the drinking water does not appear to have been confirmed.
- No information on animal husbandry was presented including pre-study acclimatisation, or environmental conditions of temperature, humidity and light/dark cycle. The method for assigning mice to study groups, to ensure a random distribution, is not stated.
- The usual and well-established parameters for assessing toxicity in a rodent study were not described (i.e. survival, bodyweight changes, clinical observations, food consumption, water consumption, clinical pathology parameters). Histopathology of the intestines was limited to five control mice and five mice in the 50 mg/kg bw/day group, a very limited sample for a study of this type.
- Changes in expression of the genes Muc2 and Defb3 were not correlated with any histological findings and are therefore of uncertain toxicological significance.
- The authors conclude that “TiO2 contributes to increased colonic macrophages and associated cytokines" on the basis of flow cytometry, and measurement of mRNA expression associated with colonic cytokines. They also conclude that flow cytometry shows a treatment-related increase in CD8+ T lymphocytes. However the graphs presented in support of these assertions are not convincing, because there is a lack of a consistent dose-response relationship.
- Notably, two photomicrographs are shown, alleging to show a difference in crypt length in the colon of control and 50 mg/kg bw/day mice. No data are presented to address the normal control range of crypt length in mice of this age and strain and strikingly, the photomicrograph of the 50 mg/kg bw/day mouse shows healthy colon tissue with no evidence of infiltration by any type of inflammatory cell, contrary to the authors' assertion that that there is “a state of colonic inflammation in TiO2 treated mice". The lamina propria of the colon from the 50 mg/kg bw/day mouse is noticeably more oedematous than that of the control mouse and therefore it is to be expected that the crypts would be shorter as a physical effect.
- The authors conclude that dietary TiO2 does not alter the microbiota composition of the small intestine and only minor changes were observed in the colonic microbiota, primarily due to a change in relative abundance of a few taxa at the genus level. It is speculated, without providing supporting evidence, that a greater impact on microbiota may occur over longer exposure periods. Microbial communities in murine models of microbiota composition are influenced by mouse breed, housing facilities and feed source and these should be considered when discussing the robustness and reproducibility of the results and inferring potential effects on human microbiota composition. For regulatory purposes, the clinical significance and reproducibility of the changes in microbiota composition have not been clearly demonstrated and the conclusions drawn by the authors seem to be somewhat speculative and exaggerated.
- The authors claim that dietary intake of TiO2 at 50 mg/kg BW mouse causes a slight but significant decrease in plasma acetate concentration compared with no supplementation with TiO2. However, no information was provided on the normal range of plasma acetate in male mice of this age and strain and no evidence is provided to support the claim that this slight decrease may partially explain changes in mucus gene expression in mice treated at this dose of TiO2 .
- The paper states that changes in plasma concentrations of trimethylamine (TMA) and choline may be associated with dietary TiO2 at 50 mg/kg BW mouse and a shift in microbiota composition, however no information on the fasting time and blood collection protocol is provided. It is therefore difficult to conclude if the observed changes were due to the TiO2 intake and altered microbiota composition or dietary choline intake and subsequent plasma pharmacokinetics of choline and TMA. Furthermore, no information is provided on the normal range of plasma TMA and choline in male mice of this age and strain and the inference that the increased TMA levels in the 10 and 50 mg/kg BW treatment groups may be associated with the development of atherosclerosis is speculative.
- In vitro biofilm assays are relied upon to assert that TiO2 induces biofilm formation in vivo. There is a failure to consider the complex nature of the colonic environment and to take into account environmental, host and other factors that influence biofilm formation in vivo. In vitro models provide a simplified version of conditions and it is important that well designed in vivo models are developed to validate in vitro results. Furthermore, intestinal colonisation, attachment and subsequent biofilm formation is host specific and care should be taken when extrapolating results from an animal model to potential biofilm formation in humans.
- The authors make no reference to the National Toxicology Program (NTP) toxicity/carcinogenicity studies of TiO2, which are available in the public domain at https://ntp.niehs.nih.gov/ntp/htdocs/lt_rpts/tr097.pdf
- The NTP studies were conducted using anatase TiO2, the same TiO2 that Pinget et al. used. Test subjects were B6C3F1 mice, 50/sex/group, and the TiO2 was administered in the diet and therefore the NTP study was more relevant to human health risk assessment. The test article was fed in the diet at concentrations of 0, 2.5 or 5% for 103 weeks. The level of TiO2 intake of the treated groups was sufficient to cause white feces in all animals. There were no significant differences in body weights or incidence of tumours in either sex. The NTP conducted a dietary study of the same test article at the same dose levels in Fischer rats and again found no significant differences in body weights or incidence of tumours in either sex. Consumption of TiO2 had no effect on the prevalence of degenerative, inflammatory or proliferative lesions in either species.
In summary, FSANZ considers that the results presented in the paper do not support the conclusions reached by Pinget et al., and do not justify any change to FSANZ's existing position on TiO2.