Dietary Titanium Dioxide (E171) Alters the Colon Transcriptome—Evidence From a Human Dietary Intervention Study Authors: Nicolaj S Bischoff, Anna K Undas, Greet van Bemmel, Marcel van Herwijnen, Marcha Verheijen, Simone G Van Breda, Jacco J Briedé, Lana Drenth‐Brink, Marcella Peelen‐Buijs, Iris Elferink, Héloïse Proquin, Henk van Loveren, Ad A van Bodegraven, Dick T H M Sijm, Theo M de Kok Published: 2026 ABSTRACT ABSTRACTFood‐grade titanium dioxide (E171) has been removed from the European food market due to concerns about its potential genotoxicity, yet human evidence remains limited. Given its continuous use outside the European Union, clarifying its potential hazard is essential for human risk assessment. In a randomized crossover study, 31 adults consumed 2 mg/kg body weight/day of E171 for 2 weeks, with biospecimens collected after both control and exposure periods. Oral intake resulted in elevated titanium concentrations in feces (CTRL: 6.3 mg/kg; E171: 377 mg/kg). Particle characterization showed a median size of 228 nm, with 9% <100 nm. Systemic oxidative stress increased, evidenced by increased superoxide radical levels in whole blood (p = 0.057). Transcriptomic profiling of colon biopsies revealed activation of 73 pathways linked to oxidative stress, cellular metabolism, and colorectal cancer‐associated processes. The observed transcriptional changes were consistent with findings of its proposed mode of action. This human intervention study characterizes the effects of dietary E171 following human gastrointestinal digestion and provides evidence that exposure to dietary E171 induces systemic oxidative stress and transcriptional changes in the colon. These findings strengthen regulatory concerns about E171 as a food additive and its potentially harmful effects on human gastrointestinal health.Keywords: colorectal cancer, E171, human dietary intervention study, RNA sequencing, titanium dioxide Abbreviations 2,2′‐Azobis(2‐methylpropionamidine) dihydrochloride 2,2′‐Azino‐bis(3‐ethylbenzothiazoline‐6‐sulfonic acid) Bicinchoninic acid Body mass index Confidence Interval 1‐Hydroxy‐3‐methoxycarbonyl‐2,2,5,5‐tetramethylpyrrolidine Counts per million Colorectal cancer Control Deoxyribonucleic acid Food‐grade titanium dioxide Ethylene diamine tetra acetic acid European Food Safety Authority Electron microscopy Electron spin resonance Food and Drug Administration False discovery rate High‐resolution induced‐coupled‐plasma mass spectrometer High‐sensitivity C‐reactive protein Food‐grade titanium dioxide Kyoto Encyclopedia for Genes and Genomes Mode of action Maastricht University Medical Center + RNA integrity number Reactive oxygen species Ribosomal RNA Standard deviation Single particle inductively coupled plasma mass spectrometer Tri chloric acid Trolox equivalent antioxidant capacity Transmission electron microscopy T helper cells 1 T helper cells 17 Titanium dioxide University Medical Center Utrecht Introduction Titanium dioxide (TiO2), widely used as a food additive (E171 in the EU and Asia, and INS171 in North America), has historically been considered safe due to its assumed inertness [1, 2]. Its high refractive index and brightness enhance the visual appeal of processed foods such as confectionery, sauces, and baked goods [1, 3, 4]. E171 is also commonly used in pharmaceuticals as an excipient because of its UV‐protective properties [5]. However, emerging evidence has challenged the assumed inertness, particularly regarding its potential effects on gastrointestinal health. In response to safety concerns, the European Food Safety Authority (EFSA) reassessed E171, and the European Union subsequently banned its use in food products in 2022, citing evidence of genotoxicity [6]. In contrast to the EU's decision, TiO2 remains permitted for use as a food additive outside Europe. Health Canada concluded that the available evidence does not warrant human health concerns, while noting that food–matrix interactions, poor oral bioavailability, and differences between dietary exposure and dispersed particles in toxicological tests affect the interpretation of evidence [7]. The United States Food and Drug Administration likewise permits TiO2 as a food additive under specified conditions and reports no evidence of genotoxicity concerns based on currently available data [8]. Despite these regulatory differences, in recent studies, it has been shown that E171 retains its toxicological properties even after digestion [9, 10]. Animal models indicate that oral E171 exposure promotes inflammation, DNA damage, and gut microbiome alterations, with some studies linking it to early CRC‐related changes [11, 12, 13, 14, 15, 16, 17]. Furthermore, by toxicogenomic analyses, transcriptional shifts in colonic tissues have been demonstrated, suggesting that E171 may be involved in pathways related to tumorigenesis and immune dysregulation [15, 18, 19]. However, most research has been limited to in vitro and animal studies, leaving critical questions about its effects on humans unresolved. A notable gap in the scientific literature is the scarcity of human data. In a recent US cohort study, Ti levels in stool samples were detected, confirming dietary exposure to E171, but its physicochemical properties following (oral) digestion were not assessed [20]. In a follow‐up study, elevated fecal TiO2 levels were linked with increased gut permeability and inflammation markers, suggesting potential adverse effects on gastrointestinal health [21]. These findings underscore the need for human‐centered data to clarify the impact of E171 on gut homeostasis, intestinal inflammation, and overall gastrointestinal health. To address this data gap, we conducted a 4‐week randomized crossover dietary intervention study in healthy volunteers, analyzing colorectal biopsies alongside whole blood, plasma, and feces to assess E171's effects on human gastrointestinal and systemic health. In this study, we examined fecal Ti levels, systemic biomarkers of inflammation and oxidative stress, as well as transcriptomic changes, in colonic tissues, providing a human dataset on E171's direct impact on the gut epithelium. By comparing our results with a postulated mode of action (MoA), we offer critical insights into the potential mechanistic causes of E171‐induced adverse effects, providing valuable information for regulators and public health policymakers. Experimental Section Study Participants This human dietary intervention study was performed according to the Helsinki Declaration of Ethical Principles for Medical Research Involving Human Subjects, following approval by the Medical Ethical Review Committee of the Maastricht University Medical Centre + (MUMC+, registration number: NL52433.068.15/METC163010) and the Office of Scientific Research, RVE Zuyderland Academy, Zuyderland (registration number: Z2022015). The inclusion and exclusion criteria are listed in Supporting Information 1. The study is registered with the International Clinical Trials Registry Platform (ICTRP) under the identifier NL8534. Patients or the public were not involved in the design, conduct, reporting, or dissemination of this research. Informed consent was obtained from all participating subjects. Human Dietary Intervention Study Design In our study, a randomized crossover design was used to assess the impact of dietary exposure to E171 on gastrointestinal and systemic health. The study spanned 4 weeks. Each participant was randomly allocated to the exposure (E171) or the control group (CTRL). Following 2 weeks of intervention, feces, whole blood, plasma, and rectal biopsies were collected, and the participants were subsequently assigned to the other group (Supporting Information 2). This was followed by another 2 weeks upon completion of the study, during which resampling and completion of the patient forms were performed (Figure 1). Each participant in the exposure group received 2 mg/kg body weight/day of E171 mixed into yogurt (Supporting Information 3). The exposure concentration was derived from the mean non‐brand‐loyal exposure scenario for adults (18–64 years), published by EFSA in 2021 [22]. Randomized cross‐over study design. The study participants were allocated to either the control group (CTRL) or the exposure group (E171) for 2 weeks, after which samples were collected, and their allocation was changed. Physicochemical Characterization of E171 and Intervention Product E171 was kindly donated by Sensient Technologies (Mexico). Pristine E171 and E171 in yogurt have been characterized via single particle‐induced coupled plasma mass spectrometry (spICP‐MS) and transmission electron microscopy (TEM), as previously described [23, 24, 25] (Supporting Information 4). Feces Analysis: HrICP‐MS and spICP‐MS Feces were analyzed for total Ti concentration via high‐resolution inductively coupled plasma mass spectrometry (hrICP‐MS) and E171's physicochemical properties via spICP‐MS (Supporting Information 4) [26]. Protein Quantification A bicinchoninic acid (BCA) protein assay was performed according to the manufacturer's protocol to quantify protein content in plasma (Thermo Fisher Scientific, Massachusetts, USA). Systemic Markers for Oxidative Stress TEAC Assay The well‐established Trolox Equivalent Antioxidant Capacity (TEAC) assay was used to measure the antioxidant capacity of plasma based on the ability of antioxidants to quench the 2,2′‐azino‐bis(3‐ethylbenzothiazoline‐6‐sulfonic acid) (ABTS•+) radical cation (Sigma‐Aldrich, A1888‐1G) (Supporting Information 5). Electron Spin Resonance Spectroscopy Electron spin resonance (ESR) spectroscopy is a highly sensitive and specific technique for detecting different types of ROS by utilizing spin‐trapping agents that form stable complexes with these short‐lived radicals [27]. The spin‐probe CMH was used to detect superoxide in whole blood (Supporting Information 6). Protein Carbonyl Content Protein carbonyl content (PCC) was determined to measure plasma protein oxidation. DNPH binds to carbonyls, forming a yellow DNP hydrazone. Unbound DNP is washed away with TCA, and absorbance was measured at 562 nm (Supporting Information 7). Systemic Markers for Inflammation Hs‐CRP High‐sensitivity C‐reactive protein (hs‐CRP) was measured in plasma samples using nephelometry to assess inflammatory status (Supporting Information 8). This analysis was conducted at the Central Diagnostic Laboratory of MUMC+, the Netherlands. Cytokine Panel A cytokine panel, including IL‐1β, IL‐6, IL‐8, IL‐10, IL‐17, IL‐18, IFN‐α, IFN‐γ, and TNF‐α, was analyzed in plasma, using a multiplex immunoassay platform based on Luminex xMAP technology at the MultiPlex Core Facility, University Medical Center Utrecht (UMCU, Supporting Information 9). Fecal Calprotectin To analyze intestinal inflammation, fecal calprotectin was measured. Ten fecal samples were selected for this analysis based on the most significant differences in E171 concentration between the exposure and control periods and the largest changes in calprotectin gene expression. Calprotectin was extracted from stool samples using the EliATM Stool Extraction Kit Plus (Thermo Fisher, Uppsala, Sweden) and quantified using the EliA 2 calprotectin assay on the Phadia 250 system, following the manufacturer's instructions. This analysis was conducted at the Central Diagnostic Laboratory of MUMC+. Transcriptomic Study RNA Isolation RNA isolation from rectal biopsies was performed using the miRNeasy Micro kit according to the manufacturer's instructions (Supporting Information 10). RNA concentration was measured using a Qubit, and RNA integrity number (RIN) was determined using either a nano‐ or pico‐chip on the Bioanalyzer (Agilent 2100 Bioanalyzer). The mean RNA concentration of all 62 samples was 484 ± 280.8 ng/mL with a RIN of 6.3 ± 1.1. Library Preparation and RNA Bulk Sequencing Samples containing purified RNA were prepared for sequencing using a ribosomal RNA (rRNA) depletion library preparation kit (Revvity, Groningen, the Netherlands) on a Zephyr G3 NGS workstation (CLS150362, Revvity, Groningen, the Netherlands). After library preparation, the samples were sequenced on an S2 flow cell (Illumina, San Diego, USA) using the NovaSeq 6000 system (Illumina, San Diego, USA) with the Rapid Directional RNA‐seq Kit 2.0 (Revvity, Groningen, the Netherlands). Pre‐Processing and Data Analysis Raw RNA sequencing data were obtained as BCL files and converted to fastq files using BCL2FastQ (v2.20.0.422). Fastq files were processed using the R‐ODAF pipeline [28] (Supporting Information 11). Gene Set Enrichment Analysis (GSEA) After preprocessing, all genes that passed the R‐ODAF filters were included in gene set enrichment analysis (GSEA) [28]. Genes were ranked using the stat value derived from the differential expression analysis, which integrates both the magnitude of expression change and the statistical evidence for differential expression between conditions. GSEA was performed in R using the clusterProfiler package with KEGG pathway annotations as the reference gene set database [29]. Statistical significance of enriched pathways was assessed using false discovery rate (FDR) correction according to the Benjamini–Hochberg procedure, and pathways with an adjusted p‐value (FDR) below 0.05 were considered statistically significant. Statistical Analysis Biomarker data were analyzed using GraphPad Prism (version 1, GraphPad Software, San Diego, CA, USA). All comparisons were strictly paired, reflecting the randomized, crossover study design, in which each participant served as their own control during the control (CTRL) and intervention (E171) periods. If one value in a pair was missing, the corresponding pair was excluded from the analysis using a complete‐case approach. Prior to statistical testing, data quality was assessed, including screening for potential outliers using robust regression and the outlier removal (ROUT) method (Q = 1%). Although this procedure identified potential extreme values, no observations were excluded solely on this basis. Flagged outliers were additionally compared with the case report forms; if no adverse event was documented and no analytical, procedural, or data‐entry errors were documented, the flagged outlier was not removed and was deemed biologically plausible. Consequently, all observations were retained to preserve the integrity of paired data. To maintain the benefit of randomization and avoid bias associated with post hoc data exclusion. Normality of the paired samples was assessed using the Shapiro–Wilk test for groups with fewer than 60 observations. Normality assessment was complemented by visual inspections of Q–Q plots and histograms and by evaluating skewness using the adjusted Fisher–Pearson coefficient of skewness to guide transformation decisions. Normally distributed data were analyzed by using a two‐tailed paired t‐test, and results are presented as mean ± standard deviation (SD). For data that were not normally distributed and exhibited right skewness, a log transformation was applied. If the distribution of paired differences approximated normality after log‐transformation, a paired t‐test was conducted on the transformed data, and results are reported as geometric means with 95% confidence intervals (CIs). If normality was not achieved after transformation, a nonparametric Wilcoxon signed‐rank test was performed, and the results are presented as medians with 95% CIs. All statistical tests were two‐tailed, and a p‐value < 0.05 was considered statistically significant. Results Participants In total, 34 healthy, participants were included in the study, and 31 completed the intervention between May 2022 and February 2024. Figure 2 illustrates the flow chart for study participant inclusion and exclusion, along with the reasons for exclusion/drop‐out. Drop‐outs occurred only due to scheduling issues, since the study's start was delayed by COVID‐related restrictions and limited access to hospital wards. Flow chart of participant inclusion, enrollment, and allocation details. A total of 31 participants finalized the human dietary intervention study. The mean sex distribution of participants was 42% male and 58% female, with a mean age of 31.2 ± 14.1 years (min: 18, max: 59) and a mean body mass index (BMI) of 23.1 ± 2.0 kg/m2 (Table 1). Body weight, caloric intake, macronutrient distribution (Figure S1), and mineral levels remained unchanged throughout the study (Table S2). Food diaries were reviewed for potential E171‐containing products, but none were identified. TABLE 1Description of the participant's age, height, and body weight, including BMI (mean ± SD).Age (years)Height (cm)Body weight (kg)BMI (kg/m2)Total (n = 31)31.2 ± 14.1173.6 ± 9.069.9 ± 9.323.1 ± 2.0Male (n = 13)34.4 ± 13.8182.3 ± 3.476.9 ± 7.223.1 ± 1.8Female (n = 18)28.9 ± 14.4167.3 ± 5.964.8 ± 7.023.1 ± 2.1 Description of the participant's age, height, and body weight, including BMI (mean ± SD). Physicochemical Characterization of the Intervention Product Feces Analysis: HrICP‐MS and spICP‐MS We measured the total Ti concentration in wet feces to assess participant compliance with the study protocol and to determine whether any additional exposure to E171 was attributable to the diet. Fecal samples were analyzed by hrICP‐MS, and a statistically significant increase in E171 (p‐value < 0.0001) was observed during the intervention period, with a median E171 concentration of 377 mg Ti/kg feces (Figure 3A). A median Ti concentration of 6.2 mg/kg was detected during the control period, which is considered to be within natural background levels. The physicochemical characteristics of E171 were determined using spICP‐MS, and a median particle size of 79 nm was observed for pristine E171 (Table 2). E171 dispersed in yogurt had a median size of 282 nm, whereas E171 detected in the feces of the human volunteers had a median size of 228 nm. The nanoparticle fraction decreased from 75% to 15% to 9% in pristine, E171‐yogurt, and E171‐feces, respectively (Table 2). Furthermore, spICP‐MS was used to confirm the presence of Ti in feces by comparing the size‐distribution histograms of the CTRL and E171 periods (Figure 3B). (A) Total titanium (Ti) concentrations in feces (mg/kg wet weight) quantified by hrICP‐MS. Ti levels were significantly higher during the E171 exposure period than during the CTRL period, indicating compliance with the intervention. Low Ti concentrations observed during the CTRL period further confirm the absence or effective washout of E171 prior to exposure. Data are presented as medians with 95% confidence intervals; statistical significance was assessed using a two‐tailed Wilcoxon rank‐sum test (each data point represents one biological replicate; n = 27). (B) Kernel density‐based particle size distribution of titanium‐containing particles. Grey curves represent samples collected during the CTRL period, whereas black curves correspond to samples obtained during the E171 exposure period. Distributions illustrate a shift in particle‐size profiles between conditions. TABLE 2Overview of the physicochemical characterization of E171 in its pristine state in E171 yogurt and participants' feces.SampleTechniqueMedian size (nm)% of particles < 100 nmPristine E171spICP‐MS7875Pristine E171TEM7973E171‐yogspICP‐MS28215E171‐yogTEMNANAE171‐fecesspICP2289E171‐fecesTEMNANAAbbreviation: NA, not applicable. Overview of the physicochemical characterization of E171 in its pristine state in E171 yogurt and participants' feces. Abbreviation: NA, not applicable. Protein Quantification We quantified plasma protein levels to standardize our assays, ensuring consistency and accuracy by accounting for potential variability in blood volume and plasma protein concentration. In the BCA assay, we found a mean protein concentration of ∼45.000 µg/mL in both groups (Figure 4). As no significant difference in plasma concentration was observed between groups, no additional correction for plasma dilution was applied for subsequent analysis. Effects of E171 exposure on plasma protein concentration, antioxidant capacity, oxidative stress, and inflammatory markers. Scatterplots display individual paired measurements for the control period (CTRL, blue) and exposure period (E171, red). Data are presented as geometric means with 95% confidence intervals (CI). Total protein concentration (each data point represents one biological replicate; n = 31), Trolox equivalent antioxidant capacity (TEAC; each data point represents one biological replicate; n = 31), and trichloroacetic acid‐precipitated TEAC (TCA‐TEAC; each data point represents one biological replicate; n = 31) did not differ between CTRL and E171 conditions. Paired comparisons using log‐transformed ratio t‐tests showed no significant effects: protein (geometric mean ratio: 1.003, 95% CI: 0.976–1.030), TEAC (geometric mean ratio: 0.979, 95% CI: 0.941–1.019), and TCA‐TEAC (geometric mean ratio: 1.016, 95% CI: 0.835–1.237). Superoxide levels (each data point represents one biological replicate; n = 31) exhibited a trend toward increased oxidative stress following E171 exposure (geometric mean ratio: 1.175, 95% CI: 0.995–1.389), although this did not reach statistical significance (p = 0.0570, log‐transformed paired t‐test). Protein carbonyl content (PCC; each data point represents one biological replicate; n = 27), a marker of protein oxidation, showed no difference between conditions (geometric mean ratio: 1.255, 95% CI: 0.771–2.028). Low‐grade systemic inflammation, assessed by high‐sensitivity C‐reactive protein (hs‐CRP; each data point represents one biological replicate; n = 29), was not significantly altered between groups (p = 0.0755), although a trend toward increased levels was observed during E171 exposure (geometric mean ratio: 1.498, 95% CI: 0.957–2.346). Systemic Oxidative Stress Markers TEAC and TCA‐TEAC We determined TEAC and TCA‐TEAC as systemic markers for the overall antioxidative capacity of our participants. No statistically significant difference was detected between the control and E171‐exposed groups in the TEAC or TCA‐TEAC assays (Figure 4). ESR Spectroscopy ESR measurements of oxidative stress indicated a trend toward increased superoxide radical concentrations in whole blood following 2 weeks of E171 exposure, approaching statistical significance (p = 0.0570). This observation suggests a potential elevation in systemic oxidative stress (Figure 4). PCC We determined PCC as a marker for protein oxidation. No statistically significant difference was detected between the control and E171‐exposed groups in the PCC assay (Figure 4). Systemic Inflammation Markers Hs‐CRP We analyzed hs‐CRP to detect early signs of increased systemic inflammation. Following 2 weeks of dietary E171 exposure, no statistically significant difference in hs‐CRP levels (p‐value: 0.0755) was detected, although a trend toward an increase was observed (Figure 4). Cytokine Panel To further analyze systemic inflammation, a cytokine panel was measured. None of the eight tested cytokines showed statistically significant differences in the plasma concentration following 2 weeks of dietary E171 exposure (Figure 5). Effects of E171 exposure on circulating cytokine levels. Scatterplots display individual paired measurements for the control period (CTRL, blue) and exposure period (E171, red), with horizontal bars representing geometric means ± 95% confidence intervals (CI). Cytokine concentrations are expressed in pg/mL; all panels include n = 30. Each data point represents one biological replicate. Pro‐inflammatory cytokines, including interleukin (IL)‐1β, IL‐6, IL‐8, tumor necrosis factor‐α (TNF‐α), and interferon‐γ (IFN‐γ), showed no statistically significant differences between CTRL and E171 conditions. Similarly, anti‐inflammatory and regulatory cytokines, including IL‐10, IL‐17, IL‐18, and interferon‐α (IFN‐α), were not significantly altered following E171 exposure. All cytokines were analyzed using paired statistical approaches (two‐tailed paired t‐tests or log‐transformed paired t‐tests, depending on distributional assumptions), and none reached statistical significance (p > 0.05 for all comparisons). Overall, E171 exposure did not induce measurable changes in circulating cytokine profiles under the conditions tested, indicating no detectable modulation of systemic inflammatory or immune signaling pathways in this cohort. Fecal Calprotectin There was no statistically significant difference in fecal calprotectin concentration between the control and exposure group (Figure S2). GSEA of Bulk RNA Sequencing Following pre‐processing with the R‐ODAF pipeline, 18 021 genes were present in the dataset. Using GSEA, we identified 73 statistically significant (FDR < 0.05) upregulated KEGG pathways after 2 mg/kg body weight/day of E171 for 2 weeks, compared with the control group (Table 3). These pathways included translation (Ribosome), cellular metabolism (Oxidative phosphorylation, Metabolic pathway, Carbon metabolism pathway, AMPK, Gluconeogenesis, Citrate cycle (TCA cycle), AMPK signaling pathway), DNA damage and cancer (Chemical carcinogenesis, Reactive oxygen species, Colorectal cancer, Viral carcinogenesis), cell proliferation (mTOR signaling pathway, FoxO signaling pathway), Oxidative stress (Glutathione metabolism, Chemical carcinogenesis—reactive oxygen species, Peroxisome), Short‐chain‐fatty acid metabolism (Propanoate metabolism, Butanoate metabolism), and signaling pathways (PPAR signaling pathway, FoxO signaling pathway, mTOR signaling pathway, AMPK signaling pathway). Following E171 exposure, no pathways were downregulated. Figure S4 depicts the CRC pathways and the effect genes that were up‐ or downregulated following 2 weeks of exposure to E171 at 2 mg/kg body weight/day, as identified via GSEA. Table 4 maps the statistically significant changes in transcriptional pathways to the postulated MoA [2, 30]. Correlation of the identified transcriptional pathway changes and the proposed MoA includes cellular uptake, the generation of ROS, and an increase in oxidative stress, local inflammation, epithelial injury, changes in the microbiome, DNA damage/genotoxic stress, increased proliferation, and the activation of the CRC pathway. We additionally mapped the transcriptional changes in the CRC pathway following E171 exposure on the identified KEGG pathway, highlighting the activation and suppression of key genes in CRC development (Figure S5). TABLE 3Significantly enriched KEGG pathways identified by GSEA, ranked by normalized enrichment score (NES).IDClassCategoryDescriptionNES p‐Value q‐Valuehsa03010Genetic information processingTranslationRibosome2.828631.00E‐102.86E‐09hsa05322Human diseaseImmune systemSystemic lupus erythematosus2.597981.00E‐102.86E‐09hsa00190MetabolismEnergy metabolismOxidative phosphorylation2.562571.00E‐102.86E‐09hsa05034Human diseaseSubstance dependenceAlcoholism2.328571.07E‐102.86E‐09hsa05208Human diseaseCancerChemical carcinogenesis—reactive oxygen species2.235161.00E‐102.86E‐09hsa05010Human diseaseNeurodegenerative diseaseAlzheimer disease2.215281.00E‐102.86E‐09hsa05016Human diseaseNeurodegenerative diseaseHuntington disease2.205721.00E‐102.86E‐09hsa05012Human diseaseNeurodegenerative diseaseParkinson disease2.191961.00E‐102.86E‐09hsa01100MetabolismEnergy metabolismMetabolic pathways2.105991.00E‐102.86E‐09hsa05022Human diseaseNeurodegenerative diseasePathways of neurodegeneration—multiple diseases2.002631.00E‐102.86E‐09hsa05014Human diseaseNeurodegenerative diseaseAmyotrophic lateral sclerosis2.021911.34E‐103.27E‐09hsa05020Human diseaseNeurodegenerative diseasePrion disease2.087437.81E‐101.74E‐08hsa05171Human diseaseInfectious disease: viralCoronavirus disease—COVID‐192.158592.10E‐094.32E‐08hsa00280MetabolismAmino acid metabolismValine, leucine and isoleucine degradation2.613973.05E‐095.82E‐08hsa05415Human diseaseCardiovascular diseaseDiabetic cardiomyopathy2.094945.77E‐091.03E‐07hsa04932Human diseaseEndocrine and metabolic diseaseNon‐alcoholic fatty liver disease2.218171.23E‐082.05E‐07hsa04714Organismal systemsEnvironmental adaptationThermogenesis1.990673.75E‐085.89E‐07hsa04141Genetic information processingFolding, sorting, and degradationProtein processing in endoplasmic reticulum2.012141.38E‐072.05E‐06hsa04140Cellular transportTransport and catabolismAutophagy—animal1.976052.53E‐073.57E‐06hsa01200MetabolismCarbohydrate metabolismCarbon metabolism2.172892.94E‐073.74E‐06hsa04142Cellular transportTransport and catabolismLysosome2.069332.91E‐073.74E‐06hsa04613Organismal systemsImmune systemNeutrophil extracellular trap formation1.967693.70E‐074.49E‐06hsa04723Organismal systemsNervous systemRetrograde endocannabinoid signaling1.927411.12E‐050.00013hsa04145Cellular transportTransport and catabolismPhagosome1.885102.07E‐050.00023hsa05017Human diseaseNeurodegenerative diseaseSpinocerebellar ataxia1.840892.29E‐050.00024hsa00640MetabolismCarbohydrate metabolismPropanoate metabolism2.212352.41E‐050.00025hsa00100MetabolismLipid metabolismSteroid biosynthesis2.146280.000110.00107hsa01212MetabolismLipid metabolismFatty acid metabolism1.999150.000130.00122hsa00650MetabolismCarbohydrate metabolismButanoate metabolism2.126930.000160.00140hsa00600MetabolismLipid metabolismSphingolipid metabolism2.034120.000160.00140hsa00520MetabolismCarbohydrate metabolismAmino sugar and nucleotide sugar metabolism2.016150.000150.00140hsa04260Organismal systemsCirculatory systemCardiac muscle contraction1.906150.000220.00180hsa00620MetabolismCarbohydrate metabolismPyruvate metabolism1.988470.000270.00221hsa00785MetabolismMetabolism of cofactors and vitaminsLipoic acid metabolism2.027630.000510.00380hsa00480MetabolismMetabolism of other amino acidsGlutathione metabolism1.933280.000530.00380hsa04137Cellular transportTransport and catabolismMitophagy—animal1.726930.000500.00380hsa04120Genetic information processingFolding, sorting, and degradationUbiquitin‐mediated proteolysis1.632350.000510.00380hsa04216Cellular processesCell growth and deathFerroptosis1.967970.000590.00412hsa04068Environmental information processingSignal transductionFoxO signaling pathway1.665900.000600.00412hsa00900MetabolismMetabolism of terpenoids and polyketidesTerpenoid backbone biosynthesis2.029670.000690.00462hsa00510MetabolismGlycan biosynthesis and metabolismN‐Glycan biosynthesis1.849300.000730.00476hsa00513MetabolismGlycan biosynthesis and metabolismVarious types of N‐glycan biosynthesis1.966130.000870.00555hsa00270MetabolismAmino acid metabolismCysteine and methionine metabolism1.877480.000900.00562hsa04144Cellular transportTransport and catabolismEndocytosis1.509600.000980.00598hsa05203Human diseaseCancerViral carcinogenesis1.519190.001030.00613hsa00630MetabolismCarbohydrate metabolismGlyoxylate and dicarboxylate metabolism2.000110.001070.00615hsa00010MetabolismCarbohydrate metabolismGlycolysis/Gluconeogenesis1.811580.001100.00615hsa04217Cellular processesCell growth and deathNecroptosis1.586300.001100.00615hsa04713Organismal systemsEnvironmental adaptationCircadian entrainment1.691360.001190.00650hsa00410MetabolismMetabolism of other amino acidsbeta‐Alanine metabolism1.974400.001440.00769hsa00512MetabolismGlycan biosynthesis and metabolismMucin type O‐glycan biosynthesis1.887830.001630.00853hsa04136Cellular processesTransport and catabolismAutophagy—other1.855250.001660.00853hsa00062MetabolismLipid metabolismFatty acid elongation1.934040.001850.00911hsa00020MetabolismCarbohydrate metabolismCitrate cycle (TCA cycle)1.903680.001840.00911hsa01250Biosynthesis of nucleotide sugarsNABiosynthesis of nucleotide sugars1.835400.001870.00911hsa00260MetabolismAmino acid metabolismGlycine, serine and threonine metabolism1.863650.002120.01012hsa04710Organismal systemsEnvironmental adaptationCircadian rhythm1.867070.002340.01099hsa04150Environmental information processingSignal transductionmTOR signaling pathway1.503150.003390.01563hsa01524Human diseaseDrug resistance: antineoplasticPlatinum drug resistance1.633660.004260.01931hsa05110Human diseaseInfectious disease: bacterialVibrio cholerae infection1.737980.004620.02058hsa01210NANA2‐Oxocarboxylic acid metabolism1.799440.004850.02125hsa04972Organismal systemsDigestive SystemPancreatic secretion1.587980.005790.02495hsa03320Organismal systemsEndocrine systemPPAR signaling pathway1.609270.005940.02502hsa01240NANABiosynthesis of cofactors1.464300.005990.02502hsa04146Cellular processesTransport and catabolismPeroxisome1.545420.007020.02888hsa04727Organismal systemsNervous systemGABAergic synapse1.560260.008070.03269hsa01040MetabolismLipid metabolismBiosynthesis of unsaturated fatty acids1.758320.008430.03366hsa04960Organismal systemsExcretory systemAldosterone‐regulated sodium reabsorption1.714790.009100.03460hsa04979Organismal systemsDigestive systemCholesterol metabolism1.689240.009190.03460hsa01230Biosynthesis of amino acidsNABiosynthesis of amino acids1.591250.009160.03460hsa04152Environmental information processingSignal transductionAMPK signaling pathway1.459520.009040.03460hsa04726Organismal systemsNervous systemSerotonergic synapse1.510560.009780.03583hsa05210Human diseaseCancerColorectal cancer1.479470.009660.03583 Significantly enriched KEGG pathways identified by GSEA, ranked by normalized enrichment score (NES). TABLE 4Mapping of significantly enriched KEGG pathways from GSEA analysis to the proposed MoA, molecular initiating events, and key events (KE), and their mechanistic interpretation following E171 exposure.AOP key eventKEGG pathways (NES)Mechanistic interpretationKE1: Cellular uptake/particle handling Endocytosis (NES ∼1.52)Phagosome (NES ∼1.60)Lysosome (NES ∼1.53)Autophagy (NES ∼1.46–1.53)Mitophagy (NES ∼1.48) Particle internalization and intracellular trafficking; activation of vesicular processing and organelle turnoverKE2: ROS generation Chemical carcinogenesis—ROS (NES ∼1.73)Oxidative phosphorylation (NES ∼1.70)Peroxisome (NES ∼1.57)Glutathione metabolism (NES ∼1.59)Ferroptosis (NES ∼1.53) Mitochondrial and peroxisomal ROS production; initiation of oxidative imbalanceKE3: Oxidative stress/redox imbalance Glutathione metabolism (NES ∼1.59)Cysteine and methionine metabolism (NES ∼1.51)Beta‐alanine metabolism (NES ∼1.37)Peroxisome (NES ∼1.57) Activation of antioxidant defense systems and redox bufferingKE4: Epithelial injury/metabolic rewiring Metabolic pathways (NES ∼1.52)Carbon metabolism (NES ∼1.50–1.57)TCA cycle (NES ∼1.50–1.58)Glycolysis/gluconeogenesis (NES ∼1.50)Pyruvate metabolism (NES ∼1.49) Cellular stress response reflected in metabolic reprogramming and energy redistributionKE5: Barrier/microbiome changes Mucin‐type O‐glycan biosynthesis (NES ∼1.84) N‐Glycan biosynthesis (NES ∼1.30)Butanoate metabolism (NES ∼1.22)Propanoate metabolism (NES ∼1.22)PPAR signaling (NES ∼1.30–1.33) Altered mucus barrier, glycosylation, and host–microbiome metabolic interactionsKE6: Local inflammation (subclinical) Neutrophil extracellular trap formation(NES ∼1.35) Necroptosis (NES ∼1.35)PPAR signaling (NES ∼1.30–1.33)Infection‐related pathways (NES ∼1.30–1.60) Potential innate immune activation; weak systemic support (cytokines NS, calprotectin NS)KE7: DNA damage/genotoxic stress Chemical carcinogenesis—ROS (NES ∼1.73)Colorectal cancer (NES ∼1.48)Viral carcinogenesis (NES ∼1.50)ER protein processing (NES ∼1.47)Platinum drug resistance (NES ∼1.46) ROS‐driven DNA damage and activation of repair/survival pathwaysKE8: Proliferation/survival signaling mTOR signaling (NES ∼1.64)AMPK signaling (NES ∼1.60)FoxO signaling (NES ∼1.52–1.54)Ribosome (NES ∼1.68) Increased translational capacity and pro‐survival signaling under stressKE9: CRC pathway overlap Colorectal cancer pathway (NES ∼1.48)PI3K/MAPK modules (within multiple pathways) Convergence on canonical CRC signaling networks (Wnt, PI3K, MAPK, TGF‐β) Mapping of significantly enriched KEGG pathways from GSEA analysis to the proposed MoA, molecular initiating events, and key events (KE), and their mechanistic interpretation following E171 exposure. Endocytosis (NES ∼1.52) Endocytosis (NES ∼1.52) Phagosome (NES ∼1.60) Phagosome (NES ∼1.60) Lysosome (NES ∼1.53) Lysosome (NES ∼1.53) Autophagy (NES ∼1.46–1.53) Autophagy (NES ∼1.46–1.53) Mitophagy (NES ∼1.48) Mitophagy (NES ∼1.48) Chemical carcinogenesis—ROS (NES ∼1.73) Chemical carcinogenesis—ROS (NES ∼1.73) Oxidative phosphorylation (NES ∼1.70) Oxidative phosphorylation (NES ∼1.70) Peroxisome (NES ∼1.57) Peroxisome (NES ∼1.57) Glutathione metabolism (NES ∼1.59) Glutathione metabolism (NES ∼1.59) Ferroptosis (NES ∼1.53) Ferroptosis (NES ∼1.53) Glutathione metabolism (NES ∼1.59) Glutathione metabolism (NES ∼1.59) Cysteine and methionine metabolism (NES ∼1.51) Cysteine and methionine metabolism (NES ∼1.51) Beta‐alanine metabolism (NES ∼1.37) Beta‐alanine metabolism (NES ∼1.37) Peroxisome (NES ∼1.57) Peroxisome (NES ∼1.57) Metabolic pathways (NES ∼1.52) Metabolic pathways (NES ∼1.52) Carbon metabolism (NES ∼1.50–1.57) Carbon metabolism (NES ∼1.50–1.57) TCA cycle (NES ∼1.50–1.58) TCA cycle (NES ∼1.50–1.58) Glycolysis/gluconeogenesis (NES ∼1.50) Glycolysis/gluconeogenesis (NES ∼1.50) Pyruvate metabolism (NES ∼1.49) Pyruvate metabolism (NES ∼1.49) Mucin‐type O‐glycan biosynthesis (NES ∼1.84) Mucin‐type O‐glycan biosynthesis (NES ∼1.84) N‐Glycan biosynthesis (NES ∼1.30) N‐Glycan biosynthesis (NES ∼1.30) Butanoate metabolism (NES ∼1.22) Butanoate metabolism (NES ∼1.22) Propanoate metabolism (NES ∼1.22) Propanoate metabolism (NES ∼1.22) PPAR signaling (NES ∼1.30–1.33) PPAR signaling (NES ∼1.30–1.33) Neutrophil extracellular trap formation Neutrophil extracellular trap formation (NES ∼1.35) Necroptosis (NES ∼1.35) (NES ∼1.35) Necroptosis (NES ∼1.35) PPAR signaling (NES ∼1.30–1.33) PPAR signaling (NES ∼1.30–1.33) Infection‐related pathways (NES ∼1.30–1.60) Infection‐related pathways (NES ∼1.30–1.60) Chemical carcinogenesis—ROS (NES ∼1.73) Chemical carcinogenesis—ROS (NES ∼1.73) Colorectal cancer (NES ∼1.48) Colorectal cancer (NES ∼1.48) Viral carcinogenesis (NES ∼1.50) Viral carcinogenesis (NES ∼1.50) ER protein processing (NES ∼1.47) ER protein processing (NES ∼1.47) Platinum drug resistance (NES ∼1.46) Platinum drug resistance (NES ∼1.46) mTOR signaling (NES ∼1.64) mTOR signaling (NES ∼1.64) AMPK signaling (NES ∼1.60) AMPK signaling (NES ∼1.60) FoxO signaling (NES ∼1.52–1.54) FoxO signaling (NES ∼1.52–1.54) Ribosome (NES ∼1.68) Ribosome (NES ∼1.68) Colorectal cancer pathway (NES ∼1.48) Colorectal cancer pathway (NES ∼1.48) PI3K/MAPK modules (within multiple pathways) PI3K/MAPK modules (within multiple pathways) Discussion This dietary intervention study is among the first to investigate systemic and local effects of dietary E171 intake in healthy humans. We established a method to monitor study compliance and colonic E171 exposure using hrICP‐MS, revealing statistically significant differences in stool Ti content between the control and E171 periods (6.3 mg/kg vs. 377 mg/kg, respectively). These results align with Putra et al. (2024), who reported Ti levels in individuals with low (3 mg/kg) and high (746 mg/kg) E171 exposure [20]. We assessed oxidative stress using TEAC, PCC, and ESR spectroscopy. While no changes in the antioxidative capacity of plasma were observed in TEAC and PCC assays, superoxide radical levels in whole blood increased significantly after E171 exposure. This suggests that E171 can induce systemic ROS following, or even despite, gastrointestinal assimilation and interactions with the food matrix, consistent with prior in vitro findings [8, 9]. However, whether ROS formation originates directly from E171's surface reactivity or secondary metabolic effects remains unclear. Potential explanations include increased mitochondrial activity and oxidative phosphorylation, as observed in E171‐exposed colon organoids. Bischoff et al. (2024) reported the innate potential of E171 to induce ROS in an aqueous environment and observed elevated superoxide levels in a human colon organoid model in vitro [25]. Notably, following oral ingestion of a 100 mg E171 capsule, blood Ti levels increase 2–6 h post‐ingestion, implying systemic uptake of E171 into the circulation [31]. Furthermore, deposition of E171 in human liver and spleen has been previously established [32]. Overall, this suggests systemic uptake of E171, which eventually leads to increased oxidative stress. Despite increased oxidative stress in whole blood, systemic inflammation markers in plasma remained unchanged. We analyzed protein levels, hs‐CRP, and cytokines (IL‐1β, IL‐6, IL‐8, IL‐10, IL‐17, IL‐18, TNF‐α, IFN‐α, and IFN‐γ) and found no alterations following E171 exposure. Additionally, unlike Putra et al. (2024), we detected no changes in fecal calprotectin concentration, suggesting that E171 did not elicit measurable intestinal inflammation in our healthy cohort exposed to 2 mg/kg bw/day of E171 [21]. We conducted bulk total RNA sequencing to investigate the molecular impact of E171 on the colonic mucosa. Gene set enrichment analysis (GSEA) identified 73 activated pathways, notably related to metabolism (Oxidative phosphorylation, Glycolysis, Citrate cycle), mirroring the Warburg effect—a hallmark of cancer characterized by a shift toward aerobic glycolysis [18, 19, 20, 21]. This metabolic reprogramming, observed in CRC patients, is often accompanied by oncogene mutations such as KRAS activation and p53 suppression, as similarly identified in our dataset. These findings align with human‐induced pluripotent stem cell (iPSC)‐derived colon organoid studies, which reported similar metabolic shifts following E171 exposure [19]. Furthermore, pathways linked to chemical carcinogenesis via ROS, glutathione metabolism, viral carcinogenesis, and CRC were activated. Increased systemic superoxide levels correlated with ROS‐driven carcinogenic pathways and heightened glutathione metabolism, indicating oxidative stress‐induced cellular responses. Notably, the CRC pathway showed suppression of AXIN, activation of APC, and increased β‐catenin production, which are key features in CRC [13, 15, 18, 19, 33]. To contextualize the findings of our transcriptome study, we mapped these results to the postulated MoA of E171 [2, 30]. Table 4 shows how the activated pathways correlate with previously reported molecular initiating events and key events (Figure S5). Our study confirms transcriptional changes related to all postulated events, starting with cellular uptake/particles internalization (KE1: Endocytosis, Phagosome, Lysosome, Autophagy, Mitophagy), ROS generation (KE2: Chemical carcinogenesis—ROS, Oxidative Phosphorylation, Peroxisome, Glutathione metabolism, Ferroptosis), oxidative stress/redox imbalance (KE3: Glutathione metabolism, Cysteine and methionine metabolism, Beta‐alanine metabolism, Peroxisome), epithelial injury/metabolic rewiring (KE4: Metabolic pathway, Carbon metabolism, TCA cycle, Glycolisis/gluconeogenesis, Pyruvate metabolism), barrier/microbiome changes (KE5: Mucin‐O‐glycan biosynthesis, N‐glycan biosynthesis, Butanoate metabolism, Propanoate metabolism, PPAR signaling), local inflammation (KE6: Neutrophil extracellular trap formation, Necroptosis, PPAR signaling, Infection‐related pathways), DNA damage/genotoxic stress (KE7: Chemical carcinogenesis—ROS, Colorectal cancer, Viral carcinogenesis, ER protein processing, Platinum drug resistance), proliferation/survival signaling (KE8: mTOR signaling, AMPK, signaling, FoxO signaling, Ribosome), and finally CRC pathway itself (KE9: Colorectal cancer pathway, PI3K/MAPK signaling). Signal transduction pathways related to cell growth and survival were also affected, including mTOR, FoxO, PPAR, and AMPK signaling. Overactivation of mTOR is known to drive tumorigenesis via uncontrolled proliferation, while FoxO upregulation suggests a compensatory response to oxidative stress [25, 26]. AMPK activation, a metabolic stress sensor, likely reflects the shift toward carbon metabolism, while altered PPAR signaling, previously linked to inflammatory bowel disease (IBD) and CRC, may contribute to tumorigenesis through disrupted lipid metabolism and inflammation [27, 28, 29]. We also identified changes in short‐chain fatty acid (SCFA) metabolism, with an activation of pathways that metabolize butyrate and propionate. Notably, changes in butyrate have been identified in the in vitro TIM‐2 model, revealing that E171 exposure shifts microbiome activity toward butyrate production. While butyrate is generally beneficial, excessive levels can have paradoxical effects, promoting inflammation and reducing the survival of CRC patients [30]. Butyrate is a ligand for GPCRs and nuclear receptors, triggering pro‐inflammatory signaling via the NLRP3 inflammasome and cytokine release [31, 32]. Our transcriptomic findings align with those of previous studies in various models. Caco‐2 microarray experiments highlighted GPCR modulation and tumor growth promotion [33], aligning with transcriptomic changes in E171‐exposed transgenic APC‐mutant mice, which also showed increased oxidative stress and CRC‐associated gene alterations [18]. Balb/c and AOM/DSS mice exhibited hyperplasia, GPCR activation, oxidative stress, and immune suppression, further supporting the findings of our human study [15, 18]. Furthermore, changes in circadian rhythm have been previously reported in a transgenic mouse model for CRC [18]. Notably, while Kulthong et al. (2021) reported metabolic downregulation in a dynamic flow chamber Caco‐2 model, the opposite effect was observed in human iPSC‐derived organoids—notably the activation of metabolic and CRC‐related pathways [34]. When comparing models, iPSC‐derived organoids appeared to be the most sensitive to E171 exposure, followed by disease‐prone animal models (AOM/DSS and APC‐mutant mice). At the same time, our healthy human data fell between these and standard Balb/c mice, with minimal overlap with Caco‐2 cells (Figure S6). A comprehensive transcriptomic analysis, employing standardized transcriptomic data processing, could provide further insights into E171's MoA across various models. While this study comprehensively evaluates dietary E171 exposure in healthy humans, certain limitations should be acknowledged. Local cellular inflammatory markers in colon tissue were not assessed. We suggest addressing this more specifically in future studies for example by evaluating parameters such as glutathione‐S‐transferase in colon tissue and α1‐antitrypsin (gut permeability) to better capture and understand localized effects. The study duration (2 weeks) and sample size were also constrained, which limited the long‐term hazard assessment. In future research, one may address gastrointestinal damage by focusing on more vulnerable populations, such as individuals with pre‐existing gastrointestinal inflammatory or precancerous conditions, or by screening for genetic predispositions, to refine hazard identification and characterization and identify susceptible subpopulations. Conclusions In conclusion, this dietary intervention study in healthy humans identified pivotal, important insights into the effects of dietary E171 exposure on the gastrointestinal tract at a human‐relevant dose. Using advanced techniques such as hrICP‐MS, spICP‐MS, and TEM, we monitored study compliance and characterized E171 particles before and after human gastrointestinal digestion. Our experiments have highlighted E171's potential to induce oxidative stress, as indicated by elevated systemic superoxide levels. A metabolic shift toward increased carbon metabolism and oxidative phosphorylation in the colon, consistent with the oncogenic Warburg effect, was suggested by transcriptome analysis. Additionally, we identified activated pathways linked to cancer, inflammation, and metabolic dysregulation, including mTOR, FoxO, PPAR, and AMPK signaling, which are known contributors to colorectal carcinogenesis. Collectively, this study highlights the potential health risks associated with dietary exposure to E171, identifies potential mechanistic insights into its MoA, and enhances our understanding of its impact on gastrointestinal health. The current findings substantiate the European Commission's regulatory decision to ban E171 as a food additive, warranting a re‐evaluation of its use in global markets and pharmaceuticals. Author Contributions Nicolaj S. Bischoff: writing – original draft preparation. Nicolaj S. Bischoff, Anna K. Undas, Greet van Bemmel, Dick T.H.M. Sijm, Simone G. Van Breda, Jacco J. Briedé, Ad A. van Bodegraven, and Theo M. de Kok: writing – review and editing. Theo M. de Kok, Dick T.H.M. Sijm, Ad A. van Bodegraven, Héloïse Proquin, and Henk van Loveren: project administration. Theo M. de Kok and Dick T.H.M. Sijm: funding acquisition. Nicolaj S. Bischoff, Dick T.H.M. Sijm, Simone G. Van Breda, Jacco J. Briedé, and Theo M. de Kok: experimental design. Nicolaj S. Bischoff, Anna K. Undas, Greet van Bemmel, Marcel van Herwijnen, Iris Elferink, Marcha Verheijen, Lana Drenth‐Brink, and Marcella Peelen‐Buijs: experimental execution. All authors have read and agreed to the published version of the manuscript. Funding Maastricht University provided internal funding, NVWA ARVODI‐2018 Overeenkomst Onderzoeksopdracht E171. Ethics Statement This human dietary intervention study was performed according to the Helsinki Declaration of Ethical Principles for Medical Research Involving Human Subjects, following approval by the Medical Ethical Review Committee of the Maastricht University Medical Centre + (MUMC+, registration number: NL52433.068.15/METC163010) and the Office of Scientific Research, RVE Zuyderland Academy, Zuyderland (registration number: Z2022015). Conflicts of Interest A.A.v.B. has served as a speaker, adviser, and/or principal investigator for AbbVie, Alfasigma, Celltrion, E‐wise, Ferring, Galapagos, Janssen, Pfizer, Roche, Takeda, and TEVA, and has received research grants from TEVA and Zuyderland MC, all outside the submitted work. All other authors declare no conflicts of interest. Supporting information Supporting File 1: mnfr70583‐sup‐0001‐SuppMat.docx. Supporting File 2: mnfr70583‐sup‐0002‐SuppMat.docx.