As the incidence of colorectal cancer (CRC) has recently increased, it has become one of the most common cancers globally. Specifically, the number of reported cases doubled between 1990 and 2019(1). According to GLOBOCAN, an estimated 1·9 million cases of CRC were diagnosed in 2020, and CRC was recognised as the third most common cancer(Reference Sung, Ferlay and Siegel2). The incidence of CRC has been documented to be greater in developed countries than in middle- and low-income countries(Reference Xi and Xu3). In the Asia-Pacific region, there is a growing public health challenge due to an increasing burden, particularly in Eastern Asia, where high incidence rates have been observed(Reference Pourhoseingholi4). South Korea is not an exception because it is one of the most commonly diagnosed cancers(Reference Pourhoseingholi4,Reference Kang, Won and Lee5) .
The increasing trend in CRC incidence is strongly affected by several risk factors, including sex, race, genetics and environmental factors related to Western lifestyles and diet(Reference Xi and Xu3). The Western diet typically involves low fruit and vegetable consumption and overuse of refined sugar and salt, which has been indicated to have a detrimental effect on the immune system(Reference Myles6). Notably, dietary Se, primarily through its incorporation into selenoproteins, plays a certain role in immunity and inflammation by inhibiting activation of NF-κB and C-reactive protein production(Reference Huang, Rose and Hoffmann7,Reference Duntas8) . Additionally, Se is recognised as a micronutrient with cancer prevention properties(Reference Fagbohun, Gillies and Murphy9).
To date, the importance of Se in human health has received much attention due to its antioxidant properties when it is consumed at an appropriate level(Reference Lener, Gupta and Scott10,Reference Combs11) . However, due to contradictory results, the existing evidence on CRC prevention efficacy is limited. Previous studies have provided insights into the role of a high Se concentration in relation to a substantially reduced CRC risk(Reference Lener, Gupta and Scott10,Reference Luo, Fang and Zhang12) . Similarly, the beneficial effect on colorectal carcinogenesis was reinforced by the findings of another study(Reference Augustyniak and Galas13). In contrast, a meta-analysis of sixty-nine studies revealed that cancer prevention agents were effective for breast, gastric, lung, oesophageal and prostate cancers but not for CRC(Reference Cai, Wang and Yu14). Additionally, there is no evidence to support a role for high Se concentrations in cancer prevention from the other meta-analysis. However, further studies should take into consideration individuals’ genetic backgrounds to explore Se in relation to cancer risk(Reference Vinceti, Filippini and Del Giovane15).
Furthermore, epidemiological evidence is sufficient to support the causal link between inflammation and cancer progression(Reference Zhao, Wu and Yan16). IL10 is known to be an anti-inflammatory cytokine that is produced by type 1 regulatory T cells and other cells(Reference Abtahi, Davani and Mojtahedi17). It may suppress the Th1-mediated immune response and cell-mediated immune response and reciprocally enhance antibody-mediated responses(Reference Kawamura, Bahar and Natsume18). IL10 plays a critical role in various physiological processes that maintain homoeostasis in the gastrointestinal tract. It helps to regulate intestinal inflammation and various pathophysiological processes, including inflammatory bowel diseases, which are related to an elevated susceptibility to CRC(Reference Banday, Sameer and Chowdri19). The association between IL10 and CRC susceptibility was well recognised in a previous study. In detail, IL10 levels were significantly greater in CRC patients than in healthy individuals. Importantly, the highest IL10 levels were identified in patients with stage IV disease, which was significantly greater than that in patients with stage I, II or III disease. Thus, an increased IL10 level was strongly associated with the progression of CRC. In contrast, no associations between IL17 or Interferons (IFN)α levels and CRC were found(Reference Stanilov, Miteva and Deliysky20). Additionally, the expression of IL10 was highlighted as an indicator of the prognosis of CRC patients in another study(Reference Li, Wang and Ma21). Notably, the IL10 gene is located at chromosomal region 1q31–1q32, and polymorphisms in this gene, especially polymorphisms in the promoter region, have been implicated in cancer because they can affect IL10 gene transcription and translation(Reference Banday, Sameer and Chowdri19,Reference Moghimi, Ahrar and Karimi-Zarchi22,Reference Tsilidis, Helzlsouer and Smith23) . Rs1800871 is an SNP in the promoter of the IL10 gene in the Korean population. Notably, the interaction effect of genetics and diet provides a unique environment for cancer development and suppression. Individuals with high-risk genetic variants and particular dietary habits may exhibit a greater cancer risk than individuals without high-risk genetic variants(Reference Kim, Cho and Choi24). Dietary Se was indicated to have an effect on cytokine secretion and immune function(Reference Zhang, Zhang and Xia25). A strong association between Se and IL10 has been documented. Se supplementation of immune cells led to increased IL10 expression in B cells and reduced induction of pro-inflammatory cytokines in B and CD4 + T cells. IL10 production in response to Se was confirmed to be linked to the activation of the ERK and Akt pathways(Reference Zhou, Yuan and Xiao26). Additionally, the oxidative stress-induced release of cytokines, including IL10, can be prevented by Se(Reference Rafferty, Walker and Hunter27). Importantly, genetic background should be considered when assessing Se intake or supplementation in relation to cancer(Reference Vinceti, Filippini and Del Giovane15). Thus, we formulated a hypothesis regarding the interactive effect of Se and the common variant IL10 rs1800871, found in the IL10 gene within the Korean population, on colorectal carcinogenesis.
To our knowledge, there are a limited number of epidemiological studies on the association of dietary Se intake with CRC development. Furthermore, the ambiguous findings of previous studies raise questions about the potential protective role of Se in CRC. Additionally, the effect of the IL10 rs1800871 polymorphism on the association between the Se concentration and CRC has not been studied. Thus, our study aimed to explore whether Se has a potential preventative effect on CRC and whether the IL10 rs1800871 polymorphism has an impact on this association.
Materials and methods
Study design and participants
We designed a case–control study by enrolling participants from the National Cancer Center (NCC) in Korea. We defined cases as individuals newly diagnosed with CRC at the Center for CRC of the NCC from August 2010 to September 2020. The participants who visited the Center for Cancer Prevention and Detection for Health Screening Program from October 2007 to December 2022 were considered controls. We used sex and age (±5 years) to match one case with two controls after excluding 290 cases and 5409 controls with incomplete questionnaires or semi-quantitative FFQ results and thirteen cases and 196 controls with implausible energy intake (< 400 and ≥ 5000 kcal/d). Additionally, we excluded fifty-seven cases with non-CRC and 1305 controls with previous diagnosis of any cancer. Finally, we investigated dietary Se intake in relation to CRC risk in 1420 CRC cases and 2840 controls. Furthermore, a total of 437 cases and 1063 controls with missing information on IL10 rs1800871 were excluded from the genetic analyses (Fig. 1). This study was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving research study participants were approved by the Institutional Review Board of the National Cancer Center, Korea (IRB numbers: NCCNCS-10-350 and NCC 2015-0202). Written informed consent was obtained from all subjects/patients.
Fig. 1. Flow chart of the study participants. After excluding participants with incomplete semi-quantitative FFQ (SQFFQ) or a self-administered questionnaire, a total of 1420 CRC cases and 2840 controls were included in our analysis to investigate dietary selenium intake in relation to CRC risk. We additionally excluded 437 cases and 1063 controls with missing information on IL10 rs1800871 for genetic analysis. CRC, colorectal cancer.
Dietary assessment
The food consumption frequency of each participant and their portion size during the previous year were collected to assess dietary intake based on the 106-item semi-quantitative FFQ. A previous report provided information on the reproducibility and validity of the semi-quantitative FFQ(Reference Ahn, Kwon and Shim28). The calculation of total energy and Se intake was performed with CAN-Pro 5.0 (Computer Aided Nutritional Analysis Program, the Korean Nutrition Society). Daily Se intake was calculated as the sum of the Se obtained from all foods consumed throughout the day (μg/d). Furthermore, information on demographics and lifestyle was provided using a self-administered questionnaire completed by participants.
Genotype measurement
We extracted genomic DNA from the blood samples of participants with MagAttract DNA Blood M48 Kit (Qiagen) and BioRobot M48 automatic extraction equipment (Qiagen). The Illumina MEGA-Expanded Array (Illumina, Inc.), which included 123K variants, was used for genotyping. A detailed description of this method has been provided in a previous publication(Reference Lu, Kweon and Cai29). The performance of genotype imputation was conducted using the Asian population (n 504) in the 1000 Genomes haplotypes phase III integrated variant set release GRch37/hg19 (https://www.1000genomes.org/) as a reference panel. Genetic markers with deviation from Hardy–Weinberg equilibrium P values < 1 × 10−6, a minor allele frequency < 0·05 and a low call rate (< 98 %) were discarded. We used SHAPEIT (v2.r837) and IMPUTE2 (2.3.2) for phasing and SNP imputation, respectively. The quality control criteria were applied after filtering for an INFO score over 0·6. Finally, IL10 rs1800871 was selected as a candidate SNP for our analysis.
Statistical analyses
We analysed the differences in demographic and lifestyle factors between the cases and controls with t tests and χ 2 tests. We adjusted the Se concentration for total energy intake using a residual method(Reference Willett and Stampfer30). The Se intake quartiles were determined based on the distribution in the control group. We examined dose–response relationships based on the median value of each Se category. We developed different models to explore Se intake in relation to CRC risk by calculating OR and 95 % CI using unconditional logistic regression. We kept missing values for each variable as a category in the analysis. Additionally, multinomial logistic regression models were utilised to examine whether dietary Se intake is associated with each anatomical subsite (proximal colon, distal colon or rectal cancer) of CRC patients. We used the Wald test to assess heterogeneity between sex groups. A dominant model was used for genetic analysis. The interaction effect between Se and SNP was analysed using the likelihood ratio test between the models with and without the interaction term (selenium*SNPs). SAS software (version 9.4, SAS Institute) was used for all the statistical analyses, and a two-sided P value less than 0·05 was considered significant.
Results
Sociodemographic characteristics of the study population
A greater proportion of CRC patients than controls had a first-degree family history of CRC (10·1 % v. 5·4 %, P < 0·001). Similarly, the cases had a higher likelihood of being former alcohol consumers and less educated compared with controls (14·4 % v. 9·6 %, P < 0·001 and 17·8 % v. 6·1 %, P < 0·001, respectively). Furthermore, in comparison with healthy individuals, CRC patients tended to have lower levels of regular exercise, lower-status occupations and lower incomes (35·5 % v. 56·0 %, P < 0·001; 23·4 % v. 27·9 %, P < 0·001; and 23·1 % v. 36·1 %, P < 0·001) (Table 1).
Table 1. General characteristics of the subjects
CRC, colorectal cancer.
Se was adjusted for total energy intake using the residuals method. Boldface is significant (P<0.05).
* t test and χ2 test were used for continuous and categorical variables, respectively.
† Mean ± sd.
Dietary selenium intake and colorectal cancer risk
The cases and controls exhibited significant differences in total energy and Se intake. Specifically, total energy intake was greater in the cases than in the healthy individuals (P < 0·001). Conversely, we observed a lower Se consumption in these patients (44·1 ± 16·9 μg/d v. 50·9 ± 19·1 μg/d) (P < 0·001).
A higher dietary Se intake seems to lead to a lower risk of colorectal carcinogenesis. Notably, significant associations were detected with both the unadjusted unconditional logistic regression model and the adjusted model; the OR (95 % CI) were 0·35 (0·29, 0·42) and 0·44 (0·35, 0·55), respectively, P for trend < 0·001. Notably, our findings consistently demonstrated a significant reduction in CRC risk for both males and females with a high Se intake (OR = 0·45 (0·33, 0·60) and OR = 0·46 (0·32, 0·68), respectively) (Table 2). Furthermore, the difference in the risk of CRC associated with Se intake was different in proximal colon cancer risks across each sex (P heterogeneity < 0·001). In detail, Se tended to reduce distal colon cancer and rectal cancer risk but not proximal colon cancer risk in male individuals, whereas it contributed to decreased proximal colon cancer and rectal cancer risk among females (online Supplementary Table 1).
Table 2. OR and 95 % CI of CRC according to the quartiles of dietary selenium intake
CRC, colorectal cancer.
Model 1: unadjusted unconditional logistic regression model; Model 2: adjusted for age, BMI, first-degree family history of CRC, regular exercise, smoking status, alcohol consumption, occupation, education and income. In the total subjects, model 2 was additionally adjusted for sex. Boldface is significant (P<0.05).
IL10 rs1800871 genetic polymorphisms and colorectal cancer risk
We investigated the IL10 rs1800871 genetic polymorphism in relation to CRC risk using the dominant model with two groups of genotypes (AA and GA/GG) after excluding participants with missing information on the IL10 rs1800871 gene. A total of 983 CRC cases and 1774 controls were involved in the genetic analyses. According to the unadjusted model, variant allele carriers were more susceptible to developing CRC (OR = 1·17 (1·00, 1·37)). However, this greater risk disappeared after potential confounders were added to the model (OR = 1·17 (0·97, 1·40)). Furthermore, a marginal significance was observed for female participants carrying a minor allele compared with those with a homozygous wild type (OR = 1·34 (0·99, 1·80)) (Table 3).
Table 3. Associations of IL10 rs1800871 genetic polymorphisms with CRC risk in the dominant model
CRC, colorectal cancer.
Model 1: unadjusted unconditional logistic regression model; Model 2: adjusted for age, BMI, first-degree family history of CRC, regular exercise, smoking status, alcohol consumption, occupation, education and income. In the total subjects, model 2 was additionally adjusted for sex.
Table 4. Interaction between IL10 rs1800871 genetic polymorphisms and selenium with CRC risk in the dominant model
CRC, colorectal cancer.
The quartiles of Se were re-calculated after excluding participants with missing information on IL10 rs1800871.
Model 1: unadjusted unconditional logistic regression model; Model 2: adjusted for age, BMI, first-degree family history of CRC, regular exercise, smoking status, alcohol consumption, occupation, education and income. In the total subjects, model 2 was additionally adjusted for sex. Boldface is significant (P<0.05).
Interaction between the IL10 rs1800871 genetic polymorphism and selenium and colorectal cancer risk
IL10 rs1800871 modified the association between Se and colorectal carcinogenesis, suggesting that an individual’s CRC risk may differ based on genetic background. An association with a reduced CRC risk tended to be limited to IL10 rs1800871 G-allele carriers with a high intake of Se; the OR (95 % CI) in the unadjusted unconditional logistic regression and adjusted models were 0·44 (0·32, 0·61) and 0·61 (0·42, 0·87), respectively. In contrast, participants with a homozygous wild-type allele exhibited a non-significant association between dietary Se intake and CRC development. Notably, Se and IL10 rs1800871 were suggested to have a significant interaction on CRC susceptibility (P for interaction = 0·043).
Discussion
In this study, involving 4260 participants, we emphasised the significant association between an appropriate Se intake and CRC risk. Specifically, a high intake of Se was associated with a reduced risk of CRC. Furthermore, we observed that the IL10 rs1800871 polymorphism had an impact on this inverse association, with the effect being allele-specific and present only among variant allele carriers.
Se has received significant attention for its potential anticarcinogenic properties, particularly in populations with low intake(Reference Vinceti, Filippini and Del Giovane15). However, the role of Se in CRC prevention is still debated, and a clear conclusion has not been reached due to previous contradictory results. For example, the Se content in the diet is suggested to be a trace element for CRC prevention because a high-dietary Se concentration has been linked to a decrease in CRC risk(Reference Augustyniak and Galas13). Similarly, the risk of colorectal carcinoma decreased significantly with higher levels of Se. The beneficial effect was more pronounced for females than for males. Thus, increasing Se intake may be considered a strategy for reducing the risk of colorectal carcinoma, especially in patients who have suboptimal Se intake(Reference Hughes, Fedirko and Jenab31). Additionally, the anticancer effect of Se intake was revealed in another study(Reference Luo, Fang and Zhang12). However, the aforementioned association was not supported by the conclusions of a previous study(Reference Cai, Wang and Yu14).
Our study adds to the growing body of evidence supporting the hypothesis regarding the significant role of appropriate intake of Se in preventing the progression and development of CRC. However, extremely high Se intake can cause adverse effects(Reference Combs11). These findings are reinforced by potential biological mechanisms that underlie the anticarcinogenic properties of Se. First, Se has been reported to have antioxidant effects, which are believed to account for its preventative effects. The antioxidant effects of Se are primarily attributed to its role as a constituent of redox-active selenoproteins, specifically those containing selenocysteine. These proteins contribute to reducing oxygen species and intra- and intermolecular disulfides or mixed disulfides/selenides(Reference Björnstedt and Fernandes32). Second, several selenoproteins have been shown to affect various biological processes(Reference Björnstedt and Fernandes32). For example, glutathione peroxidases, including GPx2, contribute to mucosal integrity through antiapoptotic effects in colon crypts and reduce peroxide levels in the gut; thioredoxin reductases are involved in controlling transcription factor expression, cell proliferation and apoptosis(Reference Augustyniak and Galas13). Third, several metals are known to increase cancer risk. Notably, Se can chemically interact with metals. Cd is a key element in the development of breast and prostate cancers, for example, Se has a protective role against Cd-induced peroxidative damage(Reference Björnstedt and Fernandes32). Fourth, the impact of Se on the p53 protein has been documented, including its ability to inhibit proliferation, stimulate DNA repair and promote apoptosis(Reference Rayman33). Fifth, NF-κB is associated with an enhanced inflammatory response, and Se may inhibit its activation by modulating the expression of selenoprotein genes(Reference Duntas8).
Chronic inflammation is well accepted to be involved in the aetiology of CRC, and inflammation occurs at the early stage of CRC progression and facilitates the progression of preneoplastic lesions into metastatic tumours. Anti-inflammatory cytokines and pro-inflammatory agents are important mediators and regulators of the immune response and contribute significantly to tumorigenesis by regulating tumour-related inflammation(Reference Banday, Sameer and Chowdri19). IL10, an anti-inflammatory cytokine, has been discussed as a regulator of carcinogenesis and tumour growth(Reference Mirjalili, Moghimi and Aghili34). The SNP located in the promoter region of the IL10 gene are in relation to changes in transcription and expression. Thus, these promoter polymorphisms are thought to be related to cancer risk(Reference Tsilidis, Helzlsouer and Smith23,Reference Mirjalili, Moghimi and Aghili34) ; however, the conclusion remains controversial. One of the IL10 promoter polymorphisms is rs1800871, which has attracted increased amounts of attention from epidemiological studies because of its unclear association with cancer susceptibility. A previous study conducted in Croatia representing a European population concluded that this polymorphism may be associated with CRC risk(Reference Cacev, Radosević and Krizanac35). In contrast, another study based on Americans in Bethesda, Maryland, indicated a non-significant association where those results were consistent with the current study(Reference Gunter, Canzian and Landi36). However, additional studies with larger sample sizes are necessary to provide reliable conclusions overall and by ethnicity(Reference Mirjalili, Moghimi and Aghili34).
A previous study suggested that focusing on genetic background is needed when assessing Se intake or supplementation in relation to cancer(Reference Vinceti, Filippini and Del Giovane15). Notably, the association between Se and CRC appears to be allele-specific in our study, further emphasising the importance of considering genetic factors in such assessments. Specifically, an inverse association between Se and colorectal carcinogenesis seemed to be limited to individuals who carry the minor G allele of IL10 rs1800871, for which the interaction effect of dietary Se-IL10 was significant. Although the exact mechanisms underlying the modification effect of IL10 rs1800871 on this relationship are not fully understood, we propose probable explanations. A previous study highlighted the relationship between Se and immune function and suggested that low and high Se levels can have an impact on cytokine secretion and impair immune function in mice(Reference Zhang, Zhang and Xia25). Furthermore, Se deficiency can lead to an increase in the expression of NF-κB and hypoxia inducible factor (HIF)-1α. Additionally, the regulation of inflammatory cytokines is affected, and there is a decrease in the expression of IL10 (Reference Li, Zhao and Zhang37). Another possible explanation may be that Se inhibits the release of cytokines such as IL10, which may suppress cell-mediated immunity(Reference Rafferty, Walker and Hunter27). Taken together, the interaction between Se and IL10 may account for the different effects of Se intake on CRC development, which depend on the genetic background of the participants.
Additionally, SNP are known to play certain roles in regulating protein expression, which contributes to differences in disease susceptibility and severity among individuals. Rs1800896, rs1800871 and rs1800872 are three common SNP located in the IL10 gene that are associated with increased production of IL10 and impact the expression and functions of proteins(Reference Zhang, Xing and Chen38). Notably, complete linkage disequilibrium was found between rs1800871 and rs1800872(Reference Zhang, Xing and Chen38). A previous study indicated that participants with the AC or AC/CC genotype of rs1800872 exhibited a reduction in CRC risk compared with those with the AA genotype(Reference Yu, Zheng and Zhang39). Thus, further studies are needed to confirm our findings, with a focus on the linkage disequilibrium between rs1800871 and rs1800872.
Furthermore, CRC risk has been shown to vary among males and females. Compared with males, females have a more aggressive form of neoplasia because the risk of right-sided (proximal) colon cancer seems to be greater. Importantly, dietary factors have been shown to be associated with tumour location. Thus, greater emphasis should be placed on sex-specific estimates of dietary risk factors to establish guidelines on dietary intake for cancer prevention(Reference Kim, Paik and Yoon40). Notably, there were discrepancies in Se levels between females and males. Sex-specific nutritional and health behaviours and variations in Se metabolism and Se distribution across body compartments may explain these discrepancies(Reference Vinceti, Filippini and Del Giovane15). Thus, we investigated the differences in the association between Se intake and CRC between males and females. We found significant associations within the total population as well as for both males and females. Importantly, there is variation in dietary patterns between sex groups, highlighting the significance of adjusting for sex as a crucial confounder(Reference Park, Lee and Oh41). Thus, our results are more accurate and reliable because we adjusted for sex. Additionally, a significant interaction effect of the IL10 rs1800871 genetic polymorphism and Se intake on CRC risk was detected in the total population but not in males or females. The limited sample size may be a possible explanation for this observation.
The colon and rectum exhibit different receptor patterns because the colon arises from the midgut and the rectum arises from the hindgut. Additionally, colon and rectal cancers have different functions and are exposed to faeces for different durations(Reference Wei, Giovannucci and Wu42). Furthermore, different genes are involved in oncogenesis in the colon and rectum. Notably, the proximal and distal colon were emphasised to have differences in clinical and molecular aspects. Familial polyposis syndrome arises first in the rectum and distal colon, whereas hereditary non-polyposis coli arises in the proximal colon(Reference Wei, Giovannucci and Wu42). Thus, susceptibility to risk factors may vary between the distal and proximal colon. Furthermore, the aetiology of the anatomical site has been documented to have sex-specific disparities(Reference Kim, Paik and Yoon40). These findings are in line with our findings, which highlight the different effects of Se on CRC risk between the distal and proximal colon in males and females.
Our study is the first to focus on the negative association between Se and CRC involving an interaction with an inflammatory gene. Additionally, a validated semi-quantitative FFQ, which was designed for the Korean population, was used in our study. Thus, we collected precise and representative dietary intake data from our study participants. Information on potential confounders was collected and adjusted for in our study. However, our study has several limitations. First, we had some degree of recall bias and selection bias due to the case–control design. Second, several important variables related to the Se content in foods were not considered. Third, although the other probable genes may affect the association between Se and CRC, these genes were not considered in our study. Additionally, IL10 rs1800871 has been demonstrated to have complete linkage disequilibrium with rs1800872(Reference Zhang, Xing and Chen38). Thus, further studies are needed to focus on other genes beyond IL10 and linkage disequilibrium between IL10 rs1800871 and rs1800872 to reach an effective conclusion. Fourth, the small number of genotypes may have affected the statistical power of the genetic associations. Fifth, information on important variables, such as the serum Se concentration and supplemental use, was not available for consideration as possible confounders in our analysis.
In conclusion, we provided evidence to support the notion that appropriate intake of Se may reduce cancer risk. However, the potential benefit of Se against colorectal carcinogenesis depends on the individual’s genetic background; in detail, high Se intake was emphasised to have a greater effect on variant allele carriers of IL10 rs1800871. Our findings suggest that individual genetic characteristics should be considered in nutritional strategies for CRC prevention. However, further studies with crossover analyses are needed to confirm the established interaction between dietary Se and the IL10 rs1800871.
Acknowledgements
This work was supported by grants from the National Cancer Center, Korea (2310470), and National Research Foundation of Korea (2021R1A2C2008439).
Formal analysis, T. T. T., M. G. and J. L.; Preparation of original draft, T. T. T.; Writing – review and editing, M. G. and J. K.; Data curation, J. L., J. H. O., H. J. C., D. K. S., A. S. and J. K.; Investigation, J. L., J. H. O., H. J. C., D. K. S. and A. S.; Methodology, J. H. O., H. J. C., D. K. S., A. S. and J. K.; Funding acquisition, J. K.; Project administration, J. K.; Supervision, J. K. All authors have critically reviewed and approved the final version of the manuscript submitted for publication.
The authors declare that they have no conflicts of interest.
Supplementary material
For supplementary material/s referred to in this article, please visit https://doi.org/10.1017/S0007114524000345
