Colorectal cancer (CRC) is ranked as one of the most common cancers globally. In 2020, there were more than 1·9 million new CRC cases and 0·9 million deaths worldwide(Reference Siegel, Miller and Fuchs1). In China, there were 0·555 million new CRC cases which ranks second among all cancers and approximately 0·286 million CRC patients died in 2020(Reference Cao, Chen and Yu2).
The cause of CRC is complex and includes lifestyle and dietary risk factors, such as alcohol drinking, high red and processed meat intake, low fibre intake, and lower physical activity(3). There is an ongoing interest in the associations of fatty acids, especially n-3 PUFA with CRC risk. n-3 PUFA include α-18:3 n-3 (α-linolenic acid, ALA), long-chain n-3 PUFA such as 20:5 n-3 (EPA), 22:5 n-3 (docosapentaenoic acid, DPA) and 22:6 n-3 (DHA). The potential protective mechanisms of n-3 PUFA against CRC risk include inhibiting the production of eicosanoids derived from arachidonic acid, which may result in decreased cell proliferation and increased cell apoptosis(Reference Larsson, Kumlin and Ingelman-Sundberg4–Reference Bartsch, Nair and Owen7). Moreover, n-3 PUFA could inhibit the mammalian target of rapamycin signalling pathway which plays an essential role in cell growth, proliferation and angiogenesis(Reference Guertin and Sabatini8,Reference Zhang, Dai and Sun9) .
Previous studies mainly focused on investigating the association between dietary intake of n-3 PUFA and CRC risk(Reference Zhong, Fang and Pan10–Reference Shin, Cho and Sandin13). However, self-report dietary questionnaires used to assess dietary n-3 PUFA intake might have recall bias and measurement error(Reference Goris, Westerterp-Plantenga and Westerterp14). Additionally, n-3 PUFA can be obtained not only through dietary intake but also from endogenous synthesis. ALA, an essential n-3 PUFA from plants, is converted to long-chain n-3 PUFA by chain elongation and desaturation in the body. However, bioconversion of ALA to EPA and DHA is limited, and marine fish is the main dietary source of EPA and DHA. DPA is present in fairly low levels in most fish oils(Reference Shahidi and Ambigaipalan15), and circulating DPA is largely derived from endogenous metabolism. Therefore, circulating n-3 PUFA, which integrate endogenous and exogenous n-3 PUFA, can be better used to investigate the association between n-3 PUFA and disease. The alterations of fatty acids and their metabolites can be observed in tumour microenvironment, and significant differences were found in circulating n-3 PUFA, including ALA, EPA, and DPA, between CRC cases and controls(Reference Pakiet, Kobiela and Stepnowski16).
So far, only seven epidemiological studies(Reference Kojima, Wakai and Tokudome17–Reference Linseisen, Grundmann and Zoller23) have explored the associations between circulating levels of n-3 PUFA and CRC risk, and the results remained inconclusive. Some studies did not find significant associations of circulating ALA(Reference Hodge, Williamson and Bassett19,Reference Kuriki, Wakai and Hirose21–Reference Linseisen, Grundmann and Zoller23) , EPA(Reference Kojima, Wakai and Tokudome17–Reference Butler, Yuan and Huang22), DPA(Reference Hodge, Williamson and Bassett19–Reference Kuriki, Wakai and Hirose21,Reference Linseisen, Grundmann and Zoller23) , DHA(Reference Aglago, Huybrechts and Murphy20–Reference Butler, Yuan and Huang22), long-chain n-3 PUFA(Reference Aglago, Huybrechts and Murphy20,Reference Kuriki, Wakai and Hirose21) and total n-3 PUFA(Reference Hodge, Williamson and Bassett19,Reference Kuriki, Wakai and Hirose21) with CRC risk. However, it was reported that plasma(Reference Hodge, Williamson and Bassett19) or erythrocyte(Reference Kuriki, Wakai and Hirose21,Reference Linseisen, Grundmann and Zoller23) DHA was inversely associated with CRC risk. Erythrocyte EPA and total n-3 PUFA were inversely associated with CRC risk(Reference Linseisen, Grundmann and Zoller23). Serum level of ALA, DPA, DHA, total n-3 PUFA(Reference Kojima, Wakai and Tokudome17) and the whole blood level of long-chain n-3 PUFA(Reference Hall, Campos and Li18) were found to be significantly inversely related to CRC risk only among males. A meta-analysis including the above-mentioned five studies(Reference Kojima, Wakai and Tokudome17–Reference Aglago, Huybrechts and Murphy20,Reference Butler, Yuan and Huang22) indicated that blood levels of n-3 PUFA was inversely associated with CRC risk(Reference Kim and Kim24). Additionally, of seven studies(Reference Kojima, Wakai and Tokudome17–Reference Linseisen, Grundmann and Zoller23), four were conducted in Western countries(Reference Hall, Campos and Li18–Reference Aglago, Huybrechts and Murphy20,Reference Linseisen, Grundmann and Zoller23) . To date, no study has reported the relationship between serum levels of n-3 PUFA and CRC risk among Chinese population. In consideration of the ethnic, genetic and dietary pattern differences, more studies are needed to perform in Chinese population.
In this context, this case–control study was conducted among Chinese population to assess whether the risk of CRC was associated with serum levels of n-3 PUFA.
Methods
Study subjects
This ongoing case–control study began in July 2010 and was conducted in Sun Yat-sen University Cancer Center, China. The ascertainment and selection of CRC case and control subjects have been previously described(Reference Zhong, Fang and Pan10). CRC patients aged 30–75 years were recruited if they were histologically confirmed and diagnosed no more than 3 months. Patients who were not able to speak or understand Mandarin/Cantonese or had a history of other cancers were excluded. Participants with familial adenomatous polyposis or hereditary non-polyposis CRC were also excluded. Between July 2010 and May 2021, a total of 3174 cases were enrolled and 2833 eligible cases completed the interview, yielding a response rate of 89·26 %. Among them, 1303 blood samples were successfully collected. Due to high cost and time-consuming of measuring circulating fatty acids, not all blood samples were tested for serum fatty acids. A total of 680 case subjects were selected by random sampling from 1303 with blood samples and included in the final measurement and analysis.
Controls were selected from two groups. The first control group was recruited from the inpatients admitted to Departments of Vascular Surgery, Ophthalmology, Otorhinolaryngology and Plastic and Reconstructive Surgery during the same time period as the cases. They were admitted for a wide spectrum of non-neoplastic conditions, such as sudden deafness, chronic otitis media, chronic sinusitis, vocal cord polyp, varicose veins, deafness, giddiness, cholesteatoma of middle ear, nasal polyp and nasal septum deviation, etc. So far, no evidence has been found that these conditions were obviously related to a dietary cause that may affect circulating n-3 PUFA. The second control group was residents in the same community as the cases and recruited through written invitations, advertisements or referrals. Eligibility criteria for controls were the same as described for the cases except that they were not diagnosed with CRC. Eventually, after frequency-matching with cases on sex and age (5-year interval), 680 control subjects (591 hospitals-based controls and 89 community-based controls) with available blood samples were included in our analysis.
The procedures for this study were approved by the ethical committee of the School of Public Health, Sun Yat-sen University (approval number: 2019-018) and were conducted according to the principles of the Declaration of Helsinki. All participants signed written informed consent forms to participate in the study before investigation.
Data collection
To collect the demographics, lifestyles, fish oil supplement use, and family history of cancer of study participants, face-to-face surveys were performed by trained interviewers with a structured questionnaire. Reproductive history was also obtained among females. Relevant diagnosis, pathological findings and tumour node metastasis stage were abstracted from the patients’ medical records. We calculated BMI by using standard method (weight (kg)/ height (m2)). Occupational activity was categorised into non-working, sedentary, low intensity, moderate intensity and vigorous intensity. We also collected information on frequency (d/week) and duration (h/d) for household and leisure-time activity. The mean metabolic equivalent (MET) task-hours value of each activity was calculated by referring to the Compendium of Physical Activities(Reference Ainsworth, Haskell and Whitt25,Reference Ainsworth, Haskell and Herrmann26) .
An 81-item validated FFQ(Reference Zhang and Ho27) was used to assess dietary intake. Information on frequency of intake and commonly used portion size during the past 12 months before diagnosis for cases or interview for controls was collected and was used to estimate the daily intake of each food item. Red and processed meat consisted of pork, beef, lamb, organ meat, sausage, ham, bacon and hot dogs. The energy intake was calculated according to the 2002 China Food Composition Table(Reference Yang, Wang and Pan28). Photographs of commonly consumed foods with usual intake portions were provided to help estimate the amounts of food consumed. The FFQ has been used in previous studies(Reference Zhong, Fang and Pan10,Reference Huang, Abulimiti and Zhang29) .
Blood collection and serum n-3 PUFA measurements
Venous blood samples were drawn in the morning of the second day of hospitalisation after 12 h fasting and before any treatment or examination. Within 2 h after collection, the blood samples were centrifuged at the speed of 3000 rpm for 15 min at 4°C. All serum samples were then stored at −80°C for further measurement.
Serum fatty acids were assayed by GC. Lipids of serum samples (50 μl) were extracted using a method previously described(Reference Folch, Lees and Sloane Stanley30). Total lipids were extracted by chloroform methanol (2:1, v/v). After extraction, samples were evaporated to dryness with nitrogen at 25°C. After methyl esterification by 14 weight percentage boron trifluoride methanol, n-3 PUFA were measured by a gas chromatograph 7890A (Agilent Technologies) equipped with a methyl polysiloxane capillary column (60-m × 0·25-mm inside diameter; 0·15-µm thickness; DB-23, Agilent Technologies).
The temperature in front injector and detector was 250°C and 280°C, respectively. The flow rate of nitrogen carrier gas was kept at 0·8 ml/min. A total of 1 µl derivatised sample was injected at a 5:1 split ratio. The initial temperature of column was held at 50°C for 1 min and then increased to 170°C at the rate of 25°C/min and held at this temperature for 10 min, and finally increased to 230°C at the rate of 2·2°C/min and kept at this temperature for 17 min. A sample was injected into the column every hour. Peak retention time were identified by injecting known standards using ChemStation software (Agilent Technologies) for analysis. The individual fatty acids of samples were separated and identified by the comparison of their respective retention time with those of fatty acids methyl ester standard. According to the retention time, we manually integrated each peak to calculate peak area. Peak area normalisation method was used for calculating relative content of compositions. The serum level of each specific fatty acid was expressed as the percentages of all detected fatty acids. The intra-day and inter-day repeatability were measured using the same serum sample extracts in every forty samples being tested. In between, the quality-control samples after extraction were kept at–20°C. The precision was expressed with the CV for corresponding peak areas. The intra-day CVs of quality-control samples were 9·47 % for ALA, 6·64 % for EPA, 12·94 % for DPA and 3·92 % for DHA. The inter-day CVs of quality-control samples were 14·46 %, 14·67 %, 13·61 % and 9·68 % for ALA, EPA, DPA and DHA, respectively.
Four n-3 PUFA, including ALA, EPA, DPA and DHA, were measured. We calculated the content of long-chain n-3 PUFA (EPA + DPA + DHA) and total n-3 PUFA (the sum of ALA, EPA, DPA and DHA).
Statistical analysis
SPSS 26.0 (SPSS Inc.) was used to conduct statistical analysis. P-values ≤ 0·05 was considered statistically significant. All data were expressed as mean and standard deviation or median (interquartile range 25th and 75th percentiles, IQR). The χ 2 -test was used to compare categorical variables between cases and controls, and t test or Wilcoxon rank-sum test was used for continuous variables. Dietary red meat and processed meat intake were adjusted for total energy intake using the residual method(Reference Willett, Howe and Kushi31).
Serum levels of n-3 PUFA were converted into categorical variables by using quartiles (Q1–Q4) according to serum levels of each fatty acid in control subjects, separately for men and women. Multivariable logistic regression models were used to estimate the associations between serum n-3 PUFA and CRC risk, and the associations were expressed as OR and 95 % CI. The lowest quartile group served as the reference group. The following potential confounding variables were included in the multivariable model: age (years), sex (male/female), residence (rural/urban), socio-economic factors (including marital status, educational level, occupation and income), occupational activity, household and leisure-time activities, regular smoking, passive smoking, alcohol drinking, first-degree relative with cancer, BMI, total energy intake, red and processed meat intake, serum SFA, serum MUFA, and serum n-6 PUFA. Linear trend was tested by entering the median of each quartile as a continuous variable in the regression models.
A sex-stratified analysis was performed. Multiplicative models were used to evaluate the interaction effect between sex and serum n-3 PUFA in relation to the odds of having CRC by including the product term in multivariable logistic regression models. Analysis by tumour site (colon cancer and rectal cancer) was also conducted. To test for heterogeneity in the OR by tumour subsite, we regarded colon/rectal status as the dependent variable (outcome) in the logistic regression models. We had greater than 99 % power to detect OR of 0·34, 0·48, 0·39 and 0·31 for the association of serum ALA, DHA, long-chain n-3 PUFA and total n-3 PUFA with CRC risk. Our sample gave us 70·88 % and 98·50 % power to detect the OR of 0·57 and 0·48 for the association of serum EPA and DPA with CRC risk at P < 0·05 (two-tailed).
Results
The characteristics of CRC case and control subjects are presented in Table 1. The proportions of cases who had been married, had low incomes, had a history of cancer in first-degree relatives and had lower education were significantly higher than those in the control subjects. Meanwhile, cases had a higher frequency of regular and passive smoking, engaged in fewer occupational activities, and had higher consumption of red and processed meat. More than 80 % of cases (565, 83·09 %) were in tumour node metastasis stage 0/I–III. Among female subgroup, more cases had earlier menarche age.
Characteristics and selected risk factors of study subjects
MET, metabolic equivalent task; IQR, interquartile range 25th and 75th percentiles; TNM, tumour node metastasis; NOS, not otherwise specified.
Normally distributed mean values and standard deviations; non-normally distributed medians and interquartile range 25th and 75th percentiles; numbers and percentages.
* Continuous variables were evaluated using t test or Wilcoxon rank-sum test. Categorical variables were evaluated using χ2 test. P < 0·05: significant.
† The consumption was adjusted for total energy intake by the regression residual method.
‡ Among female subgroup.
As shown in Table 2, serum SFA were significantly higher in cases compared with controls. Cases had significantly lower serum levels of ALA, EPA, DPA, DHA, long-chain n-3 PUFA, total n-3 PUFA and total n-6 PUFA compared with control subjects.
Serum levels of detected fatty acids among colorectal cancer cases and controls
IQR, interquartile range 25th and 75th percentiles; ALA, α-linolenic acid; DPA, docosapentaenoic acid; LC n-3 PUFA, long-chain n-3 PUFA.
Non-normally distributed medians and interquartile range 25th and 75th percentiles.
* Wilcoxon rank-sum test comparing the median levels between cases and controls.
† Among total detected serum fatty acids. The serum level of fatty acid was expressed as the percentages of all detected fatty acids.
Table 3 shows the results of the associations between serum levels of n-3 PUFA and the risk of CRC. Compared with the lowest quartile, the odds of having CRC in the highest quartile were 70 % lower for serum ALA, 43 % lower for EPA, 41 % lower for DPA, 62 % lower for DHA, 68 % lower for long-chain n-3 PUFA and 74 % lower for total n-3 PUFA. Additionally adjusting for serum SFA, MUFA and n-6 PUFA did not change the results of the associations between serum levels of n-3 PUFA and the risk of CRC, except that the association between serum EPA and odds of having CRC attenuated to statistically insignificant (adjusted ORQ4 v. Q1 = 0·72, 95 % CI = 0·50, 1·04, P for trend = 0·060).
Association between serum n-3 PUFA and colorectal cancer risk
ALA, α-linolenic acid; DPA, docosapentaenoic acid; LC n-3 PUFA, long-chain n-3 PUFA.
OR and 95 % CI.
* OR1 adjusted for age, sex, residence, occupation, educational level, marital status, income, occupational activity, household and leisure-time activities, regular smoking, passive smoking, alcohol drinking, first-degree relative with cancer, BMI, total energy intake, and red and processed meat intake.
† OR2 additionally adjusted for serum SFA, serum MUFA and serum n-6 PUFA.
Analysis stratified by sex found that serum ALA, DHA, long-chain n-3 PUFA and total n-3 PUFA were inversely related to odds of CRC among both males and females. However, serum EPA and DPA displayed significant and inverse associations with odds of CRC mainly among females. The adjusted OR values for the fourth quartile (v. the first quartile) were 0·56 (95 % CI = 0·32, 0·98) (P interaction = 0·069) for EPA and 0·22 (95 % CI = 0·12, 0·40) (P interaction < 0·001) for DPA, respectively (Table 4).
Association between serum n-3 PUFA and colorectal cancer risk stratified by sex
ALA, α-linolenic acid; DPA, docosapentaenoic acid; LC n-3 PUFA, long-chain n-3 PUFA.
OR and 95 % CI.
* Adjusted for age, residence, occupation, educational level, marital status, income, occupational activity, household and leisure-time activities, regular smoking, passive smoking, alcohol drinking, first-degree relative with cancer, BMI, total energy intake, red and processed meat intake, serum SFA, serum MUFA and serum n-6 PUFA.
† Additionally adjusted for age at menarche.
Among the 680 cases, 403 were diagnosed with colon cancer and 277 with rectal cancer. Subgroup analysis by cancer site showed that serum ALA, DPA, DHA, long-chain n-3 PUFA and total n-3 PUFA were inversely associated with the odds of having both colon and rectal cancer, except for EPA (Table 5).
Association between serum n-3 PUFA and colorectal cancer risk stratified by cancer site
ALA, α-linolenic acid; DPA, docosapentaenoic acid; LC n-3 PUFA: long-chain n-3 PUFA.
* Adjusted for age, sex, residence, occupation, educational level, marital status, income, occupational activity, household and leisure-time activities, regular smoking, passive smoking, alcohol drinking, first-degree relative with cancer, BMI, total energy intake, red and processed meat intake, serum SFA, serum MUFA and serum n-6 PUFA.
Discussion
The purpose of this study was to examine the association between serum level of n-3 PUFA and the risk of CRC among Chinese population with a relatively larger sample size. The results showed that serum total n-3 PUFA, ALA, DPA and DHA were statistically significantly inversely associated with CRC risk. In contrast, serum level of EPA displayed no significant association with the risk of CRC.
Previous studies(Reference Mika, Kobiela and Czumaj32–Reference Okuno, Hamazaki and Ogura34) showed that there were significant differences between CRC cases and controls in circulating n-3 PUFA, which indicated that it is necessary to further examine the association of circulating n-3 PUFA and CRC risk. However, few studies have examined the association between circulating ALA and CRC risk. A nested case–control study from Japan(Reference Kojima, Wakai and Tokudome17) found a significant inverse association between serum ALA and CRC risk among males, which is in line with the present study. A Mendelian randomisation study also reported that circulating ALA was inversely associated with CRC risk(Reference Khankari, Banbury and Borges35). However, some studies(Reference Hodge, Williamson and Bassett19,Reference Kuriki, Wakai and Hirose21–Reference Linseisen, Grundmann and Zoller23) did not find significant associations of circulating ALA with CRC risk. ALA cannot be synthesised endogenously in human body and therefore must be provided exogenously in the diet. The potential explanations for differences in results are likely to be associated with the difference of ALA intake across different populations. We previously(Reference Zhong, Fang and Pan10) reported that mean ALA intake was 1·0 g/d among Chinese men and women. Average ALA intake was 1·90 g/d among Japanese people aged 35–66 years(Reference Muramatsu, Yatsuya and Toyoshima36), whereas ALA intake was 0·76 g/d among Australia people aged 45–64 years(Reference Sekikawa, Steingrimsdottir and Ueshima37). Additionally, the relatively higher level of serum ALA in our study (case group: 0·29 %, control group: 0·36 %) compared with that in other studies (both cases and non-case group were 0·15 %)(Reference Hodge, Williamson and Bassett19) may help explain the reduction of CRC risk with serum ALA in the present study. Additionally, relatively larger sample size in our study might contribute to find a significant association and narrower 95 % CI. For example, the previous study had wider 95 % CI (OR 0·39, 95 % CI 0·16, 0·91) due to smaller sample size (161 cases)(Reference Kojima, Wakai and Tokudome17). Other studies did not find significant association between ALA and CRC risk which might be related to the smaller sample size, with OR (95 CI%) of 0·96 (0·69, 1·33) (395 cases)(Reference Hodge, Williamson and Bassett19), and 1·18 (0·63, 2·21) (74 cases)(Reference Kuriki, Wakai and Hirose21), and 1·70 (0·84, 3·43) (350 cases)(Reference Butler, Yuan and Huang22).
The inverse associations of serum DPA, DHA and long-chain n-3 PUFA with likelihood of having CRC observed in our study was in line with a nested case–control study from Japan(Reference Kojima, Wakai and Tokudome17). However, some studies did not observe statistically significant associations of circulating DPA(Reference Hodge, Williamson and Bassett19–Reference Kuriki, Wakai and Hirose21,Reference Linseisen, Grundmann and Zoller23) , DHA(Reference Aglago, Huybrechts and Murphy20–Reference Butler, Yuan and Huang22,Reference Khankari, Banbury and Borges35) and long-chain n-3 PUFA(Reference Aglago, Huybrechts and Murphy20,Reference Kuriki, Wakai and Hirose21) with CRC risk. Our observation of the inverse association between total n-3 PUFA and odds of having CRC was consistent with a meta-analysis(Reference Kim and Kim24) and a nested case–control study from the European Prospective Investigation into Cancer and Nutrition (EPIC)(Reference Linseisen, Grundmann and Zoller23). Consistent with some previous studies(Reference Kojima, Wakai and Tokudome17–Reference Linseisen, Grundmann and Zoller23,Reference Khankari, Banbury and Borges35) , our study did not observe a statistically significant association between serum EPA and odds of having CRC. However, in the above-mentioned nested case–control study, erythrocyte EPA was found to be inversely related with CRC risk(Reference Linseisen, Grundmann and Zoller23). In a Mendelian randomisation study with individuals of European ancestry, circulating EPA and DPA were found to be positively associated with CRC risk(Reference Khankari, Banbury and Borges35). One possible explanation for our observation of the inverse association between total and individual long-chain n-3 PUFA and odds of having CRC might be related to higher dietary n-3 PUFA intake due to higher consumption of fish in eastern compared with Western populations. Based on the national nutritional surveys, average fish intake was 92–108 g/d among Japanese middle-aged men in 2002(Reference Sekikawa, Steingrimsdottir and Ueshima37). Mean fresh fish intake was 77·51 g/d among Chinese population aged 30–75 years(Reference Xu, Fang and Chen38). However, average fish intake was 37·1 g/d and 23·7 g/d from the EPIC in France(Reference Aglago, Huybrechts and Murphy20) and among Australian population aged 18–59 years(Reference Black, Zhao and Peng39). Meanwhile, the sample size in our study is larger than that of a case–control study from Japan with seventy-four cases(Reference Kuriki, Wakai and Hirose21) and one study from France with 461 cases(Reference Aglago, Huybrechts and Murphy20). Both of these studies(Reference Aglago, Huybrechts and Murphy20,Reference Kuriki, Wakai and Hirose21) did not find significant association between long-chain n-3 PUFA and CRC risk which might be due to smaller sample size. Moreover, the relatively higher level of serum total n-3 PUFA and relatively large sample size in our study compared with previous studies(Reference Hodge, Williamson and Bassett19,Reference Kuriki, Wakai and Hirose21) may help explain our observed reduction of the odds of having CRC with serum total n-3 PUFA.
The potential beneficial effects of n-3 PUFA against the risk of CRC are biologically plausible. n-3 PUFA might reduce CRC risk by inhibiting the COX-2 enzyme and the production of eicosanoids that are derived from arachidonic acid(Reference Larsson, Kumlin and Ingelman-Sundberg4–Reference Bartsch, Nair and Owen7). A clinical trial of fish oil supplementation observed reduced proliferation in the rectal mucosa of patients diagnosed with sporadic colorectal adenomas(Reference Anti, Armelao and Marra40). There are other mechanisms by which n-3 PUFA may decrease CRC risk, including the inhibition of ornithine decarboxylase, decreased bile acid excretion and NF-κB activity, the alteration of protein kinase C activity, the activation of peroxisome proliferator-activated receptor α and γ, and the reduced nitric oxide production(Reference Larsson, Kumlin and Ingelman-Sundberg4,Reference Rose and Connolly6,Reference Bartsch, Nair and Owen7,Reference Reddy41) . Recent experimental animal study has indicated that n-3 PUFA elicit anti-CRC effect through regulating the DNA methylation process(Reference Huang, Wen and Chen42) and modulating profiles of eicosanoid metabolites(Reference Wang, Yang and Nimiya43). Moreover, the mammalian target of rapamycin signalling pathway plays a key role in physiological and pathological processes of CRC(Reference Francipane and Lagasse44). n-3 PUFA have been found to down-regulate mammalian target of rapamycin signalling pathway in CRC cells(Reference D’Angelo, Piazzi and Pacilli45–Reference Liu, Zhou and Zhang47) and contribute to the suppression of tumour initiation and progression(Reference Aoki, Tamai and Horiike48) and the reduction of proliferation in colon cancer cell lines(Reference Gulhati, Cai and Li49,Reference Roulin, Cerantola and Dormond-Meuwly50) .
In our study, significantly inverse associations of serum EPA and DPA with odds of having CRC were only observed among females. No clear explanation exists for this point. It has been reported that female sex hormones might play a role in the aetiology of CRC(Reference McMichael and Potter51,Reference Tamakoshi, Wakai and Kojima52) . Several studies suggested that women have higher endogenously synthesis of EPA and DHA than men(Reference Burdge and Calder53,Reference Giltay, Gooren and Toorians54) . Previous studies suggested that human can convert ALA to DHA in the liver predominantly(Reference Burdge, Jones and Wootton55–Reference Salem, Wegher and Mena57). After initiation of oral ethinyl estradiol treatment, conversion of EPA to DHA increased(Reference Giltay, Gooren and Toorians54). Further research is needed to clarify this issue.
The risk of CRC might differ by cancer type(Reference Inoue, Tajima and Hirose58,Reference Potter59) . Some studies showed that plasma DHA displayed inversely significant associations with rectal cancer risk(Reference Hodge, Williamson and Bassett19) and plasma ALA was inversely associated with colon cancer risk(Reference Butler, Yuan and Huang22). But several studies did not find significant associations of plasma ALA(Reference Hodge, Williamson and Bassett19), EPA(Reference Hodge, Williamson and Bassett19,Reference Butler, Yuan and Huang22) , DPA(Reference Hodge, Williamson and Bassett19), DHA(Reference Butler, Yuan and Huang22) and total n-3 PUFA(Reference Hodge, Williamson and Bassett19) with the risk of colon or rectal cancer. Plasma EPA, DPA, DHA and long-chain n-3 PUFA were not associated with colon cancer risk(Reference Aglago, Huybrechts and Murphy20). The present study also showed no significant differences between serum n-3 PUFA and the risks of colon and rectal cancer.
Our study had some strengths. First, this is the first study to comprehensively investigate the association between serum n-3 PUFA and CRC risk among Chinese people. Second, the methods commonly used to measure fatty acids include HPLC, GC–MS and GC. Due to the high cost and professional maintenance of GC-MS and HPLC, GC is therefore the most commonly used method for the analysis of medium- and long-chain fatty acids(Reference Arab and Akbar60). Third, the relatively large number of CRC cases allowed subgroup analyses by sex and tumour site. Fourth, detailed information on potential CRC risk factors was collected and could be adjusted in the multivariable models.
The limitations of our study also need to be acknowledged. Firstly, all CRC patients recruited from Sun Yat-sen University Cancer Center might lead to selection bias. However, as the largest cancer centre in Southern China, the CRC patients admitted in this cancer centre have similar clinical characteristics to those of other big hospitals in Guangdong province or in other parts of mainland China(Reference Xu, Jiang and Zhong61,Reference Li, Huang and Shi62) , which might help to minimise selection bias. Secondly, measurement errors might be present in the process of detecting serum fatty acids. To reduce this bias, laboratory technician was blind to the diagnostic status of study subjects to ensure consistency in the sample preparation. Additionally, all samples were assayed by the same technician and quality controls were also applied. Thirdly, serum samples used in our study were stored at –80°C for 5–12 years. Iso et al. (Reference Iso, Sato and Umemura63) measured serum fatty acids in 1990 and 1998 using thirty-one duplicated serum aliquots frozen at −80°C in 1990. The results showed that no changes were seen for n-3 PUFA (12·8 % v. 12·3 %, P = 0·140). This indicated that serum level of n-3 PUFA might be a stable biomarker. Fourthly, although a wide range of possible confounding factors were adjusted in the models, potential unmeasured confounders cannot be excluded. Fifthly, we had only serum samples available for fatty acids analyses. Fatty acids content can be measured in erythrocyte which reflect dietary intake over a longer period(Reference Arab and Akbar60). However, previous studies indicated good correlations between serum and erythrocyte n-3 PUFA(Reference Yanagisawa, Shimada and Miyazaki64). Sixthly, one of the major drawbacks of fatty acids measured from serum and not the putative target tissue, in this case the colon and rectum, is the associated time factor. Serum fatty acids reflect what has been consumed over a period of weeks to days, whereas tissue fatty acid levels may reflect what has been consumer in months to years. However, measuring adipose tissue level of n-3 PUFA is more invasive, and fatty acids profiles might differ across sites within a person(Reference Arab and Akbar60). Furthermore, longitudinal intra-subjects analysis illustrated that serum novel circulating long-chain fatty acids levels are relatively stable over the short term of up to 90 weeks in both late-stage CRC patients and healthy controls(Reference Ritchie, Heath and Yamazaki65). Seventhly, due to the case–control study design, CRC cases might either change their diets or n-3 PUFA metabolism might be affected by cancer itself. Therefore, the association between n-3 PUFA and CRC risk should be interpreted with caution, and further prospective studies are needed among Chinese population.
In summary, our findings suggest the protective effect of serum ALA, DPA, DHA, long-chain n-3 PUFA and total n-3 PUFA on the risk of CRC in Chinese population. Our study indicates that serum n-3 PUFA might play a critical role in the incidence of CRC and may have significant implications for directing scientific diets of CRC prevention. In the light of existing evidence, we recommend to have an increased consumption of n-3 PUFA-rich food such as ALA-rich oil or marine fish to improve health. Further prospective studies are still needed to clarify the relationship between serum n-3 PUFA and the risk of CRC.
Acknowledgements
The authors express their appreciation to the study subjects for their participation. The authenticity of this article has been validated by uploading the key raw data onto the Research Data Deposit public platform (www.researchdata.org.cn), with the approval RDD number as RDDB2022927062.
This work was supported by the National Natural Science Foundation of China (No: 81973020, 81871991) and Guangdong Basic and Applied Basic Research Foundation (No: 2021A1515011751). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Conceptualisation: C-X. Z. and D-D. S.; data curation: C.-X. Z., R-L. Z. and Z-L. Z.; formal analysis: D-D. S. and Z-L. Z.; funding acquisition: C-X. Z.; investigation: D-D S., Y-L. J., T. D., Z-L. Z. and T. M.; methodology: C-X. Z. and D-D. S.; project administration: C-X. Z.; resources: Y-J. F., Q-J. O. and C-X. Z.; supervision: C-X. Z. and Y-J. F.; visualisation: C-X. Z. and D-D. S.; Writing – original draft: D-D. S.; Writing–review & editing: C-X. Z. and Y-J. F.
There are no conflicts of interest to declare.
