Colorectal cancer (CRC) ranks third and second in incidence and mortality worldwide, respectively, and it was estimated that more than 1·9 million cases and 935 000 deaths were due to CRC in 2020(Reference Sung, Ferlay and Siegel1). A high incidence of CRC is expected to occur in high or very high Human Development Index countries compared with low or medium Human Development Index countries, where it may be considered a marker for socio-economic development(Reference Sung, Ferlay and Siegel1). A similar trend has been reported in Korea, where CRC is one of the major cancers with an age-standardised incidence rate of 28·7/100,000 for both sexes in 2019(Reference Kang, Won and Lee2).
There are several risk factors associated with the development of CRC(3). It is widely accepted that environmental factors, especially diet, play a certain role in CRC pathogenesis, among which a Western-style diet has been emphasised to have an impact on cancer progression(Reference Khil, Kim and Hong4,Reference Kant and Deepa5) . High sugar, red and processed meat, and refined grain consumption is a major determinant for the causal link between a Western-style diet and CRC. In contrast, a diet with high fruit, vegetable, wholegrain, and fibre consumption is recognised to have a protective effect against CRC development(Reference Mehta, Song and Nishihara6).
Acid–base disequilibrium is a contributor to cancer development because it affects molecular activities such as adiponectin production and insulin-like growth factor 1 (IGF-1) levels(Reference Robey7). Notably, diet composition is clearly known to influence acid–base balance(Reference Remer8). An increased dietary acid load may be attributable to a high intake of animal proteins and refined grains because they are metabolised to precursors of acid. In contrast, high consumption of fruits, vegetables and dairy products contributes to reducing dietary acid load because they are key sources of K-, Ca- and Mg-producing alkaline precursors(Reference Keramati, Kheirouri and Musazadeh9). Specifically, lysine, arginine and histidine are amino acids that promote acid production because they generate hydrochloric acid when metabolised. Additionally, methionine and cysteine contain sulphur and increase sulphuric acid. Thus, animal foods are major contributors to acid load because these foods are rich in these amino acids(Reference Osuna-Padilla, Leal-Escobar and Garza-García10). For example, sulphur-containing amino acids such as homocysteine, methionine and cysteine were reported to be abundant in meat and meat products. In comparison with certain grains and legumes, the methionine and cysteine concentrations are 2- to 5-fold higher in eggs and meat. Skinless chicken breast and tuna have 4·94 and 6·48 mg of methionine per kilocalorie, respectively, while the methionine content is below 1 mg per kilocalorie in lentils, pinto beans, brown rice and maize(Reference Müller, Zimmermann-Klemd and Lederer11). Notably, citric acid is a mild tricarboxylic acid that is naturally abundant in citrus fruits(Reference Penniston, Nakada and Holmes12). However, its influence on high acid load has not been well elucidated in the literature. In contrast, fruits and vegetables have been shown to have an alkalinising effect because they contain potassium salts (citrate and malate), which remove hydrogen ions when metabolised in the body. Thus, the potassium concentration reflects the alkalising capacity of these foods(Reference Osuna-Padilla, Leal-Escobar and Garza-García10,Reference Müller, Zimmermann-Klemd and Lederer11) . Furthermore, in comparison with animal proteins, plant proteins contain a high concentration of glutamate, which requires hydrogen ions for its metabolism and may have a neutral impact(Reference Hamidianshirazi and Ekramzadeh13). Thus, a Western diet, characterised by high consumption of acidogenic foods and low consumption of alkalising foods, can potentially impact metabolic function due to an imbalance in acid–base equilibrium(Reference Lee and Shin14).
Body buffer systems control the blood pH and are maintained in a narrow range of 7·35–7·45, which prevents acidosis (pH < 7·35) or alkalosis (pH > 7·45). Low-grade metabolic acidosis could be defined as a blood pH close to the lower limit (7·35). Diet has been well documented as a major factor that may have an impact on this condition(Reference Ronco, Martínez-López and Calderón15,Reference Carnauba, Baptistella and Paschoal16) . Specifically, an increased dietary acid load may result in an increase in (H+) and a reduction in (HCO3−))(Reference Carnauba, Baptistella and Paschoal16). Animal foods are high in sulphur-containing amino acids, which are contributors to increased hydrogen ion (H+) levels, whereas fruits and vegetables possess the potential to produce bases due to their high K and Mg content, which act as hydrogen ion consumers during metabolism(Reference Jafari Nasab, Rafiee and Bahrami17). This metabolic condition can be estimated and represented through the potential renal acid load (PRAL) and net endogenous acid production (NEAP), which have been reported to be valid and straightforward methods to identify the acid load of the diet(Reference Ronco, Martínez-López and Calderón15,Reference Remer and Manz18,Reference Frassetto, Todd and Morris19) .
To date, attention has been drawn to the link between acid–base abnormalities and cancer, especially a long-term imbalance(Reference Jafari Nasab, Rafiee and Bahrami17). However, the majority of previous studies seem to be restricted to breast cancer(Reference Park, Steck and Fung20–Reference Wu, Hsu and Pierce23), and some studies focus on lung, bladder, gastric, oral, pancreatic, and oesophageal cancers and glioma(Reference Ronco, Martínez-López and Calderón15,Reference Ronco, Storz and Martínez-López24–Reference Milajerdi, Shayanfar and Benisi-Kohansal29) . Evidence of the association of diet-induced acid–base imbalance with CRC is limited. To our knowledge, only two case–control studies have been conducted to investigate this association(Reference Jafari Nasab, Rafiee and Bahrami17,Reference Ronco, Martínez-López and Calderón30) . Thus, it is unclear whether there is a causal association between diet-dependent acid load and CRC progression. Furthermore, there is a lack of consideration regarding diet-dependent acid load in CRC in Korea, where a high incidence of CRC and a nutrition transition have been reported(Reference Kang, Won and Lee2,Reference Kim, Kim and Kim31) . Therefore, our study aimed to explore whether colorectal carcinogenesis is attributable to a diet with a high acid load in a Korean population.
Materials and methods
Study design and participants
This was a case–control study performed at the National Cancer Center (NCC) in Korea. We recruited patients with a new diagnosis of CRC by endoscopic biopsies from August 2010 to August 2013 at the Center for Colorectal Cancer, NCC. We selected controls from among participants who visited the Center of Cancer Prevention and Detection at the NCC for their health examination provided by the National Health Insurance Cooperation and who were verified to be cancer-free from October 2007 to December 2014. Of the 15 271 individuals (1070 CRC patients and 14 201 healthy participants) who agreed to participate, we excluded those who did not meet the inclusion criteria as follows: 5189 participants (145 cases and 5044 controls) with incomplete dietary information and 122 participants (2 cases and 120 controls) with improbable energy intake data. Of the remaining participants, controls were matched with cases by age (±5 years) and sex using a ratio of 2:1. A total of 923 cases and 1846 controls were included in the final analysis. The details of the case–control recruitment were described previously(Reference Kim, Lee and Oh32,Reference Jun, Lee and Oh33) . 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 NCC Korea (IRB numbers: NCCNCS-10-350 and NCC 2015-0202). Written informed consent was obtained from all subjects/patients.
Data collection
Information on sociodemographic characteristics of cases and controls was obtained by using a structured questionnaire. The participants’ diets within the past 12 months were assessed by a semiquantitative FFQ that contained information on nine categories for the frequency of food consumption and three categories for portion size. The semiquantitative FFQ includes 106 food items, and its validity and reliability have been previously reported(Reference Ahn, Kwon and Shim34). We calculated total energy and nutrient intake based on Can-Pro 5.0 software (Computer-Aided Nutritional Analysis Program, Korea Nutrition Society). Medical records were used to obtain information on specific clinical parameters of CRC.
Estimates of diet-dependent acid load
PRAL and NEAP have been documented as two indicators to estimate diet-dependent acid load. PRAL exhibits intestinal absorption rates of protein and K, Ca, Mg and phosphate dissociation. NEAP measures the potential acidification of protein and potential alkalinisation of potassium(Reference Lee and Shin14,Reference Frassetto, Todd and Morris19,Reference Remer, Dimitriou and Manz35) .
PRAL and NEAP were estimated based on the following formulas:
PRAL (mEq/d) = protein (g/d) × 0·4888 + phosphorus (mg/d) × 0·0366–potassium (mg/d) × 0·0205 - magnesium (mg/d) × 0·0263–calcium (mg/d) × 0·0125(Reference Remer, Dimitriou and Manz35).
NEAP (mEq/d) = (protein (g/d) × 54·5/potassium (mEq/d))–10·2(Reference Frassetto, Todd and Morris19).
Negative PRAL and low NEAP values imply a diet with alkaline-forming potential, whereas a diet with acid-forming potential is reflected by positive values of PRAL and high values of NEAP(Reference Park, Steck and Fung20).
Net acid excretion (NAE) is a variant of PRAL and is an alternative index to indicate acid load from diet(Reference Remer and Manz18). Thus, we performed a sensitivity analysis using this index.
NAE (mEq/d) = PRAL + (body surface area (m2) × 41 (mEq/d)/1·73 m2), in which body surface area was based on the Du Bois formula: 0·007184 × height0·725 × weight0·425(Reference Park, Steck and Fung20).
Statistical analyses
To estimate PRAL and NEAP, the intake of protein, P, Ca, K and Mg was adjusted for energy using the residual method(Reference Willett and Stampfer36,Reference Willett, Howe and Kushi37) . We conducted a comparison of the demographic characteristics of controls with those of cases by using the chi-square test and t test for categorical and continuous variables, respectively. To explore the relationship between dietary acid load and the odds of CRC, PRAL and NEAP were categorised into tertiles based on the distributions of those two variables among controls. We considered the lowest tertile group as the reference. Furthermore, we considered PRAL and NEAP scores as continuous variables to observe the odds of CRC with a 1 s d increase in these scores. We estimated OR and 95 % CI using unconditional logistic regression models after adjustment for possible confounders (age, sex, smoking status, occupation, education, income, marital status, first-degree family history of CRC, alcohol consumption, BMI and physical activity). The median values in each tertile of PRAL and NEAP were used to test for trends. We used multinomial logistic regression models to investigate the association of diet-dependent acid load with the odds of CRC stratified by anatomical site (proximal colon, distal colon and rectal cancers). Furthermore, we used various algorithms (random forest, extreme gradient boosting (XGBoost), decision tree, Gaussian Naive Bayes and AdaBoost classifier) to assess the accuracy of dietary acid load and sociodemographic characteristics in the classification of participants with CRC. Python software (version 3.7.9) with the scikit-learn library was used for this procedure. SAS software (version 9.4, SAS Institute) was used for other statistical analyses, and a two-sided P value less than 0·05 was considered significant.
Results
General characteristics of cases and controls
In comparison with healthy participants, CRC patients exhibited a higher proportion of first-degree family history of CRC (9·3 % v. 5·4 %, P < 0·001). Similarly, they were more likely to be previous alcohol consumers (14·0 % v. 9·2 %, P < 0·001). In contrast, the case group had a lower prevalence of high occupational, educational and income levels (20·5 % v. 26·1 %, 25·2 % v. 50·6 %, and 23·3 % v. 29·5 %, respectively, P < 0·001). Additionally, less physical activity was observed in the case group (33·7 % v. 56·7 %, P < 0·001) (Table 1).
General characteristics of the participants
CRC, colorectal cancer.
* t test and χ2 test are for continuous and categorical variables, respectively.
† Means ± sd.
Furthermore, a higher energy intake was observed in the case group than in the healthy group (P < 0·001). Similarly, the case group had higher dietary acid load values than the control group. PRAL and NEAP scores were 7·5 ± 8·9 v. 3·5 ± 12·3 (P < 0·001) and 52·8 ± 12·9 v. 48·7 ± 14·1 (P < 0·001), respectively (Table 2).
Comparison of total energy and diet-dependent acid load intakes between the case and control groups
PRAL, potential renal acid load; NEAP, net endogenous acid production; NAE, net acid excretion.
* Calculation of P values was based on Student’s t test.
Diet-dependent acid load and odds of colorectal cancer
Table 3 presents the CRC risk in relation to dietary acid load scores. A significantly increased OR of CRC was observed among those who had a higher dietary acid load. Specifically, a higher risk of CRC was observed with a 1-sd increase in PRAL and NEAP (OR (95 % CI) = 1·40 (1·26, 1·57) and 1·28 (1·16, 1·42), respectively). With regard to the tertile group, significant associations were consistently found for PRAL in model 1 (unadjusted model; OR = 2·56, 95 % CI 2·07, 3·18, P for trend < 0·001) and model 2 (OR = 2·31, 95 % CI 1·79, 2·99, P for trend < 0·001). When stratified by sex, although a positive association emerged for males and females, a stronger effect was found for females (OR (95 % CI = 3·09 (1·93, 4·94) and 1·71 (1·27, 2·31), respectively, P for trend < 0·001). Participants with higher NEAP values exhibited significantly elevated odds of CRC (the OR (95 % CI) in model 1 (unadjusted) and model 2 were 2·48 (2·00, 3·07) and 2·14 (1·66, 2·76), respectively, P for trend < 0·001). A stronger association was observed for females (OR = 3·13, 95 % CI 1·96, 5·00, P for trend < 0·05) than for males (OR = 1·63, 95 % CI 1·21, 2·19, P for trend < 0·05).
OR and 95 % CI for the association of diet-dependent acid load with CRC risk
CRC, colorectal cancer; PRAL, potential renal acid load; NEAP, net endogenous acid production.
Model 1, unadjusted model; model 2, adjusted for age, education, first-degree family history of CRC, occupation, physical activity, income, marital status, smoking status, alcohol consumption and BMI. For all subjects, model 2 was additionally adjusted for sex.
NAE is a variant of PRAL, which is an alternative index to indicate acid load from diet. We used this index for a sensitivity analysis, and consistent results were observed. The OR (95 % CI) were 1·30 (1·16, 1·46) for a 1-sd increase in NAE and 1·68 (1·28, 2·21) for the high tertile group compared with the low tertile group in model 2 (online Supplementary Table 1).
When stratified by anatomical site, dietary acid load scores were found to be related to increased odds of proximal colon, distal colon and rectal cancer. Specifically, the corresponding OR (95 % CI) of PRAL and NEAP values for proximal colon, distal colon and rectal cancers were 2·77 (1·69, 4·54), 2·17 (1·50, 3·14), and 2·24 (1·62, 3·10) and 2·46 (1·51, 4·03), 2·23 (1·54, 3·22), and 1·97 (1·43, 2·72), respectively (P for trend < 0·001). However, the significant association was restricted to colon cancer among males (PRAL score: OR = 1·84 (1·06, 3·17), P for trend = 0·014 for the proximal colon and OR = 2·75 (1·69, 4·48), P for trend = 0·001 for the distal colon). In contrast, dietary acid load exhibited a greater effect on rectal cancer in females (PRAL score: OR (95 % CI) for the proximal colon, distal colon and rectum were 3·48 (1·35, 9·00), 1·74 (0·94, 3·23), and 5·00 (2·47, 10·14), respectively) (Table 4).
OR and 95 % CI of CRC according to tertiles of diet-dependent acid load intake stratified by anatomical site
CRC, colorectal cancer; PRAL, potential renal acid load; NEAP, net endogenous acid production.
* Adjusted for age, income, education, first-degree family history of CRC, physical activity, marital status, smoking status, alcohol consumption, BMI, and occupation. For all subjects, the models were additionally adjusted for sex.
Furthermore, we evaluated the accuracy of dietary acid load and sociodemographic characteristics in the classification of participants with CRC. The importance of dietary acid load was expected considering the performance of the XGBoost model, with 80·2 % accuracy and an AUC of 0·876 (online Supplementary Table 2). Importantly, dietary acid load seemed to be more important than other variables in classifying cases and controls (data not shown).
Discussion
In our study, among 923 CRC cases and 1846 controls, higher dietary acid load scores were identified to be associated with CRC. Specifically, significantly increased odds of CRC were found as the PRAL score increased. A similar trend was observed for the NEAP score. Notably, females were more susceptible to this positive association than males. Furthermore, acidogenic diets appeared to affect rectal cancer more strongly than colon cancer in females.
The effect of dietary intake on health outcomes may be well reflected by overall quality rather than single foods or nutrients. Diet-dependent acid load is considered an indicator to measure the influence of overall acidogenic foods on cancer risk(Reference Park, Steck and Fung20). Recently, dietary acid load has attracted attention because it was identified as an aetiologic factor for breast cancer in a previous cohort study of 43 570 participants(Reference Park, Steck and Fung20). Dietary acid load was also determined to have an association with an elevated pancreatic cancer risk in another cohort of 95 708 individuals(Reference Shi, Wu and Hu27). Similarities were found for several types of cancers in previous case–control studies, such as bladder, gastric, oral, and oesophageal cancers and glioma(Reference Ronco, Storz and Martínez-López24–Reference Ronco, Martínez-López and Calderón26,Reference Ronco, Martínez-López and Calderón28,Reference Milajerdi, Shayanfar and Benisi-Kohansal29) . Although acid–base disequilibrium has been proposed as a factor driving cancer development and progression(Reference Robey7), epidemiological evidence is limited for CRC, making it difficult to emphasise the role of acidogenic diets in colorectal carcinogenesis. A previous study among 499 participants (259 cases and 240 controls) used PRAL as an indicator for the assessment of dietary acid load and indicated that increased risks of CRC and colorectal adenomas were attributable to higher PRAL scores(Reference Jafari Nasab, Rafiee and Bahrami17). Another case–control study among 611 cases and 2394 controls was conducted based on the established hypothesis that acidogenic diets may cause inflammation and cancer due to metabolic acidosis. The association of dietary acid load with CRC susceptibility was documented with respect to this hypothesis in this study(Reference Ronco, Martínez-López and Calderón30). Similar associations were observed in our study. Notably, the association of dietary pattern with CRC risk varied depending on the study design(Reference Tabung, Brown and Fung38). The majority of previous studies focused on the detriment effect of diet-dependent acid with case–control design; thus, further research with a cohort design is needed to confirm the findings. Moreover, our finding is in agreement with that in a previous study conducted in Korea, where a westernised pattern was reported to be related to CRC development(Reference Park, Lee and Oh39). A Western-style diet is typically associated with high consumption of red meat and processed meat and a low intake of vegetables and fruits. As a result, this diet is considered an acid-producing diet(Reference Lee and Shin14). Accumulating evidence indicates that a plant-based diet has an association with a reduction in dietary acid load. For example, a previous study among 8398 non-vegetarians and 191 lacto-ovo-vegetarians found that non-vegetarians exhibited higher median dietary acid load scores than those of vegetarians. Thus, plant-based diets were discussed as being associated with a lower acid load(Reference Storz and Ronco40). This is in agreement with recent studies, which concluded that a plant-based diet could serve as a potential strategy to reduce systemic acid load(Reference Storz and Ronco40–Reference Cosgrove and Johnston42). Dietary shifts towards plant-based nutrition were also documented to be more effective in reducing dietary-induced, low-grade metabolic acidosis in another study(Reference Storz, Ronco and Hannibal43). Similarly, a recent review reinforced the aforementioned association and further suggested that a greater emphasis should be placed on the importance of fruits and vegetables in preventing acid–base imbalance and chronic diseases. Specifically, it is deemed essential to implement policies that promote increased consumption of fruits and vegetables(Reference Carnauba, Baptistella and Paschoal16). Notably, vegetables were the primary food source, while lower consumption levels of processed meat and red meat were reported in the Korean population(Reference Gunathilake, Lee and Choi44,Reference Kim, Lee and Choi45) . Thus, a lower mean PRAL score was exhibited in our population compared with the other populations. Importantly, our study emphasised that a greater detrimental effect of dietary acid load on colorectal carcinogenesis was observed for females, especially rectal cancer.
With the global spread of the Western diet, potential mechanisms in relation to this association need to be elucidated. First, low adiponectin levels, which might be attributed to the consumption of animal protein and metabolic acidosis(Reference Keramati, Kheirouri and Musazadeh9), are associated with increased insulin resistance leading to elevated IGF-1 levels. IGF-1 and insulin bind to their receptors, leading to increased cell proliferation and inhibited apoptosis. Furthermore, low adiponectin levels result in reduced activity of peroxisome proliferator-activated receptor γ, which promotes vascular endothelial growth factor secretion and increases cancer risk(Reference Izadi, Farabad and Azadbakht46). The causal link between low levels of adiponectin and CRC was reported in a review of previous studies(Reference Izadi, Farabad and Azadbakht46). Second, an increased production of cortisol was indicated to be dependent on the protein content of the diet, which prominently comprised red and processed meats(Reference Robey7). As a result, a Western diet leads to elevated cortisol production, which can cause carcinogenesis by increasing the tryptophan metabolism pathway and driving downstream molecular events(Reference Robey7). Additionally, the immunosuppressive effect of cortisol can reduce immunosurveillance of cancer at an early stage, facilitate immune escape and acquire mutations(Reference Larsson, Lee and Kar47). Furthermore, cortisol contributes to insulin resistance and the activation of IGF-1, which induces colorectal carcinogenesis via mitogen-activated protein kinase and phosphoinositide 3-kinase pathways(Reference Robey7,Reference Jafari Nasab, Rafiee and Bahrami17,Reference Belfiore and Malaguarnera48) . Third, a long-term intake of high protein has been widely recognised to directly elevate IGF-1 production(Reference Keramati, Kheirouri and Musazadeh9). For example, a positive association between the level of plasma IGF-1 and a protein diet was reported in a previous study conducted among 753 men(Reference Giovannucci, Pollak and Liu49). Another study emphasised that sources of protein should be considered and that animal protein mainly accounted for this association(Reference Hoppe, Udam and Lauritzen50). In line with these findings, the literature has highlighted the impact of an acidogenic diet on prediabetes and diabetes development(Reference Hatami, Abbasi and Salehi-sahlabadi51,Reference Abshirini, Bagheri and Mahaki52) . Notably, prediabetes and diabetes were implicated as aetiological factors for CRC development(Reference Park, Cho and Woo53,Reference Tran, Gunathilake and Lee54) . Fourth, dietary acid load has been indicated to have an impact on insulin resistance levels, which is a well-established risk factor for cancer development. This is attributed to its role in promoting hyperinsulinaemia, elevating bioavailable IGF-1 levels and increasing the production of reactive oxygen species(Reference Keramati, Kheirouri and Musazadeh9).
Furthermore, a higher PRAL was associated with a diet with high processed meat and red meat consumption and low fruit and vegetable consumption. Thus, the observed associations of PRAL with CRC could potentially be attributed to these risk factors and protective factors, suggesting that PRAL may serve as one of the mechanisms linking these specific foods to the development of CRC(Reference Jafari Nasab, Rafiee and Bahrami17). Possible mechanisms may be proposed for these foods in relation to CRC as follows: plant foods exhibit a protective effect against CRC due to antioxidant and phytochemical compounds, which have been indicated to have the ability to reduce the concentration of plasma inflammatory markers and impact the activation of the NF-κB pathway, respectively. Additionally, it should be noted that fibre may promote cell cycle arrest and apoptosis and inhibit cell migration and invasion. Furthermore, it has been suggested that the production of high fat, polycyclic aromatic hydrocarbons and heterocyclic amines during high-heat cooking are key mechanisms underlying the detrimental effects of red meat and processed meat on health(Reference Jafari Nasab, Rafiee and Bahrami17).
Notably, a larger effect of diet-dependent acid load was found among females than among males in our study. Our study is consistent with a previous study, which indicated that a significant association was observed only in females(Reference Ronco, Martínez-López and Calderón30). This is in agreement with a previous study that compared the influence of low v. high dietary acid load on blood acid–base status. The levels of pH, bicarbonate and base excess were lower in participants who followed a diet with a high acid load compared with those with a low dietary acid load. In particular, during submaximal cycling, young women and elderly women experienced a more acidic blood condition(Reference Hietavala, Stout and Frassetto55). This response is explained by the difference in renal functional capacity between males and females, which may contribute to the difference observed. Specifically, the glomerular filtration rate is higher in males than in females. As a consequence, females are more sensitive to the effects of diet-dependent acid load(Reference Hietavala, Stout and Frassetto55). Furthermore, NEAP was observed to have a significant positive association with fat mass in females but not males(Reference Mansordehghan, Daneshzad and Basirat56). Notably, body fatness was indicated to be related to CRC in females(Reference Zhang, Wu and Giovannucci57). Additionally, lean body mass has been reported to be beneficial for health outcomes(Reference Lee, Keum and Hu58). Notably, dietary acid load was indicated to have a negative impact on lean body mass in females but not in males, although the diet was more acidogenic in males than in females. Bioavailable testosterone might partially protect males from dietary acid load-induced lean body mass loss, which may explain this discrepancy(Reference Faure, Fischer and Dawson-Hughes59).
Additionally, colon cancer and rectal cancer exhibited different levels of susceptibility to diet-dependent acid load. An agreement was observed in a previous study(Reference Ronco, Martínez-López and Calderón30). Although the reasons are unclear, the biological mechanisms underlying this difference may be the colon and rectum arising from the midgut and hindgut, respectively. Thus, differences exist in the receptor pattern of growth factors, and sensitivity to the effect of insulin could be a possible and plausible explanation(Reference Wei, Giovannucci and Wu60). Furthermore, the colon and rectum exhibit different functions and different durations of exposure to faeces. Alkaline mucus coats undigested matter when travelling through the colon. Thus, environmental factors impacting susceptibility may be partially explained by the difference in pH level(Reference Wei, Giovannucci and Wu60). Additionally, different genes participate in the oncogenesis of colon cancer and rectal cancer in different ways. Rectal cancer exhibits more nuclear β-catenin and immunohistochemical expression of p53 than colon cancer. Thus, the aetiology and molecular basis may vary by anatomical site(Reference Kapiteijn, Liefers and Los61). These biological hypotheses need to be clarified in further studies.
Our study is the first to explore and provide evidence regarding the detrimental effect of diet-dependent acid load on CRC progression stratified by anatomical site in the Korean population. Another strength was the use of a validated semiquantitative FFQ, which was designed to estimate nutrient intakes specifically for the Korean population. Thus, information on nutrient intake in our study may accurately reflect eating habits for the study population. Additionally, sufficient information on possible confounders was collected and adjusted for in our study. Furthermore, we used three indicators to estimate diet-dependent acid load, including PRAL, NEAP and NAE. However, this study had some limitations. First, although efforts were made to mitigate the limitations of a case–control study, selection bias and recall bias may be present. Second, when stratified by anatomical site, statistical power could have been affected in the investigation of specific associations for colon/rectal cancer due to a small sample size. Third, supplement use was not considered as a potential confounder because we had a low response rate regarding information on supplement use due to open-ended questions. Fourth, markers of metabolic acidosis, such as anion gap and bicarbonates, were not considered to validate diet-dependent acid load in our study because data were not available. However, metabolic acid load, which is measured from 24-h urine samples, is widely documented to be strongly correlated with PRAL and NEAP(Reference Lee and Shin14,Reference Frassetto, Todd and Morris19,Reference Remer, Dimitriou and Manz35) . Furthermore, participants with high PRAL scores were observed to have a lower pH of spot urine samples than participants with low PRAL scores in a previous study in Korea. Therefore, PRAL and NEAP are applicable to estimate diet-dependent acid load for Koreans(Reference Lee and Shin14).
In conclusion, the present study contributes to reinforcing epidemiological evidence regarding a detrimental effect of acidogenic diets on CRC. Additionally, a greater susceptibility to diet-induced acid load was suggested for females. Thus, it is important to pay attention to the balance of acidogenic and alkalinogenic foods in CRC prevention, especially for females.
Acknowledgements
This work was supported by International Cooperation & Education Program (NCCRI•NCCI 52210-52211, 2020) of National Cancer Center, Korea, and grants from National Cancer Center, Korea (2 210 990), 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 interests.
Supplementary material
For supplementary material/s referred to in this article, please visit https://doi.org/10.1017/S0007114523001691
