The potential importance of acid–base homeostasis to cardiometabolic risk factors has been recently suggested in the literature(Reference McCarty1, Reference Vormann and Goedecke2). Mild metabolic acidosis, which can be caused by diet(Reference Maurer, Riesen, Muser, Hulter and Krapf3–Reference Sicuro, Mahlbacher, Hulter and Krapf5), may adversely affect blood pressure(Reference Resnick, Gupta, Sosa, Corbett and Laragh6–Reference Sharma, Cetto, Schorr, Spies and Distler8), possibly by increasing cortisol production(Reference Maurer, Riesen, Muser, Hulter and Krapf3), increasing Ca excretion(Reference Cappuccio, Kalaitzidis, Duneclift and Eastwood9, Reference Oshima and Young10) or decreasing citrate excretion(Reference Taylor, Mount, Forman and Curhan11). Increased cortisol production caused by mild metabolic acidosis(Reference Maurer, Riesen, Muser, Hulter and Krapf3–Reference Sicuro, Mahlbacher, Hulter and Krapf5) may also have a detrimental influence on other cardiometabolic risk factors, including obesity and cholesterol(Reference Dimitriou, Maser-Gluth and Remer12–Reference Fraser, Ingram, Anderson, Morrison, Davies and Connell14).
Since acid–base status is markedly influenced by diet(Reference Remer and Manz15, Reference Frassetto, Todd, Morris and Sebastian16), diet-dependent acid–base load can be calculated based on dietary intake information. Remer and colleagues developed an equation for estimating potential renal acid load (PRAL), an indicator of dietary acid–base load, using the dietary intake of five nutrients (protein, P, K, Ca and Mg)(Reference Remer and Manz15, Reference Remer, Dimitriou and Manz17). In addition, Frassetto and colleagues proposed the ratio of dietary protein to K (Pro:K) as an indicator of dietary acid–base load(Reference Frassetto, Todd, Morris and Sebastian16). Both PRAL and Pro:K estimated from dietary intake information have been validated against objective measures of acid–base load determined from 24 h urine (i.e. PRAL and net acid excretion (NAE), respectively)(Reference Remer, Dimitriou and Manz17). Using these measures, an expected negative relationship of dietary acid–base load with bone health was demonstrated in several epidemiologic studies that relied on a dietary questionnaire for nutrient intake estimation(Reference New, MacDonald, Campbell, Martin, Garton, Robins and Reid18–Reference Alexy, Remer, Manz, Neu and Schoenau22).
Despite the potential influence of dietary acid–base load on cardiometabolic risk factors and the availability of dietary acid–base load measurements using dietary intake data, no study has examined the possible association between measures of dietary acid–base load and cardiometabolic risk factors. Here, we investigated the associations of measures of dietary acid–base load (i.e. PRAL and Pro:K), calculated using nutrient intake estimates obtained from a validated self-administered comprehensive diet history questionnaire (DHQ)(Reference Sasaki, Yanagibori and Amano23–Reference Sasaki, Ushio, Amano, Morihara, Todoriki, Uehara and Toyooka25) with several cardiometabolic risk factors including BMI, waist circumference, systolic and diastolic blood pressure, total, HDL, and LDL cholesterol, fasting TAG, fasting glucose and glycated Hb, using data gathered from a cross-sectional observational study of free-living young Japanese women.
Subjects and methods
Subjects
The present study was based on a cross-sectional multi-centre survey conducted from February to March 2006 and from January to March 2007 among female dietetic students from fifteen institutions in Japan. All measurements at each institution were conducted according to the survey protocol. Briefly, staff at each institution explained an outline of the survey to potential subjects. Those who responded positively were then provided detailed written and oral explanations of the survey's general purpose and procedure. The protocol of the study was approved by the Ethics Committee of the National Institute of Health and Nutrition, and written informed consent was obtained from each subject, and also from a parent for subjects aged < 20 years.
A total of 1176 Japanese women took part. For the present analysis, women aged 18–22 years were selected (n 1154), not only because dietetic students outside this age range are rare in Japan but also because their dietary and cardiometabolic characteristics may differ from those of dietetic students aged 18–22 years. We then excluded from the 1154 women aged 18–22 years those not completing survey questionnaires (n 1), those with extremely low or high reported energy intakes ( < 2092 or >16 736 kJ/d; n 2), those currently receiving dietary counselling from a doctor or dietitian (n 13), those with previously diagnosed diabetes, hypertension or CVD (n 1), and those without measurement of body height and weight (n 2). Additionally, women with missing information regarding cardiometabolic risk factors were excluded from the respective analyses (n 2 for BMI, n 2 for waist circumference, n 0 for systolic and diastolic blood pressure, n 16 for cholesterol (total, HDL and LDL), n 16 for TAG, n 15 for glucose, and n 16 for glycated Hb). Further, those providing non-fasting blood samples (n 34) were excluded from the fasting TAG and glucose analyses. Some women fell into more than one exclusion category. The final sample size was 1136 for BMI, waist circumference, and systolic and diastolic blood pressure, 1121 for cholesterol (total, HDL, and LDL) and glycated Hb, 1089 for fasting glucose, and 1088 for fasting TAG.
Dietary assessment
Dietary habits during the preceding month were assessed using a self-administered comprehensive DHQ(Reference Sasaki, Yanagibori and Amano23–Reference Sasaki, Ushio, Amano, Morihara, Todoriki, Uehara and Toyooka25). Responses to the DHQ, as well as to a lifestyle questionnaire, were checked at least twice for completeness. When necessary, forms were reviewed with the subject to ensure the clarity of answers. The DHQ is a sixteen-page structured questionnaire that consists of the following seven sections: general dietary behaviour; major cooking methods; consumption frequency and amount of six alcoholic beverages; consumption frequency and semi-quantitative portion size of 118 selected food and nonalcoholic beverage items; dietary supplements; consumption frequency and semi-quantitative portion size of nineteen cereals (rice, bread, and noodles), soup consumed with noodles, and miso (fermented soyabean paste) soup; and open-ended items for foods consumed regularly ( ≥ once/week), but not appearing in the DHQ(Reference Sasaki, Yanagibori and Amano23). The food and beverage items were selected as foods commonly consumed in Japan, mainly from a food list used in the National Nutrition Survey of Japan, and standard portion sizes were derived mainly from several recipe books for Japanese dishes(Reference Sasaki, Yanagibori and Amano23).
Estimates of dietary intake for a total of 150 food and beverage items (including five seasonings), energy, and nutrients were calculated using an ad hoc computer algorithm for the DHQ based on the Standard Tables of Food Composition in Japan(26). Information on dietary supplements and data from the open-ended questionnaire items were not used in the calculation of dietary intake(Reference Sasaki, Yanagibori and Amano23). Nutrient and food intake was energy-adjusted using the residual method(Reference Willett and Stampfer27). Detailed descriptions of the methods used to calculate dietary intake and the validity of the DHQ regarding nutrients have been published elsewhere(Reference Sasaki, Yanagibori and Amano23–Reference Sasaki, Ushio, Amano, Morihara, Todoriki, Uehara and Toyooka25). The Pearson correlation coefficients between the DHQ and the 3 d estimated dietary records for forty-seven women were 0·48 for protein, 0·59 for P, 0·68 for K, and 0·49 for Ca (data not available for Mg)(Reference Sasaki, Yanagibori and Amano23). The Pearson correlation coefficients between the DHQ and the 16 d weighed dietary records for ninety-two women were 0·52 for protein, 0·55 for P, 0·55 for K, 0·56 for Ca, and 0·56 for Mg (S. Sasaki, unpublished results). In addition, the Spearman correlation coefficients were 0·66 for meats, 0·55 for fish and shellfish, 0·38 for eggs, 0·61 for dairy products, and 0·40 for fruits, 0·57 for vegetables, and 0·46 for cereals in ninety-two women (S. Sasaki, unpublished results). Furthermore, the Pearson correlation coefficient between the DHQ and the 24 h urinary excretion for K was 0·40 in sixty-nine women(Reference Sasaki, Yanagibori and Amano24).
Calculation and validation of dietary acid–base load measures
Urinary NAE is an established index of net endogenous acid production(Reference Frassetto, Todd, Morris and Sebastian16, Reference Remer, Dimitriou and Manz17), which is difficult to measure directly(Reference Lennon, Lemann and Litzow28, Reference Kleinman, Lemann, Maxwell, Kleeman and Narins29). Because the sum of cations excreted in the urine equals the sum of anions, urinary NAE is also equal to the difference between the sum of the major urinary non-bicarbonate anions minus the sum of the non-titratable acid and non-ammonium cations(Reference Remer and Manz30). The amounts of these non-bicarbonate anions and mineral cations in urine (excluding organic acids (OA)) are primarily influenced by dietary nutrient intake(Reference Remer and Manz15, Reference Frassetto, Todd, Morris and Sebastian16). OA are largely independent of dietary acid load or macronutrient composition(Reference Remer and Manz15, Reference Remer, Dimitriou and Manz17, Reference Chalmers, Healy, Lawson and Watts31, Reference Berkemeyer and Remer32), and can be reasonably estimated from body surface area(Reference Berkemeyer and Remer32):
Thus, the estimate of the urinary difference in non-bicarbonate anions (without OA) and mineral cations can be considered an index of diet-induced acid load(Reference Remer, Dimitriou and Manz17). Remer and colleagues referred to this estimate as PRAL (i.e. NAE = PRAL+OA)(Reference Remer, Dimitriou and Manz17), and developed the equation for estimating PRAL from dietary information(Reference Remer and Manz15, Reference Remer, Dimitriou and Manz17, Reference Remer and Manz30):
The validity of PRAL estimated from this equation has been established against PRAL measured from 24 h urine(Reference Remer and Manz15, Reference Remer, Dimitriou and Manz17).
The rate of H2SO4 production from protein metabolism and the rate of bicarbonate generation from the metabolism of intestinally absorbed K salts are major and highly variable components of net endogenous acid production(Reference Kleinman, Lemann, Maxwell, Kleeman and Narins29). Based on this, Frassetto and colleagues proposed the ratio of dietary protein (g/d) to K (mEq/d) (i.e. Pro:K) as an index of diet-induced acid load(Reference Frassetto, Todd, Morris and Sebastian16). The validity of Pro:K has been established against NAE measured in 24 h urine(Reference Remer, Dimitriou and Manz17).
In the present study PRAL and Pro:K were used as measures of dietary acid–base load. PRAL and Pro:K were calculated according to the equations described above using crude nutrient intake data estimated from the DHQ. Higher values of PRAL and Pro:K mean more acidic dietary acid–base load. Calculated PRAL and Pro:K were then energy-adjusted using the residual method(Reference Willett and Stampfer27). Prior to the present analysis, the relative validity of PRAL and Pro:K estimated from the DHQ was examined against that from the 16 d weighed dietary records in ninety-two women aged 31–69 years. The Pearson correlation coefficient between the two methods was 0·35 for PRAL and 0·37 for Pro:K (S. Sasaki, unpublished results).
Cardiometabolic risk factors
Cardiometabolic risk factors were measured 1–3 d after completion of the questionnaires. Body height was measured to the nearest 0·1 cm with the subject standing without shoes. Body weight in light indoor clothes was measured to the nearest 0·1 kg. BMI was calculated as body weight (kg) divided by the square of body height (m). Waist circumference was measured at the level of the umbilicus to the nearest 0·1 cm. The measurement was taken at the end of a normal expiration while the subject was standing erect with her arms at her side and feet together. Systolic and diastolic blood pressure was measured on the left arm with an automatic device (Omron model HEM-770A; Omron Health Care, Kyoto, Japan) after the subject had been sitting quietly for ≥ 3 min. A second measurement was carried out about 1 min after the first, and the mean value of the two was used. Peripheral blood samples were obtained from subjects after an overnight fast. Blood was collected in evacuated tubes containing no additives, allowed to clot, and centrifuged at 3000 g for 10 min at room temperature to separate the serum. Blood samples for glycated Hb measurements were also collected in evacuated tubes containing no additives. In accordance with the survey protocol, blood samples were transported at − 20°C by car or airplane to ensure delivery to a laboratory in Tokyo, Japan (SRL, Inc. in the 2006 survey and Mitsubishi Kagaku Bio-Clinical Laboratories (MBCL), Inc. in the 2007 survey). Biochemical variables were assayed at SRL in the 2006 survey and MBCL in the 2007 survey within 1–2 d of collection to avoid significant degradation, as follows. Serum total cholesterol concentration was measured enzymically using a kit from Wako Junyaku Co. Ltd (Tokyo, Japan) at SRL and using a kit from Daiya Shiyaku Co. Ltd (Tokyo, Japan) at MBCL. Serum LDL- and HDL-cholesterol concentrations were measured enzymically using a kit from Daiichi Kagaku Co. Ltd at SRL and using a kit from Daiichi Kagaku Co. Ltd at MBCL. Serum TAG concentration was measured enzymically using a kit from Daiichi Kagaku Co. Ltd at SRL and using a kit from Kyowa Medex Co. Ltd (Tokyo, Japan) at MBCL. Serum glucose concentration was measured enzymically using a kit from Shino Tesuto Co. Ltd (Tokyo, Japan) at SRL and using a kit from Kanto Kagaku Co. Ltd (Tokyo, Japan) at MBCL. Glycated Hb was measured using whole blood by latex agglutination-turbidimetric immunoassay (Fuji Revio Co. Ltd (Tokyo, Japan) at SRL and Kyowa Medex Co. Ltd (Tokyo, Japan) at MBCL). In-house quality-control procedures for all assays were conducted at SRL in the 2006 survey and MBCL in the 2007 survey.
Other variables
In the lifestyle questionnaire, the subject reported her residential area, which was grouped into one of three regions (residential block: north (Kanto, Hokkaido, and Tohoku); central (Tokai, Hokuriku, and Kinki) or south (Kyushu and Chugoku)). The residential areas were also grouped into three categories according to population size (size of residential area: city with population ≥ 1 million; city with population < 1 million or town and village). Current smoking (yes or no) was self-reported in the lifestyle questionnaire. Physical activity was computed as the average metabolic equivalents (MET)-hours per day(Reference Ainsworth, Haskell, Leon, Jacobs, Montoye, Sallis and Paffenbarger33) on the basis of the frequency and duration of five different activities (sleeping, high- and moderate-intensity activities, walking, and sedentary activities) over the preceding month, as reported in the lifestyle questionnaire.
Statistical analysis
Measures of dietary acid–base load, i.e. PRAL and Pro:K, were examined in relation to ten cardiometabolic risk factors, namely BMI, waist circumference, systolic and diastolic blood pressure, cholesterol (total, HDL, and LDL), fasting TAG, fasting glucose, and glycated Hb. All statistical analyses were performed using SAS statistical software (version 8.2; SAS Institute Inc., Cary, NC, USA). Linear regression models were constructed using the PROC GLM procedure to examine the association between dietary acid–base load measures with cardiometabolic risk factors. For the analyses, subjects were categorized into quintiles according to the dietary acid–base load measures. The mean metabolic risk factor values (with standard errors) were calculated by quintiles of dietary acid–base load measures after multivariate adjustment for potential confounding factors. Confounding factors included residential block, size of residential area, survey year (2006 or 2007; because of the different laboratories with different kits used for blood analyses for the 2006 and 2007 surveys, even though there were no differences in the assay methods), current smoking, and physical activity (continuous). BMI (continuous) was added as a confounding factor in all analyses except for that for BMI itself. Waist circumference (continuous) was also added as a confounding factor in the analyses except for those for BMI and waist circumference. We initially intended to include estimated excretion of OA (the diet-independent acid–base load) as a confounding factor to investigate diet-dependent acid–base load (PRAL and Pro:K) and cardiometabolic risk factors. However, OA was strongly correlated with body height and weight and hence BMI (Pearson correlation coefficient with OA = 0·70, 0·95, and 0·68, respectively), as OA is just a function of the combination of height and weight as mentioned earlier. To avoid over-adjustment, we consequently did not include OA as a confounding factor. Since alcohol intake was extremely low (mean = 1·5 g/d), it was not considered as a confounding factor. Because the inclusion of measures of obesity (BMI, waist circumference, or both) as confounding factors did not influence the results materially, we present the full-adjustment models only. Linear trends with increasing levels of dietary acid–base load measures were tested for by assigning each participant a median value for the category and modelling this value as a continuous variable. All reported P values are two-tailed, and a value of P < 0·05 was considered statistically significant.
Results
Basic characteristics of all subjects (n 1136; those included in the analyses of BMI, waist circumference, and systolic and diastolic blood pressure) are shown in Table 1. Mean PRAL was 10·4 mEq/d, and mean Pro:K was 1·23 g/mEq. There was a strong correlation between these variables (Pearson correlation coefficient = 0·84). The potential confounding variables for all subjects are shown in Table 2 according to quintile of dietary acid–base load measure. Both PRAL and Pro:K were associated negatively with physical activity and positively with waist circumference. There was also a positive association between Pro:K and BMI. The dietary intakes of all subjects are shown in Table 3 according to quintile of dietary acid–base load measures. PRAL was associated positively with protein and negatively with K, Ca and Mg as well as P, while Pro:K was positively associated with K as well as protein. For foods, both PRAL and Pro:K showed a positive association with meats, eggs, cereals, and fish and shellfish, and a negative association with fruits, vegetables and dairy products. According to the quintiles of the dietary acid–base load measures, similar patterns were observed for potential confounding factors and dietary intake among those subjects included in the analyses of cholesterol (total, HDL, and LDL) and glycated Hb (n 1121), fasting TAG (n 1088), and fasting glucose (n 1089) (data not shown).
Table 1 Basic characteristics of subjects (n 1136)*
(Mean values and standard deviations)
MET, metabolic equivalents; Pro:K, ratio of dietary protein to K.
* n 1121 for cholesterol (total, HDL, and LDL) and glycated Hb; 1088 for fasting TAG; 1089 for fasting glucose.
† Energy-adjusted using the residual method.
‡ Calculated using crude nutrient intake values and then energy-adjusted using the residual method.
Table 2 Selected characteristics according to quintile of measures of dietary acid–base load (n 1136)
(Mean values and standard deviations)
MET, metabolic equivalents; Pro:K, ratio of dietary protein to K.
* For categorical variables, a Mantel–Haenszel χ2 test was used; for continuous variables, a linear trend test was used with the median value in each quintile as a continuous variable in linear regression.
† Calculated using crude nutrient intake values and then energy-adjusted using the residual method.
Table 3 Nutrient and food intake according to quintile of measures of dietary acid–base load (n 1136)
(Mean values and standard deviations)
Pro:K, ratio of dietary protein to K.
* A linear trend test was used with the median value in each quintile as a continuous variable in linear regression.
† Calculated using crude nutrient intake values and then energy-adjusted using the residual method.
‡ Energy-adjusted using the residual method.
The multivariate-adjusted mean values for cardiometabolic risk factors across quintiles of dietary acid–base load measures are shown in Table 4. After adjustment for potential confounding factors, higher PRAL and Pro:K (more acidic dietary acid–base loads) were associated with higher systolic and diastolic blood pressure (mean difference between the lowest and highest quintiles = 2·1 mmHg (P for trend = 0·028) and 1·6 mmHg (P for trend = 0·035) for PRAL, and 2·5 mmHg (P for trend = 0·012) and 2·3 mmHg (P for trend = 0·009) for Pro:K, respectively). In addition, PRAL showed an independent positive association with total and LDL cholesterol (mean difference = 59 mg/l (P for trend = 0·042) and 60 mg/l (P for trend = 0·021), respectively). Pro:K was positively associated with BMI and waist circumference independently of potential confounding factors (mean difference = 0·5 kg/m2 (P for trend = 0·024) and 0·8 cm (P for trend = 0·012), respectively). No significant associations were observed between PRAL or Pro:K and any of the other cardiometabolic risk factors examined.
Table 4 Cardiometabolic risk factors according to quintile of measures of dietary acid–base load (n 1136)*
(Mean values with their standard errors)
Pro:K, ratio of dietary protein to K.
* n 1121 for cholesterol (total, HDL, and LDL) and glycated Hb (224 in the first, second, fourth, and fifth and 225 in the third quintiles); 1088 for fasting TAG (217 in the first and fifth and 218 in the second, third, and fourth quintiles); and 1089 for fasting glucose (217 in the first and 218 in the second, third, fourth, and fifth quintiles). For potential renal acid load, median value in each quintile is almost the same (within < 0·1 mEq/d difference) in all analyses; for pro:K, median value in each quintile is the same.
† A linear trend test was used with the median value in each quintile as a continuous variable in linear regression.
‡ Calculated using crude nutrient intake values and then energy-adjusted using the residual method. Values are median.
§ Adjusted for residential block (north (Kanto, Hokkaido, and Tohoku), central (Tokai, Hokuriku, and Kinki), or south (Kyushu and Chugoku)), size of residential area (city with population ≥ 1 million, city with population with < 1 million, or town and village), survey year (2006 or 2007), current smoking (yes or no), and physical activity (total metabolic equivalent-h/d, continuous).
∥ Additionally adjusted for BMI (kg/m2, continuous).
¶ Additionally adjusted for waist circumference (cm, continuous).
Discussion
In a group of free-living young Japanese women, we found that higher PRAL and Pro:K (more acidic dietary acid–base loads) were associated with higher systolic and diastolic blood pressure after adjustment for possible confounding factors. We also found independent positive associations between PRAL and total and LDL-cholesterol, as well as between Pro:K and BMI and waist circumference. To our knowledge, this is the first study to examine the relationships between dietary acid–base load measures and cardiometabolic risk factors.
Consistent with previous studies(Reference Remer19–Reference Welch, Bingham, Reeve and Khaw21), the correlation between PRAL and Pro:K was quite high (Pearson correlation coefficient = 0·84–0·93), which indicates that these measures capture similar, but not the same, elements of dietary acid–base load. Given that only blood pressure was significantly associated with both PRAL and Pro:K in the present study, the finding on blood pressure may be more reliable than those on other cardiometabolic risk factors that are significantly associated with only one, either PRAL or Pro:K (i.e. total and HDL-cholesterol and BMI and waist circumference). Although both PRAL and Pro:K are established and reasonably valid measures of dietary acid–base load(Reference Remer, Dimitriou and Manz17), further methodological research may be helpful for clarifying which is the better indicator of diet-induced acid–base load.
Both PRAL and Pro:K were independently and positively associated with systolic and diastolic blood pressure in the present study. Several previous experimental studies have also indicated a significant rise in blood pressure caused by mild metabolic acidosis(Reference Resnick, Gupta, Sosa, Corbett and Laragh6–Reference Sharma, Cetto, Schorr, Spies and Distler8), which can be caused by diet(Reference Maurer, Riesen, Muser, Hulter and Krapf3–Reference Sicuro, Mahlbacher, Hulter and Krapf5). Although not established, several plausible mechanisms have been proposed, including increased cortisol production(Reference Maurer, Riesen, Muser, Hulter and Krapf3) increased Ca excretion(Reference Cappuccio, Kalaitzidis, Duneclift and Eastwood9, Reference Oshima and Young10) and reduced citrate excretion(Reference Taylor, Mount, Forman and Curhan11). Further research to elucidate the apparent influence of dietary acid–base load on blood pressure using a range of free-living populations is required.
We found an independent and positive association of total and HDL-cholesterol with PRAL, but not with Pro:K. Likewise, an independent and positive association was also found between BMI and waist circumference and Pro:K, but not with PRAL. No significant association with any of the other cardiometabolic risk factors was observed, including fasting TAG, fasting glucose, and glycated Hb. Increasing cortisol production caused by mild metabolic acidosis(Reference Maurer, Riesen, Muser, Hulter and Krapf3–Reference Sicuro, Mahlbacher, Hulter and Krapf5) might have a detrimental influence on cardiometabolic risk factors(Reference Dimitriou, Maser-Gluth and Remer12–Reference Fraser, Ingram, Anderson, Morrison, Davies and Connell14). Further investigation of this poorly understood field is warranted.
In the present study, PRAL showed a strong negative association with K (as well as Ca and Mg), but a weak association with protein and P. For foods, the negative association with fruits and vegetables was quite strong compared with that for other foods. Similar trends were observed for Pro:K. Thus, in Japanese populations, dietary acid–base load may be primarily determined by foods rich in K such as fruits and vegetables. This is somewhat different from Western populations, where PRAL was strongly associated not only negatively with fruits and vegetables, but also positively with meats(Reference Welch, Bingham, Reeve and Khaw21).
Mean PRAL and Pro:K were much more acidic in this study of young Japanese women (10·4 mEq/d and 1·2 g/mEq, respectively) than in other two studies of British middle-aged women (3·7 and − 7·6 mEq/d and 1·0 and 1·0 g/mEq, respectively)(Reference Remer19, Reference Welch, Bingham, Reeve and Khaw21). The intake of K, Ca, and Mg as well as P was 37–50 % lower, although protein intake was 27 % lower in the Japanese subjects than in the British population(Reference Welch, Bingham, Reeve and Khaw21), which seems to explain the more acidic dietary acid–base load in the present study.
Several limitations of the present study warrant mention. First, the cross-sectional nature of the study does not permit the assessment of causality owing to the uncertain temporality of the association. Nevertheless, there are biologically plausible mechanisms for the relationship between dietary acid–base load and the cardiometabolic risk factors, particularly blood pressure(Reference McCarty1, Reference Vormann and Goedecke2, Reference Resnick, Gupta, Sosa, Corbett and Laragh6–Reference Taylor, Mount, Forman and Curhan11), although while the premise for the hypothesis examined is that the acidosis is associated with increased cortisol production(Reference Maurer, Riesen, Muser, Hulter and Krapf3–Reference Sicuro, Mahlbacher, Hulter and Krapf5), no measurements for cortisol are unfortunately available, which would help back up findings. Second, our subjects were selected female dietetic students, not a random sample of Japanese women. In addition, because of our recruitment procedure, the exact response rate was unknown, which might have produced recruitment bias. Thus, these results may not apply to the general Japanese population, although our population was on average comparable with a representative sample of Japanese women with similar age range (18–29 years for dietary intake and 20–29 years for cardiometabolic risk factors), at least with regard to the intake of energy (mean: 7117 kJ/d), protein (64·0 g/d), P (894 mg/d), K (1965 mg/d), Ca (459 mg/d), and Mg (210 mg/d) and several cardiometabolic risk factors including BMI (20·9 kg/m2), systolic blood pressure (108·8 mmHg), diastolic blood pressure (67·0 mmHg), total cholesterol (1806 mg/l), HDL-cholesterol (689 mg/l), and glycated Hb (4·91 %) (data not available for other cardiometabolic risk factors)(34). Further, because the study population consisted of generally healthy persons, the clinical relevance of our findings remains to be elucidated. Nevertheless, our results should provide valuable insight from a prevention perspective. Third, as with other studies on dietary acid–base load(Reference New, MacDonald, Campbell, Martin, Garton, Robins and Reid18, Reference Macdonald, New, Fraser, Campbell and Reid20, Reference Welch, Bingham, Reeve and Khaw21), a semi-quantitative dietary assessment questionnaire (i.e. DHQ) was used to collect dietary data(Reference Sasaki, Yanagibori and Amano23–Reference Sasaki, Ushio, Amano, Morihara, Todoriki, Uehara and Toyooka25). As actual dietary habits were not observed, and the relative validity of the PRAL and Pro:K values derived from the DHQ against the 16 d weighed dietary records was also modest, the results should be interpreted with caution. It should be noted, however, at least for conventional nutrients, that the applicability of the DHQ is comparable to that of dietary questionnaires used in previous studies, in which the relative validity of PRAL and Pro:K was not examined(Reference New, MacDonald, Campbell, Martin, Garton, Robins and Reid18, Reference Macdonald, New, Fraser, Campbell and Reid20, Reference Welch, Bingham, Reeve and Khaw21). Additionally, the misreporting of dietary intake, particularly by overweight subjects, is a serious problem associated with self-report dietary assessment methods(Reference Livingstone and Black35). However, at least for dietary protein, K, and Na, BMI-dependent misreporting seems to be cancelled by energy-adjustment(Reference Murakami, Sasaki and Takahashi36). To minimize the influence of dietary misreporting as much as possible(Reference Murakami, Sasaki and Takahashi36), we used energy-adjusted values(Reference Willett and Stampfer27). Actually, the ratio of reported energy intake to estimated basal metabolic rate, a surrogate measure of the general quality of dietary data(Reference Livingstone and Black35) was not associated with PRAL or Pro:K (data not shown), which suggests that it is unlikely that dietary underreporting had a major effect on the observed associations. Further, because of the lack of a reliable composition table for dietary supplements in Japan, nutrient intake from dietary supplements was not included in the analysis. However, the percentage of subjects who consumed dietary supplements containing mostly protein and minerals during the preceding month was only about 2 %, and the exclusion of these supplement users did not change the results materially (data not shown). This suggests that it is unlikely that dietary supplements had a major effect on the findings. Moreover, although adjustments were attempted to compensate for a variety of potential confounding variables, residual confounding could not be ruled out. In particular, physical activity was assessed relatively roughly from only five activities, which may not have been sufficient. Finally, because the associations between diet-dependent acid load and cardiometabolic risk factors were examined using PRAL and Pro:K estimated from dietary intake, it is not known whether the observed associations were caused by a change in acid–base balance due to dietary intake or by some other mechanism. Additionally, the observed associations between PRAL and Pro:K and cardiometabolic risk factors may be simply due to the effects on cardiometabolic risk factors of intakes of some nutrients or foods associated with PRAL and Pro:K. For example, K and vegetable intake, of which not only the favourable effect on blood pressure is broadly recognized(Reference Appel, Moore and Obarzanek37–Reference Appel, Brands, Daniels, Karanja, Elmer and Sacks39) but which was also most strongly correlated with PRAL and Pro:K in the present study (Pearson correlation coefficients with PRAL or Pro:K = − 0·50 to − 0·74), was negatively associated with BMI and waist circumference (P for trend ≤ 0·001), although there was no association with blood pressure (systolic and diastolic) or cholesterol (total and LDL). However, it is not possible to understand what food sources or nutrients are driving the observed associations between dietary acid–base load and cardiometabolic risk factors, because dietary acid–base load used in the present study was derived from estimates of dietary intakes, and hence was associated with these dietary intakes. Thus the results of the present study should be cautiously interpreted without oversimplification.
In conclusion, after adjusting for possible confounding factors, we found that higher PRAL and Pro:K (more acidic dietary acid–base loads) were associated with higher systolic and diastolic blood pressure in free-living young Japanese women. We also found independent positive associations between PRAL and total and LDL-cholesterol, as well as between Pro:K and BMI and waist circumference. Because the cross-sectional nature of the present study does not allow causal inferences, any firm conclusions regarding the effects of dietary acid–base load on cardiometabolic risk factors require additional observational and experimental studies.
Acknowledgements
We thank Keika Mine, Yoko Hosoi, Mami Itabashi, Tomono Yahata, Asako Ishiwaki, and Kyoko Saito (National Institute of Health and Nutrition) for data collection. None of the authors had any personal or financial conflict of interest. This study was supported by grants from the Ministry of Health, Labour, and Welfare of Japan. K. M. contributed to the concept and design of the study, the study protocol, data collection and management, and coordinated the fieldwork, formulated the hypothesis, calculated the measures of dietary acid–base load, analysed and interpreted the data, and wrote the manuscript. S. S. is a principal researcher of this project and contributed to the concept and design of the study, the study protocol, and data collection and management, and the writing and editing of the manuscript. Y. T. and K. U. contributed to the concept and design of the study, the study protocol, and data collection. All authors contributed to the preparation of the manuscript and approved the final version submitted for publication.
