Effect of reduced-calcium and high-calcium cheddar cheese consumption on the excretion of faecal fat: a 2-week cross-over dietary intervention study

There is growing evidence that total fat, and saturated fat, consumed in different food matrices has different cardiometabolic and other health outcomes [1, 2]. The food matrix effect describes the interaction of the overall food structure and how the nutrients contained within, may differentially impact digestion and absorption [3]. The dairy food matrix is a particular example of this, with results from a number of randomized controlled trials and observational studies showing that fat from cheese consumption compared to butter is associated with differences in blood lipid profiles, whereby low-density lipoprotein-cholesterol (LDL-c) concentrations [4‐6] or total cholesterol [7] were lower following cheese intake compared to butter.

A 2015 meta-analysis of the available randomised controlled trials calculated that a weighted mean difference of 145 g (cheese vs. butter) resulted in a 6.5% reduction in LDL-c concentration [8]. Various mechanistic reasons for these differences have been postulated. These include the protein content of cheese [9, 10], the fermentation which may modify the gut microbiota [9], greater phospholipid content [11], and greater mineral (Ca and P) content of cheese [10] vs. butter. It is likely that all of these contribute to some degree, but data on these individual parameters is lacking to date, mostly as it is difficult to test the various hypothesis in a single study and or modify a single component within an experimental approach without impacting the other aspects.

With respect to the mineral content, particular emphasis has been given to dairy calcium [12‐14]. It is suggested that during digestion, this may interact with the fat in cheese to form insoluble calcium soaps, which in turn are excreted, resulting in reduced intestinal fat absorption (and, therefore, increased faecal fat excretion [12, 15]. Dairy calcium may also complex with phosphate to form CaP, which can bind intestinal bile acids, leading to their excretion in faeces, and resulting in the use of circulating cholesterol for do-novo synthesis of BA, subsequently lowering circulating cholesterol [16, 17].

While several studies to date have suggested that dairy calcium is associated with reduced LDL-c concentration [12, 13], or triglycerides [18], studies on fat excretion have not been conclusive. Dairy Ca was shown to increase faecal fat excretion in some studies [13, 14], but not significantly so in others [18]. However, these studies have used a variety of dairy foods, each with different matrices, which may have confounded the interpretation. For example, Soerensen and colleagues examined milk vs. cheese, while the study by Bendsen et al., provided the dairy fat from butter, and the Ca from a variety of low-fat dairy products. Few have examined Ca in cheese alone. Previously, we observed a significant matrix effect of cheese on blood lipids following a 6 week intervention study, controlling for other macronutrients and calcium [6]. Here, we were interested specifically in the effect of adjusting the calcium within the cheese matrix itself. The aim of this study, therefore, was to examine the impact of additional dairy calcium contained within the matrix of cheese (with a naturally enhanced Ca cheese), and outside of the matrix, using a naturally reduced Ca cheese plus a supplement, on faecal fat excretion (FFE). We hypothesized that FEE would be greater in the HCC and HCC + Supp diets compared to the RCC diet.

In this study, we aimed to examine the effect of increasing the calcium within the cheese matrix on faecal fat excretion, in a group of healthy, free-living males, who consumed a diet of higher calcium or reduced-calcium cheese for two weeks, or a reduced-calcium cheese plus a CaCO₃ supplement (the control diet), in a three-arm cross-over design. The primary outcome of the study was faecal fat excretion (FFE). We also measured post-intervention faecal and urinary calcium, as well as fasting glucose and lipid profiles and anthropometry pre and post-intervention. The mean daily FFE rates (g/day) here ranged from 4.48 to 5.72 g/day. The lowest excretion (4.48 g/day) was seen during the RCC + Supp period, while FFE was highest following the high Ca cheese diet, although not significantly so (P = 0.066). Despite the lower total fat excretion rate, faecal fat % was highest in the RCC + Supp diet, indicating that the lower g/day excretion was due to the lower dry matter excretion overall during this dietary period. It is unclear why total faecal excretion would be affected by the supplement, but other studies have observed similar variations with diets. Soerensen and colleagues (2014), in a cross-over trial with milk and cheese-based diets (which were matched for energy, fat content and fibre in a similar manner to this study), observed that the higher dairy calcium diets resulted in higher dry matter excretion. Lower LDL-c post-intervention in milk and cheese diets vs. the control were observed, indicating an LDL-lowering effect from the dairy calcium. In this study, the LDL-cholesterol concentration was significantly lower post-intervention, following the HCC diet compared to the RCC and to the RCC + Supp diets (Table 6). There were no other significant differences observed in the fasting measures of blood lipids or glucose. The lower LDL-c concentration observed following the HCC diet supports a matrix effect of fat and dairy calcium together in the cheese, since the calcium contents were matched in the RCC + Supp group; thus, the effect appears due to more than simply Ca content alone. The slightly higher FFE rate (albeit not significant) following HC cheese suggests that dairy calcium within the cheese matrix is a driver of this effect. A 2016 in-vitro digestion study of enhanced calcium cheddar cheese [20] demonstrated that additional calcium in the matrix resulted in a harder cheese, and during in vitro digestion, matrix disintegration was found to be slower than the lower calcium cheese, while the rate of lipolysis progressed more quickly in the enhanced cheese. This suggests that the calcium content has a significant impact on the overall physical properties of the matrix and may affect accessibility to nutrients within. Further, the cheeses used here were subject to the same fermentation, yet differences were noted between the diets. This suggests that the effects observed cannot be due solely to the fermentation process either. It should be noted that the amount of cheese used in this study was high (240 g per day) and is greater than the recommended portions for most cheeses, which tend to be in the region of 30 g. The large portions were for the experimental purposes described and should not be considered as a recommended daily intake.

The faecal fat excretion rates observed here (4.48–5.72 g/day) were slightly lower than those reported in Bendsen et al. (2008), (5.4 g and 11.5 g/day during two 7-day dietary periods of high and low Ca intake), while Buchowski et al. [21], in a 12-week weight loss intervention, reported levels of ranging from 3.8 to 5.9 g/day, which are more aligned with those reported here. Several aspects may have contributed to the higher FFE rates observed by Bendsen and colleagues [14]. In their study, FFE was examined in n = 11 overweight participants aged 18–50. Mean daily FE was 5.4 g /day during the low Ca diet (52 mg per MJ of dietary energy) and 11.5 g during the high Ca diet (205 mg per MJ), in a crossover study, and the dietary Ca was provided using low-fat dairy foods. Here, the HCC and RCC diets provided 218 mg per MJ and 160 mg per MJ, respectively (data not shown), meaning that the difference in Ca between the high and low periods was less here than those in the study by Bendsen and colleagues, and the dairy foods here (cheese) contained approx. 34% fat per 100 g of product, so would not be considered ‘low-fat’ dairy. Further, the diets here all consisted of 240 g daily cheese intake, which contributed significant protein (~ 60 g per day, to a mean dietary protein of between 143.6 and 147.9 g/day, or 19.03%—19.44 expressed as % energy from protein). Previous work suggests that increased protein intake can increase intestinal calcium absorption [22, 23], potentially reducing the available calcium to bind to the intestinal fat [19]. The relatively high protein content of the diets here (ranging from 19.03 to 19.44%E on average) may have attenuated differences in FFE between the high and low Ca arms. Jacobsen et al. [24] observed FFE rates of 6 g per day during a low Ca diet (500 mg) of 1 week duration, and at a low protein energy (15%E from protein), which increased to 14.2 g/day during the high Ca intervention (1800 mg Ca), but when the %E from protein was increased to 23%, the FFE was unchanged vs the low Ca, at 5.9 g/day. The protein %E in the present study (ranging from 19.03 to 19.44%E on average) falls between the two values reported by Jacobsen’s study of 15%E and 23%.

However, it should be noted that it is unlikely that the protein content here reduced the availability of Ca for soap formation, since the fat excretion rates observed here, and relatively high Ca excretion, suggest that Ca was still available, and thus does not support the formation of unabsorbed soaps as being solely responsible for the LDL–c lowering, suggested by other studies. In the initial power calculations to estimate the difference between the high and low Ca diets, we had assumed a linear relationship between Ca and fat excretion but had not considered potential effects of protein which could impact that linearity. It should also be noted that we observed lower post-intervention LDL-c in the HCC diet vs the RCC and the RCC + Supp diets, despite a lack of increase in FFE following the HCC diet, which would also indicate this was not solely due to soaps excretion.

In both studies by Bendsen et al. [14] and Buchowski et al. [21], the participants were classified as overweight/obese, whereas here, the participants in the present study were of normal BMI (18–25 kg/m⁻2). While there was 30% energy restriction in study by Buchowski et al., the food in the Bendsen study was matched to each individual, based on their basal metabolic rate (BMR) and their physical activity level (PAL). Here, they were also matched to needs, but on an averaged basis and to a lower BMI, so energy intakes were more similar to those in the study by Buchowski et al., which may explain why the excretion rates here are also similar. In addition to potential effects of protein, Kjølbæk and colleagues [25] reported that fibre may have a confounding effect on the interaction between Ca and faecal fat However, in that study, the range of fibre content was much greater: 20.2 g in Quartile 1 and 32.5 g in Quartile 3. Here, the fibre content did not differ across the diets, and was generally close to the low end of that reported by Kjølbæk et al., ranging from 23.8 to 26.8 g/day. For this reason, we do not think that fibre was a confounder in this study, but this is an important point to highlight.

Boon et al., [18], observed that 7-day intake of 1200 mg calcium, from a non-dairy source, [supplemental CaCO₃] in a mixed-sex group of n = 10 healthy normal weight subjects, led to decreased serum triglycerides. The same study did not observe an impact on triglycerides of either of 400, 1200 and 2500 mg of dairy calcium, suggesting that the CaCO3 form had a greater impact on triglycerides than the same level of Ca from dairy foods. Serum triglycerides in this study were all slightly lower post-intervention compared to pre-intervention values, but there was no difference between the diets. The CaCO₃ used here provided approx. 820 mg of non-dairy Ca, but this was consumed in addition to dairy Ca, provided in the RC cheese, i.e. the calcium during the RCC + Supp period was from a mixed source, which could account for the lack of differences observed between the diets for triglycerides vs. other studies.

No differences were observed here in the individual fatty acids, other than linoleic acid (Table S2), but after Bonferroni adjustment of the P-value cutoff (0.0013) this was not a significant result. This was not entirely unexpected, since the fat consumed was derived, for a large extent, from dairy fat, and so would have been expected to contain similar levels of fatty acids, and the diets were standardized. Although studies have suggested that SFA may be more affected by calcium in the diet [26, 27], a more recent study by Bendsen et al. [14], using dairy foods, suggested that MUFA were more affected by the differences in Ca intake than SFA. We did not observe evidence for differences in the fatty acid excretion here. Nonetheless, this information adds to the literature in this area, since few studies report the faecal fatty acid breakdown following dairy consumption.

A number of strengths and limitations to this study should be noted. Firstly, the higher than expected attrition rate was a major limitation and unfortunately, despite the best efforts of the researchers, it was not possible to extend the study, due to the cheeses being specifically made for this study and aged for 8 months prior to the interventions. Additional limitations include the lack of a habitual calcium intake measure, and that cheeses were not measured for their P content. A measure of habitual Ca would have allowed us to identify any low and high habitual consumers, who potentially might have responded differently in their lipid concentrations. Regarding the P content of the cheeses, calcium phosphate has been shown to reduce LDL-c concentration [16], and cheddar cheeses are high in both Ca and P, which may have had an additional impact on LDL-c concentration over and above Ca alone. A further limitation is that we did not measure physical activity levels during this study, and that individual energy estimates were not calculated. Despite these limitations, a particular strength was the rigorous nature of the data collection, and the controlled dietary intake from those who remained in the study. This followed best practice from previous research, and there was considerable contact with participants to encourage compliance, using an automated messaging system. The specifically produced cheeses, allowing only the calcium to be adjusted within the cheese matrix, while controlling other variables (e.g. starter culture, ripening conditions), and the supply of the drinking water, also contributed to a well-controlled dietary intake in each arm.

This study demonstrates that diets matched for fat content from cheese, but which differ in the amounts of calcium delivered within the matrix, result in differences in fasting LDL-c values, in a group of healthy males. Increasing the level of Ca within the cheese matrix resulted in significantly lower LCL-c vs. the reduced Ca cheese, and compared to the same RC cheese with matched Ca, delivered as a supplement. The mechanistic reason underlying this is unclear. Our results do not support the increased faecal fat hypothesis, but we acknowledge that the high attrition rate here limits the ability to draw strong conclusions on the mechanism. The increased faecal rates with calcium supplementation warrant further study. While the amounts of cheese eaten here (240 g/day) are greater than would be consumed more generally, these findings have implications for dietary guidelines, and provide further evidence that food matrix effects are an important consideration when studying the effects of nutrient and food intake on health outcomes.