The impact of animal products on human health: A 2020 vision of the evidence

Importance of animal origin foods and trends of consumption

Given their high nutrient density, animal-origin foods have been staples of the human diet along our evolutionary history. Indeed, evidence of meat and bone marrow consumption in hominins can be traced back to around 2.5-2.8 million years ago (De Heinzelin et al., 1999, Thompson et al., 2019). On the other hand, milk from ruminants was introduced into our diets more recently, at least 8500 years ago (Curry, 2013). Key to early development of mammals, milk constitutes a major source of energy, high-quality protein, and vitamins and minerals such as vitamin D, calcium, and potassium (Smilowitz et al., 2005; Gaucheron, 2005). These nutrients are of particular importance in populations of children experiencing undernutrition, for which this can translate into long-term effects including reduced cognitive and physical development (Black et al., 2013). Data form the WHO shows that stunting (low height for age from chronic under nourishment), affected around 21.3% children under age five in 2019 (Data World Bank, 2020). Importantly, stunting begins in utero and is greatest for the first 1000 days of life, making the adequate nutrition of the mother a key factor (Victora et al., 2010, de Onis and Branca, 2016).

Indeed, beyond its key role in child nutrition, milk and milk-derived products are also major sources of nutrients for adults. The ability of humans to carry on drinking milk through adulthood has developed gradually over the past eight millennia, in association with agriculture and dairying (Curry., 2013), and it is due to the persistence of the enzyme lactase past early childhood. The rise of distinct genetic mutations for lactase persistence developed in at least four different geographic regions on the planet (i.e., lactase hotspots), and, arguably, provided a major selective advantage (Bersaglieri, et al., 2004). Today, one-third of humans produce lactase during adulthood, with prevalence being greater in places like northern Europe, where over 90% of people can drink milk, but also in West Africa, the middle east, and south Asia (Curry, 2013; Liebert et al., 2017).

As the human population continues to grow, demand for animal-origin food follows, and it is estimated that by 2050 the greatest food security challenge will not be the provision of adequate calories, but rather the access to nutrient-dense foods, such as meat and dairy (Nelson et al., 2018). Despite the arguably obvious benefits that animal-origin products can provide to human populations as a source of key nutrients, the image of milk and meat has blemished and their place as components of a healthy diet has been put into question. Indeed, over the past five decades, changes in consumer perception, dietary guidelines, public health messages, and policy, have all resulted in a shift in the patterns of consumption of meat and dairy products, particularly those of higher fat content.

A generalized “fear of fats” has spread in recent decades and shaped consumer’s choices, who nowadays tend to look for reduced-fat foods as healthier alternatives to full-fat options, in hopes of reducing intake of fat and ‘calories’, and the risk of heart problems and obesity. Simultaneously, milk avoidance has become more prevalent, with a linear decrease in Canada, starting at the end of the 1970’s (Figure 1; Statistics Canada), and new beverages of plant origin with low contents of saturated fats have become available, replacing cow’s milk in the diet (e.g., the so –called soy and almond milks). Furthermore, the most recent nutrition guidelines in Canada suggest, to limit our intake of foods containing saturated fats, which is based on the interest in reducing LDLcholesterol (Health Canada, 2019). In contrast to current perceptions and official dietary advice, an important body of literature has emerged over the past decade that challenges the contemporary views associating saturated animal fat consumption to human disease. Such studies suggest that dairy in general, as well as full-fat dairy, may decrease the risk for CVD. Furthermore, recent studies suggest that full fat dairy may actually protect from obesity and associated chronic diseases.

Figure 1. Consumption trends of fluid milk in Canada from 1960 to 2018 Source: Statistics Canada

A Note on the Assessment of Evidence: An example from the association between meat and cancer

In 2015, the world health organization (WHO) released a report where a meta-analysis of 10 cohort studies yielded a positive relationship between red and processed meat and colorectal cancer (Bouvard et al., 2015). The WHO initially considered 800 observational studies regarding the association between cancer and meat. However, their findings are based on 56 studies looking at colorectal cancer. Their assessment estimated a 17% increased risk per 100 g/d of red meat, and a 18% increase per 50 g/d of processed meat. This resulted in the classification of red meat as ‘’probably carcinogenic’’ and processed red meat as ‘’carcinogenic’’. Importantly, these conclusions are primarily based on observational studies which are limited in their ability to establish causal inferences and are at high risk of confounding.

In contrast, Jonhston et al. (2019) performed 4 parallel systematic reviews studying the impact of red and processed meat on cardiometabolic and cancer outcomes suing both randomized trials and observational studies. A panel of 14 members from 7 countries was formed to provide a recommendation based on the quality of the evidence. The panel recommended that adults continue the consumption of both types of meat. Importantly the recommendations from this committee were based on strict methodology, including the Nutritional Recommendation (NutriRECS) guideline development process and GRADE (Grading of Recommendations, Assessment, Development and Evaluation) methods (Johnston et al., 2019).

To explain this discrepancy, an analysis by Fogelholm et al. (2015) shows how epidemiological research on meat and chronic disease can be riddled with confounding factors. For instance, there is a strong association between meat consumption and a ‘lower-quality’ diet, smoking, and lack of physical activity in both men and women, which complicates interpretation of observational studies (Fogelholm et al., 2015). Furthermore, evidence for claims related to food and human health come from a range of studies with varying degrees of causal strength, going from observational studies to controlled randomized clinical trials. Young and Karr (2011) give some perspective on the frequency at which observational claims fail to replicate, at an alarming rate of 80%. Further, in about 10% of the cases, when the claims from observational studies were tested in clinical trials they moved significantly in the opposite direction (Young and Karr, 2011). These authors suggested that ‘‘any claim coming from an observational study is most likely to be wrong – wrong in the sense that it will not replicate if tested rigorously’’. Therefore, assessing the observational evidence that relates dietary intakes to common disease outcomes is, at the very least, problematic (Prentice, 2014). For instance, the validity of observational studies can be compromised by the necessary reliance placed on self-reported food intakes and can further be complicated by the effects of confounders, such as lifestyle factors that may impact the association. This highlights some of the problems with claims originated from ecological studies, such as the diet-heart hypothesis, which now seems to lose validity in the face of stronger pieces of evidence, such as RCTs and prospective cohort studies, which contradict its original presuppositions (e.g., Micha and Mozaffarian, 2008; Ramsden et al., 2016). While bearing these limitations in mind, the interpretation of observational studies relating dairy intake and health outcomes can be in many instances a useful starting point for investigation and validation. Regardless, causation may only be derived from properly controlled experimentation.

The Diet-heart Hypothesis - Background and Classic Studies

The now prevailing concept that saturated fats of animal origin are detrimental to human health can be traced to some suggestive pieces of evidence originated in the 20th century. First, the ground-breaking studies of Ignatowski in 1908 and Anichkov in 1913 demonstrated the ability of animal fats and, specifically, cholesterol, to cause atherosclerotic lesions, raise plasma cholesterol, and cause death in rabbit models of atherosclerosis (Konstantinov et al., 2006). Second, during the 1950’s, Ancel Keys produced epidemiological data that seemed to identify dietary fat as a major cause of heart disease. In two commonly known studies, Keys and collaborators showed some seemingly strong associations between national death rates for middle-aged men from arteriosclerotic and degenerative heart disease and the proportion of fat-calories available in their national diets (Keys, 1953; Keys et al., 1966). His data relating availability of energy from fat and cardiovascular-related death led him to conclude that “dietary fat somehow is associated with cardiac disease mortality, at least in middle age”. These studies propelled the extensively known diet-heart hypothesis that related dietary factors to the incidence of cardiovascular disorders. The validity of this hypothesis was quickly challenged by Yerushalmy and Hilleboe (1957), and its acceptance has remained far from unanimous ever since it was first proposed. In their methodological note, the authors attempted to evaluate whether the proposed hypothesis could actually reflect “known or ascertainable facts” which may allow for the generalization of this premise. The authors pointed at several limitations, including 1) small sample size (i.e., only 6 countries were used in Key’s original study); 2) the possible effects of unaccounted confounders (e.g., underreporting in countries with lower economic status who cannot afford meat and dairy); 3) the fact that no actual data on fat consumption was used (i.e., availability was used); 4) the lack of specificity of the relationship (i.e., protein consumption also related to death); 5) variation in deaths from cardiac disease are largely variable across countries at any given fat availability category. Some interesting points emerging from their methodological analysis include: first, the strength of the association is greatly reduced when more countries are included (n= 22); second, the relationship is not specific to fat; third, “almost no association” was found when correlating fat or protein with all causes of death. This last point merits attention because CVD is the leading cause of death in industrial countries, and consequently, the relationship in Keys’ studies would be expected to also hold for all-cause mortality. More recently, others (Willet, 2012) have pointed out that the countries chosen by Keys to represent low fat intake and low incidence of CVD were in fact less industrialized and showed differences in smoking habits, physical activity and obesity, thus complicating the generalization of the diet-heart hypothesis.

Dietary Lipids and CVD Risk

The importance of cholesterol in the development of CVD in humans was first suggested by data extracted from the Framingham study, which enrolled 5127 men and women aged 30–59 years in Massachusetts, starting between 1948 and 1950. Following six years of longitudinal evaluation, cholesterol was identified as one of three risk factors for CVD (Kannel et al., 1961)1. Several other studies between the 1950’s and 1980’s found positive associations between serum (total) cholesterol and risk of CVD. With the advent of techniques to identify and measure circulating lipoproteins, it was further shown that cholesterol contained in very low-density lipoproteins (VLDL-C), as well as low-density lipoproteins (LDL-C; i.e., ‘bad cholesterol’), correlates positively with CVD risk, while that found in high density lipoproteins (HDL-C; i.e., ‘good cholesterol’), correlated negatively (see review by Parodi, 2009). In fact, LDL-C became the lead marker for atherogenicity and CVD risk, and thus, the main target and factor guiding CVD treatments in the last few decades (Stone et al., 2014). However, the role of LDL-C recently has been put into question as it is considered a very poor predictor of CVD (Sachdeva et al., 2009; Ravnskov et al., 2018). On the other hand, HDL-C (an indicator of cholesterol efflux) levels are used as a marker of reduced CVD risk (Rohatgi et al., 2014; Monette et al., 2016), particularly when used as a ratio of total cholesterol to HDL-C (Castelli, 1988). In addition, increased triglyceride to HDL-C ratio is a more powerful predictor of coronary heart disease (Luz et al., 2008).

While much of the focus has been historically placed on cholesterol, other factors seem to be important to predict CVD risk. Some of these include obesity, serum triglycerides, inactivity, hypertension, cigarette smoking and diabetes. Results from the Framingham cohort illustrate the importance of these cofactors, as, for example, accounting for glucose intolerance, high systolic blood pressure, smoking, and left ventricular hypertrophy increased CVD risk to 60.2, compared to only 3.9 when cholesterol alone was used (Kannel et al., 1979). Furthermore hyperinsulinemia has been identified as the most important predictor of CVD in some studies, (Després et al., 1996, García et al., 2011), which may explain the association between CVD and previously identified risk factors such as obesity.

Saturated Fats and Blood Lipids

The ability of saturated fatty acids (SFA) to raise blood LDL-C is consistent across the literature, and it is well documented (Micha and Mozaffarian, 2008). Interest in this relationship stems from hypothesis that LDL particles may increase cholesterol accumulation in arterial walls, facilitating the formation of atheromatous plaque, and therefore increasing CVD risk (Kruth et al., 2001). Consequently, efforts have been made to determine the atherogenic potential of individual SFA as a marker of CVD. For example, Ulbricht and Southgate (1991) proposed the atherogenic index, which is calculated by dividing the sum the SFA lauric (12:0), myristic (14:0) and palmitic (16:0), by the sum of omega 3 and 6, 18:1c9, and other monounsaturated fatty acids. Based on Hegsted’s work (1965), each factor is multiplied by an empirical constant according to its capacity to raise or decrease cholesterol, using a value of 1 for all fatty acids and a value of 4 for 14:0. Because these presumably atherogenic fatty acids (12:0, 14:0, 16:0) represent 30-40% of cow’s milk triglycerides (Jensen, 2002; O'Donnell-Megaro et al., 2011), some have concluded that dairy fat is a potential cholesterol-raising food, and, consequently, consumers have reduced the consumption of full-fat dairy (Wang and Li, 2008). Simultaneously, official advice has focused on reducing the consumption of fat and saturated fat specifically. The dietary fat guidelines introduced in 1977 (US) and 1983 (UK) (Harcombe et al., 2015) recommend to 1) reduce overall fat consumption to 30% of total energy intake and 2) reduce saturated fat consumption to 10% of total energy intake. It is important to note that, with these guidelines, the ability of SFA to raise blood HDL-C, and to reduce CVD risk, is implicitly ignored (Parodi, 2009). In fact, a meta-analysis of 60 controlled trials showed that SFA have no effect on the ratio of total cholesterol to HDL-C (lower is better) when SFA replace dietary carbohydrates (Mensink, et al., 2003). Furthermore, the allegedly “atherogenic” lauric acid reduces the ratio, mostly by increasing HDL-C. These observations challenge the notion that SFA are indeed atherogenic and that SFA sources like milk fat may have health-adverse effects.

Furthermore, low-fat diets increased circulating triglycerides and the ratio of total cholesterol to HDL-C (Total:HDL-C; i.e., the atherogenic ratio), both proxies for increased CVD risk. In this way, elevations observed by reducing dairy fat intake cast some concerns about the efficacy of low-fat diets to reduce CVD risk. Arguably, increased CVD risk may actually result from following this type of low-fat approach. Although factors like adiposity and insulin resistance of subjects may have influenced the responses to low-fat diets in this study, it is evident that the effects of such diets on the commonly used biomarkers (e.g., LDL-C, HDL-C, Total:HDL-C, and triglycerides) were certainly not what many may expect in terms of alleviation of CVD risk. Moreover, the validity of CVD biomarkers like LDL-C has dwindled progressively, and the role of lipoproteins on CVD has shifted focus into their size (i.e., small and dense LDL profile is worse), number (i.e. Apolipoprotein B as a marker of atherogenic particles), and oxidation propensity, rather than their cholesterol content (Lamarche et al., 1997; Krauss, 2005; Parodi, 2009). An example comes from the guidelines of the American Heart Association, who, based on a recent review of available data, reported being unable to find evidence to support continued use of specific LDL–C and/or non-HDL–C treatment targets. In this way, the “bad cholesterol” is no longer the main factor guiding treatment (Stone et al., 2014).

Even leaving CVD lipid biomarkers aside, under the diet-heart hypothesis, the effects of SFA on actual clinical outcomes should reflect harmful consequences on cardiovascular endpoints.

Contrary to this expectation, a meta-analysis of prospective cohort studies that followed 347,747 subjects for 5 to 23 years found no association between saturated fat intake and CVD, both fatal and non-fatal (Siri-Tarino et al., 2010). Similarly, in a systematic review and meta-analysis of prospective cohort studies and RCTs (n = 78) with 649,812 participants, Chowdhury et al. (2014) reported no increase of relative risk of coronary outcomes associated with dietary or circulating SFA. Moreover, the authors reported an inverse association between circulating margaric acid (17:0, a marker of dairy fat intake) and coronary disease. Taken together, available evidence from prospective epidemiologic studies and RCT does not support guidelines encouraging reduced saturated fat consumption, particularly those from dairy. Whether this evidence can be considered sufficient to totally vindicate SFA and dairy, is still a matter for discussion; however, currently available data suggest the heavy focus on saturated fats may be not only unnecessary, but perhaps also detrimental. This is particularly true when considering that dairy fats may have been replaced with industrial trans fats of plant origin (e.g. margarines with high trans fat content), as well as refined sugars (i.e., fructose). Some trends exemplifying these dietary substitutions can be seen in Figure 1. Full fat milk has been partially substituted with lower-fat versions, while soft drinks consumption and fructose has risen in linearly. Importantly, there are cogent reasons to believe that the simultaneous reduction in consumption of some dairy products and the increase industrial trans fats and sugars could be detrimental. For example, non-ruminant (i.e., industrial) trans fats are nowadays recognized as harmful (Micha and Mozaffarian, 2008) and they relate strongly to heart disease and all-cause mortality (Oomen et al., 2001; de Souza et al., 2015). Similarly, a growing body of evidence indicates that most US adults currently consume excess added sugar (a source of fructose), and this is significantly associated with obesity, metabolic syndrome, and CVD mortality (Johnson et al., 2009; Lustig et al., 2010; Yang et al., 2014). In fact, the American Heart Association has recommended the reduction of dietary sugar intake by more than half (Johnson et al., 2009).

Saturated Fats and CVD Risk and CVD Mortality

As discussed previously, the relationship between SFA consumption and CVD is not straightforward, and it was historically derived from the diet-heart hypothesis, with two important premises: 1) SFA can influence circulating cholesterol (i.e., increase LDL-C), and 2) cholesterol is a risk factor for CVD. The resulting assumption was, therefore, that SFA consumption can cause CVD. Given the disconnect between SFA consumption and the anticipated clinical CVD outcomes (e.g., Siri Tarino et al., 2010; Chowdhury et al., 2014), the strength of the diet-heart hypothesis has been questioned. Considering that dairy products may have protective effects against obesity, T2D, and metabolic syndrome, it is important to elucidate whether this may also be true for CVD, which remains a major cause of death in the United States (Mozaffarian et al., 2015). When looking at the effects of dairy consumption on CVD risk factors, the most salient finding is that, contrary to expectations, reducing SFA intake from dairy increases CVD risk, as determined by commonly used markers (Lefevre et al., 2005). This sobering observation seems to receive further support from other studies that report significant associations between milk-derived fatty acids and a more favorable LDL particle size distribution (i.e., reduction in small dense LDL particles; Sjogren et al., 2004). Furthermore, some SFA found in milk fat, such as lauric acid, are actually associated with a reduction of CVD risk (Micha and Mozaffarian., 2010). Importantly, these pieces of evidence align with the solid, general observation, that SFA are neutral or even beneficial in terms of CVD risk.

In contrast, a recent meta-analysis by the presidential advisory of the AHA concluded that saturated fatty acids (SFA) should be replaced with unsaturated fats, in particular with PUFA (Sacks et al., 2017). Using data from randomized controlled trials, Sacks et al., (2017) showed that lowering intake of SFA by replacement with PUFA results in 29% lower CHD. This is in contrast to the meta-analysis of Hamley (2017), who reported replacing saturated fat with PUFA had no effect on major CHD events or CHD mortality. A beneficial effect of PUFA was only observed when inadequately controlled trials were included. Those trials were deemed inadequately controlled when important factors, other than the SFA replacement by PUFA, were different between groups (Hamley, 2017). In support of this, in an extensive review of the meta-analyses of observational and controlled studies looking at the replacement of SFA with PUFA, Heileson (2019) points out that 3 of the 4 core clinical trials used in the analysis of Sacks et al., (2017) contained design and methodological flaws, which biased their conclusions. Briefly, the Finnish Mental hospital study [FMHS; (Turpeinen et al., 1979)], the Oslo Diet-Heart Study [ODHS; (Leren, 1970)], The Los Angeles Veterans Administration Trial [LA Vets; (Dayton and Pearce, 1969)], The British Medical Research Council study [MRC; (MRC, 1968)], were included in the AHA study by Sachs et al. (2017) despite several flaws including differences in smoking habits between groups, confounding dietary factors, lack of individual randomization and low adherence across 3 of them (ODHS, LA Vets, and FMHS). In view of the results of all meta-analyses included in his review, Heileson (2019) concludes that observational and properly controlled clinical trials show lack of association of SFA and heart diseases, suggesting the current stance of the AHA may need to be revaluated.

The Effects of Dairy on CVD

Focusing on evidence from prospective data, Elwood et al., (2008) conducted a meta-analysis of 15 prospective cohort studies reporting the association between milk and dairy consumption and the incidence of vascular diseases in the UK. The relative risk (RR) of stroke and/or heart disease was significantly reduced in subjects with high milk or dairy consumption (RR =0.84 and 0.79, respectively), compared with the risk in those with low consumption. These findings highlight once more, the disconnect between the hypothesized effects of SFA-containing foods like dairy, and the actual clinical outcomes of interest. Similarly, a systematic review of the available literature indicates that most studies do not support the expected effects of dairy fat on CVD, and those discrepancies may be associated to country-specific effects (Kratz et al., 2012). Specifically, the Nurses’ Health study (from the US) found a consistent positive association between dairy fat intake and CVD, while 11 other studies across Europe, Costa Rica, and Australia, showed either no association or an inverse relationship between CVD and dairy fat intake. Only one of these 11 studies reported a discrepancy, as it found an inverse association in men, but a positive one in women (Kratz et al., 2012). The authors suggested that residual confounding from lifestyle factors associated with dairy intake, as well as differences in food sources of dairy fat, may help explain the discrepancy between US and non-US data. Relevant to this point, the recently published results from the Prospective Urban Rural Epidemiology (PURE) study evaluated the effects of dairy consumption on death and major CVD events across 21 countries and 5 continents in an 9-year follow-up (Dehghan et al., 2018). Dietary intakes of dairy products for 136,384 individuals were recorded using country-specific validated food frequency questionnaires. Dairy foods evaluated included milk, yoghurt, and cheese, and these were grouped into whole-fat and low-fat dairy. Dairy intake above 2 servings per day reduced the risk of total mortality, CVD mortality, major CVD, and stroke, relative to no intake. Similarly, whole-fat dairy (> 2 servings per day) was inversely associated with total mortality and major CVD. Interestingly, the CVD response to whole-fat dairy appeared to be dose-responsive, as it increased progressively from <0.5 to 0.5-1, 1-2, and >2 servings per day. Cheese consumption (>1 serving per day) was associated with reduced mortality and major CVD, while the effect of butter was neutral (i.e., no increase in risk). The PURE study suggests that dairy intake, especially whole-fat dairy, might be beneficial for preventing deaths and major cardiovascular diseases. Moreover, there seems to be no disadvantage associated with the consumption of full-fat dairy, compared with the low-fat counterparts. The authors conclude that consumption of dairy products should not be discouraged and perhaps should even be encouraged, particularly in low-income and middle-income countries where dairy consumption is low.

The Effects of Dairy on other health outcomes

Obesity and Type 2 Diabetes

One arguably important reason for the current trends of dairy fat avoidance (Figure 1) is related to the interest in reducing excess energy intake. The common presumption is that dairy fat can be stored as body fat and thus contribute to weight gain, obesity, and cardiometabolic risk. This has driven dietary guidelines to recommend the consumption of low-fat dairy (Jensen et al., 2014). In contrast to guidelines and prevailing public sentiment, available evidence indicates dairy fat consumption is not related with the risk of weight gain. The comprehensive review of Kratz et al. (2013), which used a combination of observational and controlled studies, indicates that dairy fat consumption, both recorded or assessed via odd-chain fatty acid content in blood (e.g., 15:0 and 17:0), was inversely related with obesity risk. Similar findings were reported in a cross-sectional evaluation of full-fat milk consumption in three-year-old children (Beck et al., 2017). The multivariate analysis included potential demographic and nutritional confounders. The authors reported reduced odds for severe obesity in association with higher milk fat consumption, suggesting a protective effect of dairy fat against obesity in three-year-olds. Similarly, in a prospective cohort study of 18,438 healthy middle-aged women followed during 11 years and belonging to the Women’s Health Study, greater consumption of total dairy products reduced the risk of becoming overweight or obese. Furthermore, the lowest risk was observed at the highest quintile of high-fat dairy product intake (Rautiainen et al., 2016). Finally, in a meta-analysis of 29 RCTs, Chen et al. (2012) reported dairy consumption does not increase body weight gain or body fat gain. Moreover, dairy consumption results in modest beneficial effects on weight loss in short term and energy-restricted RCTs.

Americans today are eating 25-28% more calories per day, which amounts to an extra 425-800 kcal/d relative to 1961, and is explained mostly by an increase in intake of cereal grains and vegetable oils (Guyenet and Schwartz, 2012). Poortvliet et al. (2007) reported an increase in voluntary energy intake in subjects fed a higher carbohydrate (Fetuccine-based meal) relative to a higher protein (Chicken-based) meal. This was observed despite meals being isoenergetic and was explained by reduced hunger throughout the day when the chicken-based meal was offered (Poortvliet et al., 2007). These observations illustrate an important effect of diet on dysregulation of intake (Guyenet and Schwartz, 2012), which recently has been shown to also be impacted by the consumption of processed foods (Hall et al., 2019). In a randomized controlled trial, Hall et al., (2019) compared two iso energetic diets varying in degree of processing. Relative to a diet rich n dairy, meat, fruits and vegetables, an ultra-processed diet resulted in higher energy intake, and increased body weight (+1.8 kg) and fat mass (0.7kg) after 14 d of intervention (Hall et al., 2019).

Type 2 diabetes (T2D) is rapidly rising worldwide, paralleling the epidemic increase in obesity. Because of its high content of calcium, magnesium, vitamin D, and whey proteins, which could reduce insulin resistance, dairy products could be hypothesized to protect against T2D (Rice et al., 2011). The meta-analysis of Aune et al. (2013) evaluated the association between intake of dairy products and the risk of T2D from prospective cohort and nested case-control studies (n=17). Nonlinear, inverse associations were found between the risk of T2D and intakes of dairy products, low-fat dairy, yogurt, and cheese, the latter being the highest in fat content. The risk responses to dairy intake were dose-dependent, and flattened at higher intakes. Interestingly, high fat dairy did not alter the T2D risk in this study, although a meta-analysis focused specifically on butter consumption (Pimpin et al., 2016; 11 country-specific cohorts and 201,628 participants) reported butter intake was associated with a reduction of T2D risk. This discrepancy may suggest that the effects of dairy fat may be food-specific (e.g. cheese different from butter), a concept that merits further investigation. Other recent meta-analyses add support to the protective effects of dairy consumption against T2D (Forouhi et al., 2014; de Souza et al., 2015; Yakoob et al., 2016). For example, the prospective associations between circulating fatty acids in phospholipids and T2D were reported in individuals from the EPIC-InterAct case-cohort study (17,928 T2D subjects and 16,835 participants in a random subcohort; Forouhi et al., 2014). By design, this study combines the temporal sequence and power advantages of a larger prospective cohort, with the measurement efficiency of a case-control. Forouhi et al., 2014 reported reduced hazard ratios for incident T2D in association with the odd chain SFA 15:0 and 17:0, both of which are mostly derived from dairy products. Similarly, using two prospective cohorts with 3333 adults aged 30 to 75 years, free of T2D at baseline, and followed during 15 years, Yakoob et al. (2016) found that individuals at the highest quartile of plasma 15:0, 17:0, and t-16:1n-7 had reduced risk of incident diabetes mellitus (-44%, -43%, and -53%, respectively; Figure 2). This last finding is of particular interest given that other studies have showed that circulating trans-palmitoleic acid (t-16:1n-7) is associated with lower insulin resistance, atherogenic dyslipidemia, and incident diabetes (Mozaffarian et al., 2010; de Souza et al., 2015). Importantly, whole-fat dairy consumption is most associated with elevated plasma concentrations of trans-palmitoleic acid (Mozaffarian et al., 2010). Whether the apparently beneficial effects of dairy on T2D risk are mediated by trans-palmitoleic acid or other components of dairy, remains to be experimentally elucidated; regardless, this possibility constitutes an exciting new direction for fatty acid research.

Figure 2. Circulating biomarkers of dairy fat intake and risk of incident diabetes mellitus in two large prospective cohorts using 3,333 adults in a 15-year follow-up (Adapted from Yakoob et al., 2016). Solid-black and dashed-red lines represent hazard ratios (HR) and their 95% confidence intervals, respectively, for plasma A)15:0, B)17:0, and C) t-16:1n-7.

Circulating biomarkers of dairy fat intake and risk of incident diabetes mellitus in two large prospective cohorts using 3,333 adults in a 15-year follow-up (Adapted from Yakoob et al., 2016). Solid-black and dashed-red lines represent hazard ratios (HR) and their 95% confidence intervals, respectively, for plasma A)15:0, B)17:0, and C) t-16:1n-7.

Metabolic Syndrome

Metabolic syndrome (MetS) consists of a cluster of cardiovascular risk factors that include central obesity, hyperglycemia, hypertriglyceridemia, low HDL-C, and hypertension (Alberti et al., 2009). Moreover, MetS is closely associated with CVD risk, T2D, all-cause mortality and cancer (Saely et al., 2007; Wu et al., 2010; Esposito et al., 2012). Indeed, epidemiological evidence suggests that several types of cancer such as liver, gallbladder, and pancreas cancer are related to obesity, which could be mediated by insulin resistance (Calle and Kaaks, 2004).

Despite the recognition of the potential of dairy products to prevent or alleviate CVD, T2D and blood pressure in adults (e.g., 2010 Dietary guidelines for Americans; USDA/USDHHS, 2010), data on the relationship between dairy consumption and MetS is very limited. Chen et al. (2015) evaluated currently available data from cross-sectional/case control studies (n=16) and prospective cohort studies (n=7) in two different meta-analyses. Comparing high vs. low dairy products intake, both meta-analyses showed a reduction in MetS risk with high dairy consumption, an observation that was maintained when the studies were evaluated by stratified subgroups (e.g. geographic region, sex, type of dairy, and follow-up duration). Finally, a dose response analysis of prospective cohorts in the same study showed an inverse relation between MetS risk and dairy consumption (Chen et al., 2015). Interestingly, the reduction in risk became evident when dairy intake was higher than 2 servings per day, and behaved linearly thereafter. This would suggest a minimum amount of dairy may be needed to positively impact MetS risk in a significant manner.

Cancer incidence

The link between dairy and cancer has been the subject of multiple studies in recent years, and it is an active area of discussion. A recent overview analysis of published meta-analyses and systematic reviews aggregated all results looking at the relationship between dairy consumption and cancer incidence (Jeyaram et al., 2018). Their analysis concluded that available studies were mostly low to moderate in terms of quality, which limits the capacity to evaluate this association and provide guidelines for consumption dairy.

Prostate cancer: Plaza et al. (2019) summarized the available reviews and meta-analyses investigating the link between dairy consumption and prostate cancer. Their analysis shows that although some data support a positive association between dairy and prostate cancer, there is a high degree of variability in the quality of published studies, including weak control of confounding factors, rendering this association inconclusive (López-Plaza et al., 2019).

Bladder cancer: A meta-analysis of cohort and case-control studies found high milk intake was associated with decreased risk of bladder cancer (Mao et al., 2011). In contrast, a recent metaanalysis of similar studies by Bermejo et al. (2019) reported that high intake of whole milk increased bladder cancer risk, whereas medium intake reduced risk, although authors warned results may be affected by high heterogeneity of the studies included in the analysis.

Colorectal cancer: Barrubés et al. (2019) showed cheese consumption was inversely related to colorectal cancer in a meta-analysis of 29 observational studies, however other types of dairy such as fluid milk and fermented milk were not associated to colorectal cancer. A randomized controlled trial, published in the Journal of the American Medical Association, showed dairy intake reduces proliferative activity of colonic epithelial cells and restores markers of normal cellular differentiation in patients determined to be at risk of colonic neoplasia (Holt et al., 1998). Such observation could be explained by calcium and/or vitamin D in milk (Holt, 1999) similar to studies in rats where markers of dimethylhydrazine-induced colorectal cancer were reduced by dietary calcium (Pierre et al., 2008). Furthermore, a meta-analysis of 60 observational studies and 26,335 cases of colorectal cancer, showed that dietary calcium and dairy intake were inversely related to cancer risk (Huncharek et al., 2008).

Breast cancer: Several meta-analyses of prospective cohorts and/or case-control studies have looked at the association between breast cancer and dairy intake, showing either no association (Missmer et al., 2002) or an inverse association (Dong et al., 2011, Zang et al., 2015). For instance, in the largest meta-analysis including 22 prospective cohort and 5 case-control studies across Asian and Western populations, Zang et al. (2015) report effects may depend on dairy type, dose, and time, such that high (>600 g/d) and intermediate dairy consumption (400-600 g/d) reduce breast cancer more pronouncedly, whereas a similar effect by fermented dairy products is detected after 10 years of follow up, but only in American women. The mechanism behind these associations is not yet clear, but may be partly mediated by anti-inflammatory effects of dairy. Rashid et al. (2006) reported fermented milk reduces the growth rate of mammary tumors in a murine model, an effect mediated by reduced production of pro-inflammatory cytokines. In line with this observation, a systematic review of RCTs (Ulven et al., 2019) documented a significant reduction of inflammation in both healthy and unhealthy individuals consuming dairy. Furthermore, the anticancer effects may also be related to bioactive components found in cow’s milk (Parodi, 2009), some of which are reviewed in the following section.

Milk components with anticancer properties

The mechanistic modes of action by which milk and dairy products may be protective against cancer, CVD and other health outcomes is likely complex and still requires investigation. It is important to bear in mind that bovine milk contains an outstanding number of bioactive components (Park, 2009), which may interact additively, synergistically, or antagonistically. The heavy focus on single nutrients during the past decades (e.g., saturated fats should be avoided) has proven narrow in scope and of limited ability to predict health outcomes. As proposed by others (Mozaffarian, 2014), food-based guidelines that reduce confusion for consumers and are based on prospective evidence for effects on clinical endpoints are needed.

Bovine milk fat is highly complex, containing up to 400 different types of lipids, and although a high proportion of those are saturated, others, which are considered to be bioactive, are also present in milk fat (Jensen, 2002), and may explain the previously discussed positive health outcomes associated with dairy consumption. Most milk lipids (approx. 98%) are found in the form of triglycerides, with the remainder composed of diglycerides, phospholipids, and cholesterol (Jensen, 2002). The following is a list of some molecules found in milk which have some potential as anticarcinogenic agents:

Butyric acid (4:0): Ruminant milk is the main dietary source of butyric acid as it derives from ruminal fermentation processes. Butyric acid has been shown to inhibit chemically induced mammary tumors in rats (Yanagui et al., 1993; Belobrajdic and McIntosh, 2000). As reviewed by Parodi (2009), butyric acid has anti-inflammatory properties, which may be associated with its potent anticancer effects.

Conjugated linoleic acid (CLA): Bovine milk fat contains over 20 positional and geometric CLA isomers, although cis-9, trans-11 CLA (also known as rumenic acid) is the predominant isomer and accounts for ~75-90% of the total CLA (Lock and Bauman, 2004). Importantly, these isomers are unique to ruminant-origin products since they are produced in the rumen by microbial biohydrogenation (BH) of dietary PUFA. The main pathway of BH yields cis-9, trans-11 CLA and other downstream intermediates such as trans-9 18:1, and may be progressively shifted by dietary factors impacting ruminal bacteria, leading to increased production of other isomers such as trans-10, cis-12 CLA, a potent inhibitor of mammary lipogenesis (Bauman and Griinari, 2001, Jenkins et al., 2008, Rico et al., 2015). Importantly, the most studied CLA isomers in humans or in animal models are cis-9, trans-11 and trans-10, cis-12 CLA, given the anticarcinogenic and anti-inflammatory properties of the former and the anti-lipogenic properties of the latter (Parodi, 2009, Foote et al., 2010, Kennedy et al., 2010).

Sphingolipids: they represent 25 - 35% of milk phospholipids (Jensen, 2002) and show potential to reduce the incidence of inflammation-related chronic diseases, by acting directly on the microbiota, on the action of bacterial lipopolysaccharide (LPS), and on anti-inflammatory receptors (Norris and Blesso, 2017). In addition, sphingolipids such as sphingomyelin have been shown to inhibit certain types of cancer in mice (Vesper et al., 1999). Importantly, the sphingolipid content and profile of milk may vary depending on the cow's physiological status (Jensen, 2002), but the effects of cow diet on their concentrations in milk fat are not known.

Branched chain fatty acids (BCFA):they are bioactive food components that make up about 2% of dairy fat in cows (Kuzdzal-Savoie, 1964, Ran-Ressler et al., 2014). The anti-cancer capacity of 15:0-iso, a BCFA, was similar to that of CLA (Wongtangtintharn et al., 2004). In addition, 15:0-iso inhibited the growth in vitro and in vivo of various cancer cell lines by inducing apoptosis without toxic side effects (Yang et al., 2000). Branched chain FA are produced by rumen microorganisms, for which they constitute an important part of cell membranes. Once synthesized in the rumen, these fatty acids are absorbed in the intestine, delivered to the mammary gland and incorporated into milk fat (Fievez et al., 2012, Baumann et al., 2016). Similar to other milk FA, their concentrations can also be modified by the diet, in particular fiber and PUFA content of diets have been shown to alter the concentrations of these FA in milk (Villeneuve et al., 2013, Saliba et al., 2014, Baumann et al., 2016).

Conclusion

Production of nutrient-dense foods will be a major challenge as global demand from a growing population continues to increase. Re-evaluation of the validity of classic literature and the emerging abundance of new evidence over the past decade, strongly contradict the long-held idea that dietary saturated fats cause adverse effects on health. Moreover, as shown in this review, current evidence indicates that dairy products, including full-fat dairy, may exert protective effects on metabolic health, reducing the incidence of obesity, T2D, MetS, CVD, several types of cancer and all-cause mortality. The mechanisms behind such effects has not yet been completely elucidated, but may be related to the wide range of bioactive compounds, such as butyric acid, conjugated linoleic acid, branched chain fatty acids and sphingolipids found dairy. In light of this, a general call to revise the guidelines on dairy consumption seems strongly justified and necessary, particularly as dairy products may help combat the spread of chronic diseases. Moreover, the historic focus on individual nutrients (e.g., fat, calories) has proven limited in terms of predicting clinical outcomes. In this sense, a whole-food approach to studying the effects of the ensemble of nutrients contained in animal-origin foods on human health outcomes seems warranted. Lastly, policy changes should be guided by a more nuanced interpretation of observational studies, and reflect the higher value of repeatable, randomized controlled studies, as the latter may provide insights on causality and the role of dairy on public health.

Published in the proceedings of the Animal Nutrition Conference of Canada 2020. For information on the event, past and future editions, check out https://animalnutritionconference.ca/.

The impact of animal products on human health: A 2020 vision of the evidence

What is Rendering?

E.S.E. & INTEC