A whole foods plant based diet could prevent Alzheimer’s
Disease by reducing cardiovascular risk factors
Abstract
Alzheimer’s Disease (AD) is a debilitating disease for the patient, the caregiver, and the healthcare system. Because there is no medication to alter the course of the disease, a better preventive approach is required.
The vascular hypothesis of AD explains amyloid fibrillogenesis as a result of a comorbid cluster of cardiovascular diseases, including atherosclerosis, hyperlipidemia, type II diabetes and hypertension. The evidence base of this hypothesis is reviewed using current clinical and epidemiological evidence on associations between AD and these cardiovascular risk factors. It is concluded that there is sufficient evidence that AD is related to these cardiovascular diseases.
A whole foods plant based (WFPB) is originally based on diets of non-industrialized cultures, where prevalence of these cardiovascular diseases is low to nonexistent. In clinical trials, the WFPB diet has shown unmatched results in reducing and reversing cardiovascular diseases and is therefore a promising candidate to prevent AD.
The diet rests on much of the evidence base of existing diets used to lower the risk of AD, such as the MIND, DASH and Mediterranean diet, but applies the evidence more directly by not compromising to match Western eating habits. Combined with the capability to reverse cardiovascular risk factors, strict avoidance of harmful foods, and higher consumption of beneficial foods, the WFPB diet is likely to be a superior alternative for those willing to make more rigorous changes. A clinical trial is warranted to confirm whether the WFPB reduces AD progression or incidence more than existing diets.
Name: Kasper van ‘t Veer Student number: 10002960
Study programme: MSc Brain and Cognitive Sciences Course: Literature Thesis
Institution: University of Amsterdam
Supervisor and assessor: Ondine van de Rest (WUR) Examiner: Paul Lucassen (UvA)
Date: Match 16, 2020 - June 29, 2020
Overview
1. Introduction and problem definition 3
2. Critical discussion: Vascular components of Alzheimer’s disease 5
2.1 Amyloid hypothesis vs vascular hypothesis 5 2.2 Cholesterol and atherosclerotic disease 9
2.3 Diabetes 16
2.4 Hypertension 19
3. Critical discussion: Effect of WFPB diets on vascular risk factors 22
3.1 Introduction 22
3.2 Cholesterol and atherosclerotic disease 27
3.3 Diabetes 39
3.4 Hypertension 43
4. Conclusion 47
5. Critical opinion 51
1. Introduction and problem definition
PrefaceDementia is the single most common and fastest growing cause of death in the Netherlands (Volksgezondheid en Zorg, 2018). As many as 280.000 people in the Netherlands have dementia, 70% of which is Alzheimer’s disease (AD), followed by 16% for vascular dementia (VaD) (Alzheimer Nederland, 2019). In this introduction, the impacts of AD on society are discussed, together with how exploring a whole foods plant based (WFPB) diet may be a promising strategy for its prevention. Then, the vascular hypothesis and amyloid cascade hypothesis of AD are compared, and existing evidence for the contribution of several cardiovascular risk factors to AD is reviewed. Afterwards, we will review the WFPB diet, how it compares to other scientifically studied diets and how this diet may alleviate cardiovascular risk factors. The article finishes with a conclusion about the applicability of WFPB diets for the prevention of AD and a personal critical opinion about dietary prevention strategies. The aim of this review is to evaluate the likelihood of a WFPB diet being a safe and improved option for the prevention of AD.
Alzheimer’s Disease’s damage to patients and their families
The possibility of losing your memory and other cognitive functions to AD is one of the most feared and devastating aspects of aging (Coyle, Price & Delong, 1983). Its distressing impact on society was captivatingly described by the following piece:
“Eventually, Alzheimer’s kills, but not before it takes everything away from you. It steals a person’s memories, judgment and independence. It robs spouses of lifetime companions and children of parents and grandparents” (Alzheimer’s Association, 2011)
This is not only debilitating for the patient themselves, but also for their loved ones. Because AD is such a disabling disease, patients in later stages of the disease often require around the clock care, the responsibility of which is often felt strongly by the patients' children. A particularly heartrending aspect of AD is that before succumbing to the disease, patients may fail to recognize children, spouses or lifelong friends, while they are often spending extraordinary effort caring for them. Caretakers have been found to have a higher incidence of mental health issues, financial problems and have their careers affected (Black et al., 2010). The mental toll on a patient’s relatives is exacerbated by the patient’s often changing personality, compounded by anger, confusion and incoherent speech and behaviour. Any new methods to reduce the incidence of AD could save a great deal of suffering and offer hope to those at risk and those in fear of contracting the disease, as well as their families.
Alzheimer’s Disease’s damage to the healthcare system
The great amount of care required by AD patients is also very problematic for the healthcare system. At 9.1 billion euros, care for dementia currently takes up 10,3% of all Dutch government healthcare expenses or 36,0% of all their mental disorder expenses (Volksgezondheid en Zorg, 2017). This figure is expected to increase explosively in the future as well, due to aging of the population. In less than twenty years, cases for dementia are expected to nearly double to over half a million (Alzheimer Nederland, 2019). A vast increase in patients requiring such a high degree of care is likely to overwhelm the system. Current medications for AD only have small effect sizes and do not clearly alter disease progression (Barnes & Yaffe, 2011), making it unlikely that a cure for AD is going to be available quickly enough. If proven effective, a preventive lifestyle programme is not only much cheaper than constant care and/or medication, but may also prevent people from depending on healthcare in the first place. It may therefore be a better strategy to win this race against time to prevent a financial disaster in healthcare, especially since the side effects would only be positive, as other ailments may be reduced as well, as is discussed in chapter 3.2, 3.3 and 3.4.
The case for a whole foods plant based diet
Alzheimer’s disease has been described as a vascular disorder (Kelleher & Soiza, 2013; de la Torre, 2004), while a whole foods plant based (WFPB) diet, consisting of minimally refined plant-based foods and avoiding animal-based or processed foods, has been reported to reverse cardiovascular disease (Esselstyn et al., 2014; Ornish et al., 1998). Though there have been no human trials directly testing the effects of a WFPB diet on the progression of AD, this may be speculated on in a mechanistic manner by literature analysis. If evidence for both a cardiovascular basis of AD and for the ability of a WFPB diet to reduce cardiovascular risk factors is strong enough, the diet is likely to be effective for the prevention of AD as well. Furthermore, the diet also bears several similarities with existing diets that have been found to be neuroprotective, though a couple of differences also exist, which are later discussed. To determine whether this diet may be more beneficial for the prevention of AD, this thesis offers a comprehensive review on the vascular components of AD in chapter 2, along with a review on the properties of a WFPB and current evidence of its ability to reverse these cardiovascular factors in chapter 3.
2.
Critical
discussion:
Vascular
components
of
Alzheimer’s disease
Back in 1907, Dr. Alois Alzheimer published his first article, describing what later became the discovery of Alzheimer’s Disease. In this original article, Alzheimer described the findings of an autopsy and prior behavioural symptoms of Auguste Deter, a woman in her early fifties with severe dementia. Vascular dementia was already known, but Deter’s young age and autopsy suggested a different pathological process, that was later coined Alzheimer’s Disease. The autopsy revealed characteristic changes in the neurofibrils andsites of deposition of a peculiar substance in the cerebral cortex (Stelzmann, Norman Schnitzlein & Reed Murtagh, 1995) , which are now understood to be tangles of hyperphosphorylated tau protein and β-amyloid plaques, two key anatomical features of AD . Much less attention seems to have initially been caught by two of his other findings described in the same article: rigid radial arteries and arteriosclerotic changes in the larger cerebral vessels. In hindsight, this may have been a clue, as in the last few decades substantial evidence has emerged for a cardiovascular basis of AD pathology. The next sections take a theoretical approach and describe the amyloid and vascular hypotheses of AD, followed by several chapters that review specific cardiovascular components and evidence for their relation to AD. Studies not differentiating between AD and VaD were not included, because the aim is to determine the vascular basis of AD only. VaD by definition has a vascular basis and its presence in data is likely to disproportionately colour the results.
2.1 Amyloid hypothesis vs vascular hypothesis
Amyloid cascade hypothesisThe most widely accepted hypothesis on the pathogenesis of AD is the amyloid cascade hypothesis described by Hardy & Higgins (1992). This hypothesis describes β-amyloid deposits as the causative agent of AD pathology, while neurofibrillary tangles, cell loss, vascular damage, and dementia follow as a direct result. The hypothesis describes how breakdown of amyloid precursor protein generates the polypeptide β-amyloid, the main component of the senile plaques implicated in AD, and further mechanisms through which this polypeptide initiates the pathological cascade of AD.
Criticism
Though the hypothesis has been underpinned with large amounts of novel research, it has been met with increasing criticism in recent years. Critics of the theory note that growing amounts of data have accumulated that are inconsistent with the hypothesis on grounds of biochemistry, animal models, pathology, clinical findings and epidemiology (Herrup, 2015). Drachman (2014) concludes that though there is no doubt amyloid is associated with AD, it is not the main causal factor. His article lists a dozen examples where doubt is casted on direct causality, most notably the lack of amyloid clearing
agents’ ability to improve cognition or arrest decline. As an alternative, Drachman proposes to focus on vascular risk factors and suggests trials aimed at improvement of microvascular and endothelial function as a promising topic for future research. The endothelium is the inner lining of arteries that allows them to expand. Endothelial dysfunction, often as a result of lifestyle, is a key pathological process underlying many cardiovascular diseases and discussed in later sections.
Vascular hypothesis
The idea of a vascular hypothesis of AD was coined by de la Torre & Mussivan (1993). This hypothesis poses that vascular degeneration (as an effect of aging or lifestyle for example) disturbs cerebral blood flow, which causes a cascade of events ultimately resulting in AD pathology. An important effect of this cascade is that cerebral hypoperfusion impairs the delivery of nutrients such as oxygen and glucose to cerebral neurons and thereby increases the risk of neuronal ischemia. Amyloid deposition is actually one of the outcomes of these events, making the vascular hypothesis “hook into” some of the mechanisms from the amyloid hypothesis. The 1993 publication suggests this happens in the following way: neuronal ischemia causes astrocytosis, the resulting reactive glial cells may then express amyloid precursor protein, whose proteolysis generates β-amyloid. The outcome of these interrelated events would be progressive neuronal degeneration and brain tissue death. In later publications (de la Torre, 2004; 2013), the basis for the hypothesis grew towards newly found associations between AD and vascular pathology (the likes of which are discussed in the next sections), and these factors’ neurodegenerative effects through cerebral hypoperfusion. Interaction between both hypotheses
Interestingly, rather than being mutually exclusive, the two theories have also been suggested to interact. Van Norden et al. (2012) proposes a model where both theories consist of separate chronological steps, both ultimately leading to AD pathology, but with many interrelations between the steps of the different theories (Figure 1).
Figure 1. “ Interaction between the amyloid hypothesis and the vascular hypothesis in the etiology of Alzheimer's Disease”. Reprinted from van Norden et al. (2012).
Proposed mechanisms
Kelleher & Soiza (2013) propose additional mechanisms of interaction between vascular dysfunction and the deposition of amyloid. Lack of cerebral blood flow causes impairment of solute transport and a decrease in glucose delivery, which results in expression leading to an increase in amyloid fibrillogenesis. Additionally, they note that high blood pressure, one of the risk factors for AD, is associated with an increase in angiotensin, which is converted by angiotensin-converting enzyme, which is not only associated with AD, but also cleaves β-amyloid. This cleavage may contribute to β-amyloid misfolding implicated in AD. Though they conclude more research is needed on this topic, this theory is consistent with the finding that an amyloid precursor protein mutation that results in reduced β-amyloid is protective of AD in the elderly (Jonsson et al., 2012).
Milionis, Florentin & Giannopoulos (2008) suggest an association between AD and “metabolic syndrome”, a clustering of cardiovascular risk factors of metabolic origin. This includes obesity, type II diabetes, atherogenic dyslipidemia (elevated serum cholesterol, especially LDL cholesterol) and hypertension. They present several pathways through which these risk factors may affect cognitive dysfunction and AD. Obesity often causes hypercortisolemia and hyperleptinemia which are both associated with cognitive dysfunction. Type II diabetes results in increased blood viscosity, which
decreases cerebral perfusion, that in turn causes cognitive dysfunction and AD. Furthermore, the hyperglycemia (elevated blood glucose) resulting from diabetes can cause alterations in cerebral capillaries that increase ischemia risk, which can lead to β-amyloid deposits and AD. Another effect of type II diabetes is high insulin levels, which may promoteβ-amyloid secretion and tau protein phosphorylation, which may play an important role in the metabolism of β-amyloid and tau-protein. Also, hypertension has been associated with silent brain infarctions, brain atrophy, endothelial dysfunction and cerebral hypoperfusion and may exacerbate AD through any of these factors. Interestingly, they conclude lifestyle and diet modifications as some of the recommended approaches to reduce AD risk by managing the aforementioned risk factors.
Though evidence exists for each of the individual steps of all these mechanistic pathways, the pathways as a whole remain merely hypotheses. However, the vascular risk factors at the start of each of these pathways are all interrelated and have been associated with AD in humans, the latter being the topic of the next chapter.
2.2 Cholesterol and atherosclerotic disease
Implications of cholesterol and atherosclerosisAtherosclerotic plaques in the walls of an artery narrow its inner diameter and may also cause arterial stiffness (van Popele et al., 2001), both of which not only cause infarctions (Zeiher, Drexler, Saurbier, & Just, 1993), but also chronically restrict blood flow. As discussed in the previous section, this decreased blood flow may play a role of its own in AD pathology and cognitive dysfunction. In this chapter, we review current evidence for direct associations between AD and atherosclerosis, but also cholesterol. Cholesterol, and especially LDL cholesterol is of interest because it has recently been causally linked to atherosclerosis (Ference et al., 2017; Piepoli et al., 2016). Benjamin & Roberts (2013) even postulate that cholesterol may be the only factor necessary to form plaques and that other risk factors merely mediate the effects of cholesterol. The effects of cholesterol on AD are therefore particularly interesting for this thesis, because of the influence of diet on serum cholesterol, as will be discussed in chapter 3.2.
Postmortem intracranial atherosclerosis Roher et al. (2011) found significantly more extensive atherosclerotic occlusion in the circle of Willis arteries in AD patients (n = 36) than in age-matched non demented controls (n = 61) within a population of deceased Americans. They provided cross section images showcasing the difference in inner diameter between clean and heavily occluded arteries (Figure 2).
Yarchoan et al. (2012) performed a similar study, also on Americans, and found extensive circle of Willis atherosclerosis more often in the AD group (n = 410; 77%) than in controls (n = 59; 47%). They also found atherosclerosis ratings to be associated with neuritic plaque, tau neurofibrillary tangle and cerebral amyloid angiopathy ratings in the entire sample and between groups after controlling for age and sex. Beach et al. (2007) compared circle of Willis atherosclerosis between an AD (n = 215), VaD (n = 30), non-AD dementias (n = 60) and non-demented elderly controls (n = 92) group, also in an American population. They found the most severe atherosclerosis in the AD group, followed by the VaD group, while the non-AD dementias and control groups had similar, lower
amounts. They controlled for age, gender and ApoE-ε4 genotype, and concluded that the relation between intracranial atherosclerosis and AD is not an artifact of diagnostic misclassification or ApoE-ε4 genotype.
The above studies show a clear association between atherosclerosis in the brain and AD. This provides evidence for shared pathological mechanisms between intracranial atherosclerosis and AD, but being a mere association, however, it does not guarantee that preventing or reversing any form of atherosclerosis does the same for AD.
Atherosclerotic cardiovascular disease
In a population study in 284 Dutch dementia patients (207 with AD) by Hofman et al. (1997), carotid atherosclerosis severity was found to be associated with both AD (OR = 1.8) and VaD (OR = 3.2) and interacting with ApoE-ε4 genotype. Presence of the ApoE-ε4 genotype increased the association between atherosclerosis and AD or VaD. In the case of both atherosclerosis and presence of ApoE-ε4, odds ratios were 3.9 for AD and 19.8 for VaD.
In a longitudinal study, Zhu, Wang, Deng & Zhou (2014) followed 423 Chinese patients with mild cognitive impairment (MCI) for 4 years and found that moderate to severe intracranial stenosis increased the odds (p < 0.001) of progressing to AD, and increased the rate of cognitive decline. They also found high blood pressure and diabetes to further increase the odds of progressing to AD, but not high cholesterol.
Li et al. (2011) also followed 837 Chinese MCI patients for 5 years and found high cholesterol, high blood pressure and diabetes to increase the risk of MCI progressing to AD. ApoE-ε4 was associated as well, though the combination of the aforementioned vascular risk factors increased the odds more than ApoE-ε4.
Luchsinger et al. (2005) followed 1138 non-demented Americans for an average of 5.5 years and found only diabetes (HR = 4.8), but not heart disease (HR = 1.2) or hypertension (HR = 1.5) to significantly increase the risk of developing AD on their own. However, in hypertensive patients, heart disease (HR = 2.7) or diabetes (HR = 2.7) did significantly increase the risk.
These studies suggest that AD could to some extent be part of a cardiovascular risk factor cluster that includes atherosclerosis, diabetes and hypertension and may be mediated by ApoE genotype.
Serum cholesterol
Some cholesterol studies specifically researched the relation between serum cholesterol and AD as well. Kivipelto et al. (2001) followed 1449 Finns for an average of 21 years and found high systolic blood pressure (OR = 2.3) or serum cholesterol (OR = 2.1) during midlife (40-64 years) to significantly increase the risk of AD in late life (65+ years). Participants with both risk factors were at an even higher risk (OR = 3.5). In a subsequent study (Kivipelto et al., 2002), they found these risk factors to individually be stronger predictors of AD than ApoE-ε4 genotype.
Helzner et al. (2009) followed a cohort of 4166 Americans aged 65 and over for a mean of 3.5 years. At the end, there were 417 subjects with incident AD. Only 156 subjects
were included in the analysis due to missing lipid data, death, too recent diagnoses or being lost in follow-up. Higher pre diagnosis total or LDL cholesterol and diabetes, but not hypertension were associated with faster cognitive decline. Interestingly, despite the association found with cholesterol, they found no association with heart disease. Notkola et al. (1998) performed a retrospective study using a sample of 444 Finnish men aged 70-89 years and found high serum cholesterol at age 40–59 to be a significant predictor (OR = 3.1) of AD at age 70–89 after controlling for age and ApoE-ε4. However, in those who did develop AD, serum cholesterol dropped before clinical manifestation of AD.
This is consistent with later findings by Solomon et al. (2007), who reexamined the same population used by Kivipelto et al. (2001) and found that while high midlife cholesterol is a risk factor for future AD, decreasing cholesterol during late life is also a risk factor. Their results were controlled for various cardiovascular risk factors, age, APOE-ε4 and midlife cholesterol.
The reasons for this peculiar difference seem unclear, though could be related to decreases in body weight and malnutrition commonly associated with (early) AD (Shatenstein, Kergoat & Nadon, 2001) which might decrease serum lipids.
Cholesterol contradictions
This discrepancy between midlife high cholesterol and late life decreasing cholesterol being associated with AD may explain why there exists contradiction between studies. Li et al. (2005) found no association between current and previous serum cholesterol in a sample of 2356 Americans. However, at enrollment, the participants were on average older than 75 and pre-enrollment cholesterol levels were determined as the mean of measurements dating back 6-8 years before the study, making the full span of measurements fall under late life for the average participant.
Romas, Tang, Berglund & Mayeux (1999) measured HDL cholesterol, LDL cholesterol, and triglycerides in a multiethnic cohort of 987 participants aged 76±6 years from New York City. The only association they found was that people with reduced total cholesterol were at a higher risk for AD, independent of ApoE genotype. Reitz et al. (2004) also found no association between any type of cholesterol and AD in a population of 4316 Americans aged 77±7 years, though 45% of participants dropped out of the study. Both of these studies were cross-sectional and with elderly people, their findings are thus also in line with the idea that cholesterol is not a risk factor when measured during late life.
There are exceptions to this rule though. Tan et al. (2003) examined data from 1026 subjects from the Framingham study cohort who were free of dementia and prior stroke at baseline. All participants had between 9 and 20 cholesterol measurements every two years leading up to the baseline moment, at which time they were 76.1±5.3 years old. They found no association between current serum cholesterol or prior serum cholesterol and who developed AD in the following decade. Prior cholesterol was calculated by averaging all measurements of at least 5 measurements (10 years) before
baseline. The study provided no data on the distribution of the measurement count or participant age, but since measurement count implies that the data dates back 16-38 years before the baseline at which participants were 76 years old, one could speculate that measurements were a roughly equal mixture of mid and late life. Lack of descriptive statistics on measurement count makes this uncertain.
Warren, Hynan & Weiner (2012) performed a cross-sectional study on 148 AD subjects (aged 79.5) and 197 non-demented controls aged 70, all Americans. When comparing these groups, they found LDL cholesterol to be positively associated (OR = 1.017) with AD, while HDL was negatively associated (OR = 0.976) with the disease, a pattern also typically seen for atherosclerosis. Though significant, odds ratios are very small and much smaller than ApoE-ε4 (OR = 4.002). Still, these results stand out from other findings, as cholesterol was still associated, despite AD subjects being well into late life at 79.5 years old.
Conclusively, midlife cholesterol levels tend to predict AD relatively well, but this effect diminishes, if not disappears completely during late life. This has been postulated to be reverse causation, though more research is required to determine why this peculiar discrepancy exists.
Cholesterol and β-amyloid
There is evidence for an association between cholesterol and β-amyloid deposition as well. Reed et al. (2014) examined 74 Americans at cognitive or vascular risk aged 78±6 years. Using PET scanning, they measured cerebral β-amyloid, which was quantified using the global PIB index. They found more cerebral β-amyloid in those with higher serum LDL cholesterol and lower HDL cholesterol, independent of ApoE genotype. A similar pattern was revealed in an autopsy study on Americans by Kuo et al. (1998) on 64 subjects with AD and 36 non-demented controls. They found lower levels of serum HDL and higher levels of LDL in the AD subjects, and also more amyloid deposition in AD subjects with higher levels of LDL, independent of ApoE genotype. This is consistent with causal evidence from animal models, where high cholesterol diets caused more amyloid deposition in mice (Refolo et al., 2000; Shie et al., 2002) and rabbits (Sparks et al., 1994), and where a cholesterol-lowering drug reduced β-amyloid pathology in mice (Refolo et al., 2001).
These studies suggest that lipid profiles with high LDL and low HDL, currently known to be atherogenic, may also be amyloidogenic.
Statins
Artificially lowering cholesterol using statin drugs has shown relatively limited results in humans with AD. Wolozin et al. (2000) examined hospital records of 57104 Americans aged 74±1 using either statins or other cardiovascular medications. They found 69.6% less probable AD in patients using statins compared to patients using non-statin medications. This is an interesting figure, though associative in nature, in a diseased and medicated population without healthy controls. To determine the causality of these kinds of effects, many randomized controlled trials were performed
where AD patients were given statins, with limited results.
Sparks et al. (2006) performed a double-blind randomized controlled trial in an American population, where 80mg atorvastatin was administered daily to 32 AD patients aged 78±1 for one year, who were compared with 31 AD patients receiving placebo. Participants performed the ADCS-CGIC test for clinically significant change, the ADAS-cog cognition test as well as the MMSE mental state test every three months during the study. No effect was found for the ADCS-CGIC or MMSE test and for the ADAS-cog test the significance threshold was only crossed at the 6 month mark, where the experimental group scored higher. The authors conclude the effect was lasting because the MMSE results at the 12 month mark were close to the significance threshold.
Feldman et al. (2010) also gave 80mg atorvastatin daily in a double-blind randomized controlled trial, but using a larger population of 314 experimental subjects and 326 control subjects aged 74±9, across 10 studies, and using a longer time frame of 72 weeks. 65.9% of experimental subjects and 75.2% of control subjects completed the study. Though lipid profiles improved, they found no effects on either the ADAS-cog test or the ADCS-CGIC test at the end of the study.
Sano et al. (2011) also performed a double-blind randomized controlled trial and gave 40 mg simvastatin to 204 experimental subjects and compared them with 202 controls, all American, with AD and aged 75±9 for 18 weeks, using a half dose during the first 6 weeks. They found no significant difference in score on the ADAS-Cog or MMSE test or any of the other conducted tests at any of the 3-month measurement intervals.
In another double-blind randomized controlled trial, Simons et al. (2002) gave up to 80mg simvastatin daily for 26 weeks to 24 experimental subjects and compared them with 20 placebo subjects aged 68±9. Two isoforms of β-amyloid in the cerebrospinal fluid were measured at the beginning and end of the study, as well as several other lipids. They found no main effects, only at post-hoc analysis a decline in one isoform of β-amyloid was found in patients who scored highest on the MMSE.
Overall, these randomized controlled trials all show practically no benefit in using cholesterol lowering agents in AD disease. This creates an interesting contradiction. The previously discussed studies in this section where atherosclerosis and total or LDL cholesterol, the very factors these drugs are proven to reduce, were quite consistently associated with increased risk of AD pathology.
This discrepancy could be because these randomized controlled trials looked at people that were already afflicted by AD at the beginning of the study. AD is generally considered to be preventable, though largely irreversible (de la Torre, 2010), which may explain why these kinds of trials have a hard time eliciting improvement. A recent meta analysis by Chu et al. (2018) that explicitly excluded randomized controlled trials found statins to be associated with a reduced risk of AD (aRR = 0.719). Thus, one could postulate the hypothesis that the lipid lowering effect of statins might be preventive, but have little effect once the damage is already done. Another potential reason could be the side effects of statins, which include memory loss, dizziness, drowsiness, confusion and
fatigue, all of which could relate to cognition (Rojas-Fernandez & Cameron, 2012). Though these side effects normally occur in a relatively small portion of the population, if they would occur in otherwise subthreshold levels, a small effect of the lipid lowering impact of these drugs could have been cancelled out by a similarly small level of cognitive side effects. Most trials also used a very high dosage of 80 mg of atorvastatin, which may increase the prevalence of side effects compared to a more common dosage. Lowering serum lipids by eating more plant based and less processed is likely to have only positive side effects and thus provides an alternative without this possible disadvantage. Evidence for this is discussed in chapter 3.2.
Cholesterol oxidation
One interesting finding that complicates the relation between cholesterol and AD is that cholesterol can not efficiently pass the blood-brain barrier (Gamba et al., 2012). This means cholesterol from the bloodstream can not get into the brain, but also that the brain can not normally excrete cholesterol. Because the brain has no ability to break down cholesterol, to prevent its accumulation, cholesterol may be oxidized into oxysterols, since in oxidized form, cholesterol is able to pass the blood-brain barrier (Gamba et al., 2012). This strategy is likely not harmless though, for oxysterols may be between one and two orders of magnitude more toxic than cholesterol, and promote inflammation, oxidative stress and atherogenesis (Otaegui-Arrazola, Menendez-Carreño, Ansorena & Astiasarán, 2010). These findings complicate the relation between cholesterol and AD, due to the possibility of interrelations between cholesterol, oxysterols, antioxidants and different parts of AD pathology.
Oxysterols are a source of oxidative stress, which can contribute to AD and other chronic diseases through several mechanisms (Poli, Biasi., & Leonarduzzi, 2013). Oxidative stress may play a key role in AD and is associated with β-amyloid deposition, neurofibrillary tau tangles and AD (Bonda et al., 2010). Nunomura et al. (2001) found oxidative stress to be greatest early in the disease, while it reduces with disease progression.
Oxidized cholesterol may also contribute to hypoperfusion directly, because it has been shown to stimulate monocyte endothelial interactions (Berliner et al., 1990). These interactions are one of the first stages of cardiovascular disease and cause atherosclerosis and artery stiffness (Mestas & Ley, 2008).
Enzymes that catalyze the oxidation of cholesterol have also been found to be abnormally increased in astrocytes of AD brains (Bogdanovic et al., 2001) and colocated with β-amyloid plaques (Brown et al., 2004). Furthermore, brain oxysterol concentrations have been found to be higher in more severe stages of AD (Testa et al., 2016).
Different forms of oxidized cholesterol may play different roles as well. Heverin et al. (2004) compared eight autopsied AD brains with eight age and gender matched controls, and found higher levels of 27-OH-cholesterol in AD brains, but not for 24S-OH-cholesterol. Heverin et al. (2005) demonstrated in vivo that different types of
oxidized cholesterol in the blood may also be deposited or excreted in the brain. Oxysterols were measured in the arteries towards and veins coming from the brain in 12 and 8 healthy volunteers in two subsequent studies. By comparing differences in oxysterol concentrations, a net uptake of 27-OH-cholesterol and a net output of 24S-OH-cholesterol by the brain was suggested, showing evidence for exchange of different oxysterols between the brain and the bloodstream.
Some clues for a potential causal mechanisms were found in an animal model by Marwarha et al. (2010). They showed that 27-OH-cholesterol increased β-amyloid, phosphorylated tau levels and leptin expression in rabbits. Interestingly, treatment with leptin reverses β-amyloid and phosphorylated tau levels. They also showed that a high cholesterol diet reduces leptin expression in rabbits and conclude that a high cholesterol diet can cause AD pathology through a reduction in leptin expression.
A recent meta-analysis (Wang et al., 2016) compared studies looking at levels of cholesterol, 24-OH-cholesterol and 27-OH-cholesterol in the cerebrospinal fluid in AD patients. Both oxysterols were found to be significantly elevated in patients with AD, though the difference in unoxidized cholesterol was insignificant. They propose cholesterol homeostasis is disturbed in preclinical AD, whereas metabolite dysregulation occurs throughout the disease process, with these oxysterols being important biomarkers in the disease.
In conclusion, atherosclerosis in the brain and in the rest of the body are both associated with AD, as well as serum cholesterol levels, the main biomarker for the disease. Interestingly, some studies even suggest a relation with β-amyloid plaque, which may pose interesting options for future research to further assess the atherosclerosis-AD relation. Time of measurement of serum cholesterol is important, as late life measurements are likely confounded by a yet to be described reverse causation mechanism. Treating high cholesterol with lipid lowering statins does not seem effective in reducing AD risk. More research is needed to assess the effects of a similar lowering using dietary measures with less side effects. Oxysterols is a diverse and relatively recent research topic with potentially different pathological effects being exerted by different oxysterols. Cholesterol oxidation products and the resulting oxidative stress they are associated with may play key roles and be a driving force in the development of AD (Gamba et al., 2015). This is especially interesting in the context of nutrition, because the antioxidant capacity of one’s diet may play a role in mediating these processes.
2.3 Diabetes
Vascular implications of diabetes
Diabetes is characterized by chronically elevated blood glucose levels (hyperglycemia). Type II diabetes, also known as adult-onset diabetes, is by far the most common type of diabetes and results from impairments in insulin functioning that prevent glucose from being absorbed (insulin resistance). Diabetes may not immediately be thought of as a vascular disorder, however, it has many vascular implications. The resulting hyperglycemia increases blood viscosity from an early stage of the disease, which has long been known to impair blood flow (Grotta et al., 1982) and can be caused by oxidative stress (Richards & Nwose, 2010). Furthermore, in later stages of diabetes, conditions such as diabetic peripheral vasculopathy (lack of blood flow to extremities) and diabetic angiopathy (blood vessel impairment) may occur (Donnelly, Emslie-Smith, Gardner, & Morris, 2000). Diabetes is also highly comorbid with heart disease, hyperlipidemia and hypertension (Sullivan et al., 2005). Heart disease is also the most common cause of death in diabetics, rather than the disease itself (Gu, Cowie & Harris, 1998). In this chapter, we will review current evidence on the relation between diabetes and AD, and how reductions of diabetes symptoms can change AD risk.
Diabetes measured against other vascular risk factors
Some studies already discussed in the cholesterol chapter also looked at diabetes. The study by Luchsinger et al. (2005) on 1138 elderly Americans found diabetes to be the only factor increasing AD risk on its own (HR = 3.6), while heart disease and hypertension were only significant combined with other risk factors. The Helzner et al. (2009) study on 4166 Americans of whom 156 who developed AD, found only cholesterol and diabetes to increase the rate of cognitive decline, but not hypertension. Li et al. (2011) who followed 837 Chinese MCI patients found an adjusted hazard ratio of 1.6 for developing AD in individuals with diabetes, similar to the adjusted hazard ratio found for hypertension and larger than hypercholesterolemia. These multivariate studies suggest that diabetes may be equally harmful in increasing AD risk as high cholesterol and atherosclerotic heart disease.
Epidemiology of diabetes and AD
Glucose has long been suggested to play a role in AD pathology (Messier & Gagnon, 1996). One of the first longitudinal studies on this topic (Yoshitake et al., 1995) followed 828 nondemented Japanese people (aged 74±6) for 7 years of which 42 developed AD. They found a relative risk of 2.18 to contract AD in those with diabetes, though this result was just shy of significance (95% CI: 0.97-4.90).
Leibson et al. (1997) followed 1455 elderly American individuals with preexisting adult onset diabetes for a total of 9981 person years. Their rates of incidence AD were compared with data from nondiabetic patients from the same region. Diabetes was found to increase the risk (RR = 2.27 for men; RR = 1.37 for women) of developing AD. This effect has also been explored within other ethnicities. Luchsinger et al. (2001)
followed a cohort of 1262 elderly individuals, of whom 32% black and 44.5% hispanic, for an average of 4.3 years. An association between diabetes and cognitive decline was found (HR = 1.6), but the association between diabetes and AD did not reach significance (HR = 1.3).
Peila et al. (2002) examined 2574 elderly Japanese Americans that had survived the Honolulu-Asia aging study. Type II diabetes status was associated with both AD (RR = 1.8) and VaD (RR = 2.3). Individuals with both type II diabetes and ApoE-ε4 genotype were at the highest risk (RR = 5.5). Autopsy data was available for 216 subjects that were deceased before the end of the study. For autopsied individuals, type II diabetes was not associated with neuritic plaques or neurofibrillary tangles in the cortex or hippocampus, nor with cerebral amyloid angiopathy.
Arvanitakis et al. (2004) followed 824 elderly catholic American nuns, priests and brothers for up to 9 years, who underwent annual cognitive assessment. They found those with diabetes to be at increased risk for developing AD (HR = 1.65), as well as lower global cognitive function, visuospatial ability, episodic, semantic and working memory. Diabetes was also associated with a faster ratio of decline in perceptual speed. Not all evidence for a link between diabetes and AD points in the same direction. Curb et al. (1999) examined 3774 Japanese-American men at the ages of 45-68 and again after an average of 15 years at the ages of 71-93. An effect for diabetes at baseline was found on all dementia (RR = 1.37), VaD (RR = 1.53), but not for AD (RR = 1.00).
MacKnight, Rockwood, Awalt & McDowell (2002) found similar findings following 5574 nondemented elderly Canadians for 5 years. Diabetes at baseline was significantly associated with VaD (RR = 2.03), but not with AD (RR = 1.30).
Akomolafe et al. (2006) examined 2210 elderly subjects of the Framingham Study who were still alive at the 16th biennial examination. At that time, 237 subjects had developed AD, 17 of which had diabetes at baseline. Diabetes was not found to be a significant predictor (RR = 1.15) of AD.
Management of blood glucose
Xu et al. (2009) followed 1248 nondemented people from Stockholm aged over 75 for 9 years. At the end of the study, 320 developed AD. No significant effect was found for diabetes on the chance to develop AD. However, individuals who had uncontrolled diabetes did have a significantly increased chance (HR = 3.29; 95% CI: 1.20–9.01) to develop AD. Uncontrolled diabetes was defined as having had a >11 mmol/l blood sugar measurement, which included subjects without a diagnosis. This suggests that diabetes would increase the odds of developing AD, but not if the disease is successfully attenuated by medication. This is consistent with Freiherr et al. (2013), who reviewed a number of studies suggesting that intranasal insulin may be of help in managing AD and concluded these to be a promising strategy.
This idea is complicated by the findings of Huang et al. (2014), where 71433 newly diagnosed diabetics (age 59±14 years) from Taiwan were followed for 10 years and compared with 71311 nondiabetic matched controls of similar age, gender,
hyperlipidemia, hypertension and stroke status. Significant effects on AD incidence were found for diabetes (HR = 1.76), hypertension (HR = 1.3) and stroke history (HR = 1.79). However, they further found that for diabetics that had insulin treatment, risk of developing AD was significantly higher. Risk of AD was the highest (HR = 2.17) in patients using insulin combined with other diabetes medications.
Similar effects were found in Ott et al. (1999), who followed 6370 elderly Dutch subjects of whom 692 had diabetes at baseline for an average of 2.1 years. At follow up, 89 had developed AD and diabetes was found to significantly increase the chance of developing AD (OR = 1.9), even more so for diabetics using insulin (OR = 4.3).
In conclusion, studies suggest that diabetes and insulin treatment are both associated with an increased risk of developing AD. It is important to consider the methodological differences between these studies, however. Not all diabetics take or require treatment with insulin. Only when the disease reaches a certain level of severity, insulin therapy is started. Therefore, patients on insulin are likely to be the more severe cases, for whom a higher risk of AD is imaginable as well. Conversely, higher AD incidence in those with uncontrolled diabetes found in Xu et al. (2009) was effectively a measure of highly elevated blood glucose levels, which may also be indicative of more severe diabetes pathology. Therefore, all these results ultimately support the idea that more severe diabetics results in a higher increase in AD risk. To verify this, more research in the form of clinical trials or large population studies proportionally (not categorically) controlling for the degree of medication usage is required.
2.4 Hypertension
Implications of high blood pressure
When the pressure exerted on the artery walls is chronically elevated, risk increases for both ischaemic and hemorrhagic stroke (O’Donnell et al., 2010). However, people with hypertension who do not experience a full stroke, can still be at increased cognitive risk through several mechanisms, one of which is in the form of microbleeds (Henskens et al., 2008). Microbleeds are tiny deposits (<5.7mm in diameter) of blood degradation products that tend to be colocated with structurally abnormal blood vessels (Martinez-Ramirez, Greenberg & Viswanathan,
2014) and can be made visible with MRI (Figure 3). Though these microbleeds are often not noticeable when they occur, their long-term aggregation may have a crucial role in the pathophysiology of AD, because microbleeds are often found adjacent to β-amyloid deposits and may therefore be another missing link between the amyloid cascade and the vascular hypothesis (Cordonnier & van der Flier, 2011). Hypertension is additionally interesting within the context of AD because of the possible effects of its comorbidities, including previously discussed factors such as atherosclerosis and diabetes (Milionis, Florentin & Giannopoulos, 2008), but also cerebral hypoperfusion (Nobili et al., 1993), which plays a key role within the vascular hypothesis of the disease. In this chapter, we review current evidence for an association between hypertension and AD. Hypertension measured against other vascular risk factors
Some of the studies discussed on cholesterol and diabetes also looked at hypertension. In 828 Japanese people, Yoshitake et al. (1995) found no significant associations between any vascular risk factors and AD incidence, though systolic and diastolic blood pressure had a lower relative risk (1.01; 1.13) and were much further from significance than diabetes (RR = 2.18).
Kivipelto et al. (2001) found that in 1449 Finns, midlife systolic blood pressure over 160 predicted (RR = 2.3) late life AD similarly well as high cholesterol (RR = 2.1) and more so than ApoE-ε4 genotype (Kivipelto et al., 2002).
Luchsinger et al. (2005) found neither hypertension (RR = 1.5; 95% CI: 0.9-2.6) nor heart disease (RR = 1.2; 95% CI: 0.4-3.5) to increase AD incidence on their own in 1138
Americans, but combined, the risk increased significantly (RR = 2.7; 95% CI: 1.5-4.7), though still less than diabetes on its own (RR = 4.8; 95% CI: 1.9-11.6).
Li et al. (2011) found that in 837 Chinese MCI patients, the risk of AD was significantly increased by hypertension, diabetes and cholesterol, with hypertension being the largest risk factor (HR = 1.8).
In a large Japanese population, Huang et al. (2014) found the risk of developing AD to be increased in those with hypertension (HR = 1.30; p = 0.01), which was more than hyperlipidemia (HR = 1.06; p = 0.74), though diabetes increased the risk the most (HR = 1.76; p < 0.001).
The ratio of risk factors is different between these studies. In general, hypertension looks to be part of a comorbid cluster that increases risk for AD. However, it may be less strong of a risk factor than diabetes, though this may depend on the population. In the rest of this chapter, more evidence on the relation between hypertension and AD is reviewed.
Midlife vs late life hypertension
As with cholesterol, whether blood pressure is high in mid or late life is an important distinction on its ability to predict AD. One of the earliest studies suggesting this was Burke et al. (1994), who found blood pressure to gradually decrease between the third and ninth year after diagnosis in three confirmed AD patients. This is suggested to happen due to degeneration of vasomotor neurons responsible for regulating blood pressure. These findings were confirmed in Skoog et al. (1996), who assessed the blood pressure and dementia status of 382 70-year olds from Gothenburg every 5 years. They found high diastolic blood pressure at age 70 to increase the risk of developing AD at age 79-85. Though blood pressure generally dropped after the age of 75, in subjects who developed AD between ages 79-85, both systolic and diastolic blood pressure dropped more than in subjects who remained non-demented. This indicates there may be large differences between mid and late life hypertension in their ability to predict AD.
Midlife hypertension
Launer et al. (2000) followed 3703 Japanese-American men, initially non-demented and aged 53±5 for 25 years. They found those with high diastolic blood pressure at the beginning to be more likely to develop AD by the end of the study (OR 3.49 for 90-95 mmHg, OR 4.67 for >90 mmHg). Untreated systolic high blood pressure was not a significant predictor. Interestingly, no increased risk for any type of hypertension was found for participants that got treatment for their blood pressure, suggesting that medication can make a substantial difference in attenuating any neurodegenerative effects of hypertension.
Wu et al. (2003) followed 16488 Chinese aged >50 years old at baseline for 15 years. 301 of these developed AD, which were compared to 301 age, gender and location matched controls. Both systolic and diastolic pressure at baseline were found to be associated with AD and a significant dose-response relationship was found between
systolic blood pressure at baseline and the odds of having developed AD at the end of the study. Risk was lowest in those with a blood pressure less than 110/80. They further confirmed that diastolic, but not systolic blood pressure dropped dramatically between baseline and the end of the study for both AD subjects and controls.
Ninomiya et al. (2011) followed for 17 years 534 non-demented Japanese whose blood pressure had been measured 15 years prior to the study when they were 50-64 years old. While an association with prior hypertension was found for VaD, they did not find any such effects for AD.
Late life hypertension
Posner et al. (2002) examined the blood pressure, dementia and cognition status of 1259 elderly Americans every 18 months for a period of 6 years. During the experiment, 157 subjects developed AD. No association was found between late life hypertension, cholesterol or diabetes and AD or cognitive decline, while VaD was found to be associated with vascular factors.
Qiu, Winblad, Viitanen & Fratiglioni (2003) followed 1270 non-demented Swedes aged over 75 for a median of 4.7 years. Neither systolic nor diastolic hypertension at baseline predicted the development of AD, which happened in 256 of the subjects over the course of the study. However, pulse pressure, the difference between systolic and diastolic pressure did predict AD both in those higher (RR = 1.4) and lower (RR = 1.7) than the median. They conclude that high pulse pressure likely increased disease risk because it is caused by artery stiffness and atherosclerosis, while low pulse pressure could have done so as a result of cerebral hypoperfusion.
Mielke et al. (2007) followed 135 Americans with AD for a mean of 3 years and regularly assessed their blood pressure, as well as their cognition using the Clinical Dementia Rating and the Mini-Mental State Examination. An average of 2.1 measurements were performed. They found systolic hypertension and angina to be associated with more rapid cognitive decline on both tests. This association was even stronger for those with a higher age at baseline. Hypertensive medications were associated with a slower rate of decline, as were diabetes and bypass surgery, though the authors conclude that this effect may have been found due to selective survival. Management of late life hypertension was further examined by Khachaturian et al. (2006), who followed a cohort of 3308 elderly (>65 years) Americans for three years, a period during which 104 cases of AD occurred in the population. They found use of antihypertensive medication be negatively associated with incident AD (HR = 0.64). In conclusion, these studies show that AD is associated with hypertension, but not as much as VaD is. Differences between studies may exist due to different biases between studied populations with regards to when AD, VaD or both are diagnosed. Though most studies suggest that hypertension is not as strong of a risk factor as diabetes, large differences in risk between treated and untreated hypertensive patients suggest there is value in reducing the condition.
3. Critical discussion: Effect of WFPB diets on vascular
risk factors
3.1 Introduction
DefinitionA whole foods plant based diet implies two inherent properties. Whole foods, meaning minimally processed foods should be favoured over processed foods and, plant-based, meaning a variety of plant-based foods should be consumed instead of animal-based foods. Avoidance of animal foods means all types of meat, eggs or dairy should not be consumed. Minimal processing entails that foods that have been concentrated or have edible nutritious components removed, such as white flour, white rice and regular pasta should be substituted for their “whole” counterparts, such as whole grain flour, brown rice and whole grain pasta, respectively. Minimal processing also means avoiding added sugar, excess salt and artificial flavourings. In most versions of the diet, minimally processed is understood as avoiding all oils as well.
Essentially, the diet consists of any kind of legumes, whole grains, fruits, vegetables, nuts and seeds. Only health promoting seasoning is used, such as herbs, spices, aromatic vegetables, vinegar or nutritional yeast flakes. Sauces or condiments are also prevalent in the diet, as long as these don’t contain added sugar, excess salt or other aforementioned ingredients to be avoided. Some examples of fitting condiments are ketchup, mustard, salsa, pico de gallo, hummus, tahini, nut butters and guacamole. Store-bought versions of these products may contain processed ingredients, so followers of the diet may opt for homemade versions using healthier substitutions. Some examples of substitutions are replacing sugar with a few blended dates, replacing oil and animal fats with small amounts of avocado or tahini if part of a sauce, or replacing dairy milk with unsweetened plant-based milk substitutions.
Often, consumption of more specific food types deemed particularly healthful is additionally encouraged, such as dark green leafy vegetables (kale, spinach, arugula etc) for their nitrate and micronutrient content, berry fruits (blueberries, strawberries, raspberries, blackberries etc) and grapes for their high amount of antioxidants or flax seeds for their lignan compounds and omega-3 fatty acids.
Versions of the diet meant for patients with or at risk for cardiovascular disease may also recommend to eliminate coconuts, avocados and nuts to minimize fat intake, and recommend a very high intake of green leafy vegetables to maximize endothelial recovery (Esselstyn et al., 2014).
Over the last decade, WFPB diets have seen increased media coverage in online groups and in somewhat sensationalist documentaries such as What the Health, Forks over Knives, Plantpure Nation, Food Choices and Eating You Alive.
Nutritional profile of WFPB diets
Though WFPB diets and vegan diets are both void of animal products, the difference is extremely important. Any foods that do not contain animal-based ingredients qualify as vegan, regardless of their level of processing. Though a vegan diet usually contains many whole foods, a diet consisting of nothing but fries, crisps, candy bars and soft drinks could technically still qualify. Since in the WFPB diet, processed foods are replaced with minimally processed ones, there are more guarantees with regards to nutrient density. For this reason, low intake levels of specific amino acids, iron or calcium commonly feared in vegan diets are much less likely, especially because legumes and green leafy vegetables are staples in the diet and high in these nutrients. In many older trials, diets are used that are labeled as low-fat vegan, which also avoid oil and are generally based around whole foods and therefore included in the review. WFPB diets tend to be very high in fiber, antioxidants and many micronutrients (Karlsen et al., 2019). Conversely, they are very low in fat, especially trans and saturated fat and completely void of cholesterol (Jak še et al., 2020). However, the diet is also very low in vitamin B12 and vitamin D (Karlsen et al., 2019), the supplementation of which is therefore non-optional. Though this may raise understandable concerns, the Academy of Nutrition and Dietetics (the largest organisation of nutrition professionals in the world) has deemed appropriately planned vegan diets to be suitable for all stages of life, including pregnancy, lactation, infancy, childhood, adolescence, adulthood and also for athletes (Melina, Craig & Levin, 2016). Indeed, their definition of appropriate planning includes taking a B12 supplement, and a vitamin D supplement for those getting low amounts of sun exposure. Insufficient vitamin D levels are extremely common among the elderly, regardless of diet (Linnebur et al., 2007), so supplementation is likely a good strategy for those at risk for AD in any case, especially because a deficiency may increase AD risk (Littlejohns et al., 2014). Similarly, all adults over the age of 50 are recommended to consume vitamin B12 in supplemented form as well (National Institutes of Health, 2020). Conclusively, supplementing these two vitamins absent in this diet is the recommended strategy for anyone in the age range where AD is a concern, regardless of their diet.
Next to vitamin B12 and D, calcium is the only other nutrient that tends to be low in WFPB diets, as the estimated intake (959 mg/day) does not reach the recommended intake for one demographic: women aged 51-70 (Karlsen et al., 2019). It is important to consider that these recommendations vary widely even in the developed world. In contrast to the American standards (1200 mg/day) used in this study, this daily intake would have been enough for this demographic in countries such as Korea (700mg/day; The Korean Nutrition Society, 2010), the United Kingdom (700mg/day; British Nutrition Foundation, 2016) and in Scandinavia (800mg/day; Norden, 2012). Furthermore, a typical modern diet low in fruits and vegetables, and high in animal products is known to cause an acidic serum pH, which results in excess calcium loss as the body buffers the pH change (Adeva & Souto, 2011). Conversely, an opposite diet high in plant-based foods, has a calcium retaining effect and predicts greater bone
density in children and postmenopausal women (Adeva & Souto, 2011). A diet consisting of whole plant foods with calcium intake not too far below the recommended intake may thus not be as problematic, since recommended intakes in Western countries are based on individuals eating acidosis causing Western diets.
Though the WFPB diet is high in short chain omega-3 fatty acids found in many whole plant foods, especially seeds, it is low in long chain omega-3 fatty acids such as EPA and DHA, commonly found in fish and seafood. Short chain omega-3 fatty acids are metabolised into long chain omega-3 fatty acids, though conversion rates are dependent on gender and hormone balance (Giltay et al., 2004). This could be a potential problem, as long chain omega-3 fatty acids (Cunnane et al., 2009) and fish (Wu et al., 2015) have been associated with positive AD outcomes. Other studies however, show fish (Danthiir et al., 2014), seafood (Masley, Masley & Gualtieri, 2012) or plasma omega-3 fatty acids (Laurin et al., 2003) to be associated with negative AD outcomes. A meta-analysis of randomized controlled trials failed to show a benefit of omega-3 fatty acids for AD as well (Sydenham, Dangour & Lim, 2012), though another meta-analysis from the same time period suggests a benefit in non-demented but cognitively declined subjects (Mazereeuw et al., 2012). A randomized controlled trial using very large doses also failed to show an effect (van de Rest et al., 2008). A recent Cochrane review also showed no effect on a variety of measures (Burckhardt et al., 2016). Clearly, data on this subject is still conflicting and consensus may change in the future. If at any point benefits of direct intake of long chain omega-3 fatty acids are shown more clearly, or there is reason to believe an individual does not metabolize short chain omega-3 fatty acids properly, these can be supplemented. These supplements are generally made using fish, though plant-based versions of these supplements, often extracted from algae, are available as well. Plant-based EPA/DHA supplements have the added benefit of being lower in persistent environmental pollutants that commonly accumulate in fish (Winwood, 2013), which have been shown to cause oxidative stress in animal models (Hong et al., 2015). In chapter 2, oxidative stress has been discussed as being a driver in AD, which could theoretically explain some of the mentioned undesirable outcomes. Extending the plant-based principle to omega-3 supplementation may thus be beneficial.
Plant-based diets in non-industrialized societies
WFPB diets were originally based on traditional diets of less industrialized cultures where prevalence of chronic (cardiovascular) disease is very low. The Tarahumaras in northern Mexico eat mostly corn and beans, with very low amounts of dietary cholesterol (71 mg/day), fat (12% of calories) and saturated fat (2% of calories) (Connor et al., 1978). Their serum cholesterol was 125±26 mg/dl and correlated strongly with dietary cholesterol intake (r = 0.874). Their average blood pressure was below 111/73 for all subgroups. The population was nearly free of diabetes and hypertension and there was no age related serum cholesterol increase as is common in western cultures.
The Papua highlanders of New Guinea eat a diet consisting of a very high amount of sweet potatoes and a negligible amount of cholesterol (Sinnett & Whyte, 1973). In the full sample of 779 people, diabetes was completely absent, rates of hypertension and coronary heart disease were very low. Their average serum cholesterol was also relatively low at 153 mg/dl. Strikingly, no age-related increase in blood pressure or cholesterol was found.
In rural China, people ate 30% more calories than Americans, though rates of obesity are far lower (BMI 20.5 vs 25.8), though this very likely also depends on differences in physical activity (Campbell, Parpia & Chen, 1998). Furthermore, their total serum cholesterol was very low (127 mg/dl). Prevalence of coronary artery disease was at least an order of magnitude smaller than in the US. Many dietary parameters were vastly different between rural China and the US, including much lower fat intake (14% vs 36%), increased fiber intake (11 vs 33 g/day) and much lower protein intake from animal sources (11% vs 70%). Another study on several rural Chinese cultures found even in the ethnic group with the highest blood pressure of the three studied groups, blood pressure was still very low at 107/71.
Granted, rates of physical activity are also much higher in all these cultures, though even in highly trained western athletes, cholesterol (Cardoso et al., 1994) and rates of hypertension (Berge, Isern & Berge, 2015) do not reach nearly as low.
Biomarker guidelines
The mean serum lipid and blood pressure of the cultures described above are one of the lowest in the world and well below tolerable limits set by health organizations. However, this does not necessarily mean that adopting a diet strict enough to be able to reach these values provides no further benefit. A large meta analysis by Law, Morris & Wald (2009) including 147 randomized controlled trials, showed that the rate of coronary heart disease events keeps decreasing until a blood pressure as low as 110/70 is reached. For cholesterol, O'Keefe et al. (2004) postulates that modern guidelines are based on what is normal within a relatively diseased and unhealthy population. Based on cultural studies and a linear regression model using clinical data, atherosclerosis was estimated not to progress when LDL cholesterol is below 70 mg/dl. For total serum cholesterol, this figure is estimated to be 147 mg/dl (Ezzati & Riboli, 2012). These values are very unlikely to be found in the developed world, except in specific vegan populations (De Biase, Fernandes, Gianini & Duarte, 2007). The reason these values are not the recommended ones, despite their benefits, has been postulated to be due to unwillingness to accept that a very large fraction of the Western population actually has unhealthily high levels (Steinberg, 2006). Another possible reason could be that medication is generally thought of as the main method to reduce cholesterol and blood pressure to amounts this low, meaning that lowering the guidelines immediately labels a large portion of the population unhealthy and a candidate for medication. Significant reductions in these values may however also be achieved by consuming minimally processed plant-based foods. In the next chapters, we will assess the evidence for why
this dietary pattern is optimal for the prevention of the studied risk factors, taking into careful consideration that evidence for intake of foods being health promoting depends on what the foods are compared with.
3.2 Cholesterol and atherosclerotic disease
In chapter 2.2 we reviewed the evidence for a relation between AD and hypercholesterolemia and atherosclerotic disease. Furthermore, it was discussed how cholesterol, specifically LDL cholesterol has been causally linked to atherosclerotic heart disease and that atherosclerosis is unlikely to develop if LDL levels are kept low. In this chapter we will look at evidence on how a WFPB diet may reduce these risk factors, which could be an important mediator through which the diet may reduce the risk of AD.
The Esselstyn study
In “A way to reverse CAD?” by Esselstyn et al. (2014), 198 non-smoking patients (aged 63±10, 91% men, 98% with coronary artery disease diagnosis), many of which severely ill and with multiple cardiovascular comorbidities, voluntarily asked for counseling in plant based nutrition for disease treatment. Patients were encouraged to consume a balanced variety of plant foods, take a vitamin B12 supplement and consume ground flaxseed daily to ensure omega-3 needs were met. As is typical for a WFPB diet, patients were asked to avoid all animal and processed foods, as well as oil, excess salt, caffeine and added sugar. Unique to this study, was avoidance of avocados and nuts, and the recommendation of consuming a generous amount of dark green leafy vegetables several times a day. Exercise was encouraged, but not required.
At the beginning of the study, each participant attended a 5-hour seminar in a small group and was asked to bring a spouse or a partner. The seminar included presentation of research on plant-based nutrition, detailed information on the biological mechanisms through which vascular health may be restored, angiograms of disease reversal, information on techniques to alter existing recipes, a talk from a senior WFPB practitioner on reading nutrition labels and preparing food, a testimonial from previous participants, a plant-based meal and a Q&A session. Patients took home a copy of Esselstyn’s Prevent and Reverse Heart Diseasebook, as well as a recipe book and two scientific articles. Contact information was also given and patients were encouraged to communicate any concerns and subsequent cardiac events and test results of lipid profile, angiograms and stress tests. Patients were also asked to complete and return a food diary for the three first weeks of the intervention and keep taking any prescribed medications.
At follow-up (44±24 months), patients who still reported avoiding all animal products and (knowingly) any oils were considered adherent. Of 177 adherent patients, 81% improved (105 with reduction and 39 with reversal), 9% stayed stable and 10% got worse (1 stroke, 1 stenting, 2 coronary bypasses, 9 with unrelated adverse outcomes and 5 non-cardiac deaths). Of 21 non-adherent patients, nobody improved, 38% stayed stable and 62% got worse (2 strokes, 4 stentings, 3 coronary bypasses, 1 endarterectomy, 1 heart transplant, and 2 cardiac deaths). Some images illustrating reversal were included in the article (Figure 4).
Figure 4. Positron emission tomography showing an increase in myocardial perfusion after three weeks of plant-based nutritional intervention (left) and coronary angiography showing a fully recovered distal left anterior descending artery after 32 months of plant-based nutritional intervention (right) in selected patients. Reprinted from Esselstyn et al. (2014).
The main shortcomings of this study are lack of a placebo group, the potential for bias in categorizing patients and a high variance in followup time between patients. Still, the high rates of improvement, combined with a complete absence of cardiac deaths in all adherent participants, make these results unmatched for an intervention based solely on nutrition. The authors note the avoidance of oil, high intake of dark green leafy vegetables and high compliance as key reasons for these positive results. Communication with the patient and the elaborate seminar were hypothesized as key elements to achieve compliance of the relatively strict diet. The authors strongly emphasize the need for patients to internalize that they themselves can take control over the course of their disease.
Other studies on the reversal of atherosclerosis
Ornish et al. (1990) performed a 1 year randomized controlled trial using a diet close, but not equal to WFPB, combined with lifestyle interventions. Subjects followed a low-fat vegetarian diet, where all animal products were excluded, except for a non mandatory allowance of egg whites and one daily cup of non-fat dairy. Lifestyle interventions included stretching exercises, breathing techniques, meditation, relaxation techniques, social support groups and an exercise program mostly based on walking with a minimum of 3 hours a week. Included in analysis were 28 participants in
