Novel therapy of hyperhomocysteinemia in mild cognitive impairment, Alzheimer's disease, and other dementing disorders

Abstract

Objectives

Studies have produced conflicting results assessing hyperhomocysteinemia (HYH) treatment with B vitamins in patients with normal cognition, Alzheimer's disease and related disorders (ADRD). This study examined the effect of HYH management with L-methylfolate (LMF), methylcobalamin (MeCbl; B12), and N-acetyl-cysteine (CFLN: Cerefolin®/Cerefolin-NAC®) on cognitive decline.

Design

Prospective, case-control study of subjects followed longitudinally.

Setting

Outpatient clinic for cognitive disorders.

Participants

116 ADRD patients (34 with HYH, 82 with No-HYH) met inclusion and exclusion criteria to participate. No study participant took B vitamins.

Intervention

HYH patients received CFLN, and No-HYH patients did not.

Measurements

Cognitive outcome measures included MCI Screen (memory), CERAD Drawings (constructional praxis), Ishihara Number Naming (object recognition), Trails A and B (executive function), and F-A-S test (verbal fluency). Dependent or predictor measures included demographics, functional severity, CFLN and no CFLN treatment duration, ADRD diagnosis, memantine and cholinesterase inhibitor treatment. Linear mixed effects models with covariate adjustment were used to evaluate rate of change on cognitive outcomes.

Results

The duration of CFLN treatment, compared to an equivalent duration without CFLN treatment, significantly slowed decline in learning and memory, constructional praxis, and visual-spatial executive function (Trails B). CFLN treatment slowed cognitive decline significantly more for patients with milder baseline severity. CFLN treatment effect increased as baseline functional severity decreased. The analytical model showed that treatment duration must exceed some minimum period of at least one year to slow the rate of cognitive decline.

Conclusion

After covariate adjustment, HYH+CFLN significantly slowed cognitive decline compared to No-HYH+No-CFLN. Longer CFLN treatment duration, milder baseline severity, and magnitude of homocysteine reduction from baseline were all significant predictors. There are a number of factors that could account for disagreement with other clinical trials of B vitamin treatment of HYH. Moreover, CFLN is chemically distinct from commonly used B vitamins as both LMF and MeCbl are the fully reduced and bioactive functional forms; CLFN also contains the glutathione precursor, N-acetyl-cysteine. The findings of other B vitamin trials of HYH can, therefore, only partly account for treatment effects of CFLN. These findings warrant further evaluation with a randomized, placebo-controlled trial.

Introduction

Hyperhomocysteinemia (HYH), an abnormally high blood level of homocysteine (often defined as >15 μmol/L), is a common condition worldwide, with prevalence in population studies ranging from 5.1% to 29% in persons aged 65 years and older (1., 2., 3.). HYH is associated with a wide variety of developmental and adult disorders. HYH-associated disorders include mental retardation, pregnancy complications, certain cancers, chronic renal disease, osteoporosis, diabetes, metabolic syndrome, eating disorders, sleep-wake cycle disorders, coronary artery disease, peripheral vascular disease, cerebrovascular disease, depression, schizophrenia, Parkinson's disease, multiple sclerosis, mild cognitive impairment (MCI) (4), and progression to dementia due to Alzheimer's disease (AD) and related disorders (ADRD) (5).

HYH also has a wide variety of pathophysiological mechanisms (Table 1), some of which may play a role in many of the HYH-associated disorders. One such pathway – the one-carbon metabolism of the methionine cycle – regulates the normal circadian cycle of histone methylation, gene expression, RNA transcription, and cellular protein synthesis. In the methionine cycle, methylfolate and B12 convert homocysteine to methionine. Methionine then converts to S-adenosyl methionine (SAM) – a major donor of methyl molecules onto DNA and histones. After methylating DNA and histones, SAM converts to homocysteine. Homocysteine either excretes in the urine, or methylfolate and B12 reconvert it to methionine. HYH disrupts this normal methylation process, which alters the circadian clocks that regulate the function of the nervous system and the body's organs (6).

Table 1.

Pathophysiological Mechanisms of Hyperhomocysteinemia

Physiology	Sample	Mechanism
One Carbon Metabolism	Human	Abnormal Methylation of DNA and histones disrupts normal circadian cycle of protein transcription and cellular function, which disrupts energy-related pathways, increases excitotoxicity, disturbs cell cycle signaling, generates synaptic abnormalities, and impairs cellular defense mechanisms (6)
Cholesterol Metabolism	Human Cross-Sectional	Normal homocysteine levels associated with in- verted-U relation between total cholesterol and cognition. In HYH, there was no association between total cholesterol and cognitive performance (38)
Protein Transcription	Chemical Analysis	HYH, via abnormal histone methylation, produces a stable disulfide bond with cysteine residues of albumin, fibronectin, transthyretin, and metallothionein to inactivate or impair the function of these proteins (39)
Beta Amyloid Intra-neuronal deposition	Human Prospective Autopsy, Transgenic Mouse Model	HYH, via abnormal histone methylation, genetically up-regulates Pre-Senilin-1 and Beta Amyloid Cleaving Enzyme, which increases beta-amyloid, intra-neuronal deposition (40, 41)
Neurofibrillary Tangles	Human Prospective Autopsy	HYH associates with increased Neurofibrillary Tangle load (40)
White Matter Ischemia	Human Prospective Autopsy	HYH associates with increased periventricular white matter hyperintensities (40)
Prolonged HYH Exposure	Hippocampal Neuron Culture	HYH lowers neuronal input resistance in hippocam- pal neurons to reduce both spontaneous and induced action potential firing (42)
Oxidized Homocysteine Products	Wistar Rat Embryonic Cortical Neurons	Homocysteine oxidation products (homocysteine-sul- finic acid and homocysteic acid) reduce spontaneous network firing of cortical neurons (43)
Hippocampal Neurogenesis	Mouse Model	Folic acid interacts with HYH in a dose dependent manner to impede neurogenesis in the hippocampal dentate gyrus (32)
Alcohol Toxicity	Chemical Analysis	High levels of alcohol interact with folate to disrupt conversion of homocysteine to methionine, which leads to HYH (33)
Pre- Post-Natal Folate Supplementation	Human	Alters post-developmental response to B12 and folate in converting homocysteine to methionine, leading to higher risk of cognitive impairment in HYH with low B12 and high folate levels.

There is evidence that this disrupted circadian cycle of protein synthesis is causally linked to diabetes (7), metabolic syndrome (8), cardiovascular disease (9), obesity (10), sleep disorders of night shift workers (11), certain cancers (12), behavioral and sleep disorders of AD and of Lewy body disease (13), and in the circadian rhythm of human dorsolateral prefrontal cortical neuronal activity (14). Consequences of HYH-mediated disturbance of the methionine cycle have been studied in AD animal models, which have shown depleted energy-related pathways, increased excitotoxicity, disrupted cell cycle signaling, synaptic abnormalities, and impaired cellular defense mechanisms (15).

HYH is easily managed. Significant cost savings and improved health would occur if HYH treatment reduced risk or progression of one or more of its associated disorders. For example, a study on cardiovascular disease and stroke in Switzerland estimated that HYH accounts for 10% of their total costs (16).

However, some clinical trials of B vitamins to treat mild to moderately demented patients with AD or vascular dementia (VD) have shown no effect of lowering HYH (17, 18). These trials may have shown no treatment effect because patients were too severe (17, 18) or because patients with normal homocysteine levels were included in the trial (18). Also, factors that interact with homocysteine could affect outcomes of HYH-lowering clinical trials (Table 2).

Table 2.

Factors that Interact with HYH and could Account for Conflicting Results in HYH trials with B Vitamins

Factor	Effect of Interaction Between Factor and HYH
Cerebral Infarcts	Increases NFTs in presence of HYH
Study Duration	Greater pathology and cognitive impairment with longer HYH duration
Cognitive Task Type	HYH has selective effects on cognitive task performance
ADRD Severity	HYH effects on cognitive task performance also depend upon severity
Folate During Development	High folate + low B12 levels associate with greater cognitive impairment.
No Folate During Development	High folate + low B12 levels associate with lower risk of cognitive impairment.
Low Folate Levels	Low folate + HYH is associated with the poorest cognitive performance
Homocysteine Oxidation	The levels of homocysteine oxidation
Products	products may account for cognitive impairment, and not HYH.
Folate Levels	Dentate gyral hippocampal neurogenesis depends on folate and HYH levels
Alcohol	High alcohol levels disrupt folate conversion of homocysteine to methionine, which leads to HYH
Total Cholesterol	High total cholesterol associates with cognitive impairment if homocysteine levels are normal, but not if they are high.
CFLN Chemistry	L-Methylfolate is not the same as folic acid. NAC is only present in CFLN

An alternative treatment of HYH that may be more effective than B vitamins is a prescription medical food containing L-methylfolate, methylcobalamin (MeCbl), and N-acetyl-cysteine (CFLN: Cerefolin®/CerefolinNAC®). Previously, we retrospectively evaluated CFLN treatment effect in cognitively impaired or demented, HYH patients vs. matched controls (19). CFLN significantly slowed cognitive decline in HYH patients, relative to controls, for cognitive tasks assessing memory, visual-spatial executive function, object recognition, fluency and processing speed.

Therefore, the present study examines the cognitive outcomes of a prospectively followed sample of HYH patients with AD, VD, and mixed dementia treated with CFLN (HYH+CFLN). The HYH+CFLN group's cognitive course was compared to that of a control sample of matched patients without HYH and without CFLN (No-HYH+No-CFLN). Note that HYH patients decline more rapidly than patients without HYH. To demonstrate a statistically significant treatment effect, CFLN must therefore overcome the more rapid decline due to HYH, plus slow decline beyond that of patients without HYH. To avoid any confounding effects of B vitamins, all patients taking supplemental B vitamins were excluded from the study.

Method

Study Protocol

Institutional review board approval was obtained for this study. Patients were recruited from a single-site community memory clinic. Those patients meeting eligibility criteria were consented for study enrollment. Inclusion criteria were that patients had: 1) a final ADRD diagnosis based on standardized evaluation with laboratory, imaging, cognitive and functional assessment; 2) at least one prior quantitative MRI (qMRI) study; 3) previous plasma homocysteine level measured to determine whether or not they had HYH; and 4) either HYH treated with CFLN for some portion of follow up period, or had no HYH. Patients who had been taking supplemental B vitamins were excluded from the study. Available longitudinal data for HYH patients had periods with and without CFLN treatment. This allowed a within-subject analysis of HYH patients that could evaluate the effect of HYH with and without CFLN treatment on cognitive decline. Figure 1 summarizes the study protocol.

Figure 1.

Study Protocol

Patients were assigned to one of two groups based on CFLN treatment status – HYH+CFLN (N=34) vs. No-HYH+No-CFLN (N=82) groups. For all patients, blood was taken to measure homocysteine level plus measure homocysteine-related genes with known single nucleotide polymorphisms (SNP) as well as ApoE allele. qMRI was also obtained for all HYH+CFLN patients and for a subset of No-HYH+No-CFLN patients (N=29). The results of the SNPs analysis and qMRI will be reported elsewhere.

All patients were managed and assessed every 3 to 6 months at the memory clinic. Clinical information captured at each assessment included ADRD diagnosis, pharmacologic treatment and duration, plus cognitive and functional assessments consistent with the study protocol. Patients diagnosed with AD were treated with a standardized protocol of memantine plus a cholinesterase inhibitor. Patients with VD were treated by managing vascular risk factors, by anti-platelet inhibition, and memantine when clinically indicated. All patients were maintained on stable doses of their treatment. These longitudinal clinical data were combined with the study data to obtain longitudinal measures of up to 8 years. The “study baseline” was defined as the first available data point after clinical diagnosis of HYH and initiation of CFLN treatment. The “study endpoint” was defined as the last data point collected under the present study protocol.

Study Group Differences

Table 3a-d summarizes the study group differences. Statistically significant differences between the groups were that the HYH+CFLN group: 1) were older at baseline; 2) had lower baseline MPI scores (20); 3) had a higher proportion of VD, and 4) had greater reduction of homocysteine levels from baseline to study endpoint. However, there were no significant differences between the groups for functional severity measured by the Functional Assessment Staging Test (FAST), endpoint homocysteine levels, study duration, and memantine and cholinesterase inhibitor treatment duration.

Table 3.

Group Differences

(a) Demographic
		No-HYH+No-CFLN	HYH+CFLN
	Total (female/male)	82 (45/37)	34 (22/12)
	% Female	55%	65%
	Age at Baseline (years)	73.7 +/- 7.6	77.6 +/- 6.9
	Education Level (years)	15.6 +/- 3.2	15.6 +/- 2.3
	Baseline MCI Screen MPI Score	43.7 +/- 14.2	41.2 +/- 14.5
	Baseline % Demented (FAST 4-6)	54%	59%
LR chi2(3) = 8.07, Prob > chi2 = 0.0446. Likelihood ratio, assuming unequal covariances.
(b) Diagnosis
	No-HYH+No- CFLN	HYH+CFLN	Fisher’s Exact
Pure or Mixed AD	53	18	0.296
Pure or Mixed VD	14	11a	0.084
Non-AD+Non-VD	22	9	1.000
a. P=0.084
(c) Homocysteine Level (a multivariate vector of means testa)
		No-HYH+No-CFLN	HYH+CFLN
	Endpoint Homocysteine (µmol/L)	8.63+/- 2.6	7.9 +/- 2.1b
	Homocysteine Level: End Point minus Baseline (µmol/L)	-0.66 +/- 2.2	-3.1 +/- 4.2c
	Study Duration (months)	27.6 +/- 19.2	24.5 +/- 17.1d
a. Wilks’ lambda = 0.8687, df(3, 112), F=5.64 P = 0.0012; b. P=0.1043; c. P=0.0025; d. P=0.3951. All hypothesis test were 2-tailed and assumed unequal veriances.
(d) CFLN, Memantine and Cholinesterase Inhibitor Treatment Duration (months)
	Treatment	No-HYH+No-CFLN	HYH+CFLN
	CFLN	0.0±0.0	20.3±17.4a
	Memantine	17.1±16.4	22.6±30.5b
	Cholinesterase Inhibitor	17.0±25.6	19.6±23.6c

^aP= 0.0000 for CFLN; b. P=0.3276; c. P=0.6009.

Assessments

The cognitive assessment tasks included MCI Screen (MCIS) for memory (learning, working memory, episodic memory) (21, 22), CERAD drawings for constructional praxis, Ishihara Number Naming for object recognition, Trails A and B for processing speed, working memory and set-shifting (executive function), F-A-S for verbal (phonemic) fluency, and verbal working memory. The FAST was used to assess functional severity. The FAST has 7 major stages and 16 sub-stages, each with a published mean duration of the untreated course of AD (23).

Outcome Measures

The primary outcome measure was the Memory Performance Index (MPI) of the MCIS – a measure of working and episodic memory. Secondary outcome measures included: 2) Trails A and B time plus total correct and error scores; 3) F-A-S verbal fluency total correct plus error scores; 4) Ishihara object recognition time plus total correct; and 5) CERAD drawing time plus total correct and error scores. The specific measures of these cognitive outcomes were:

• •MCIS MPI
• •MCIS Total Score of the 3 Immediate Free Recall Learning Trials
• •MCIS Total Score of the Delayed Free Recall Trial
• •MCIS Total Score of the Delayed Recognition Memory Task
• •MCIS Total # of Correctly Identified Target Words of Delayed Recognition Task
• •MCIS Total # of Correctly Identified Distractor Words of Delayed Recognition Task
• •MCIS Total Score of the Animal Delayed Free Recall
• •CERAD Drawings Task Total Score of the Correct Drawing Figures
• •CERAD Drawings Task Total Error Score
• •CERAD Drawings Task Drawing Times of the Circle, Diamond, Rectangles, and Cube
• •Ishihara Color Plates Number Naming Task Total Score of Correctly Identified Numbers
• •Ishihara Color Plates Number Naming Task Time to Perform
• •Trails A and B Tasks Times to Perform
• •Trails A and B Error Scores of the Sequence and Set Loss Errors
• •Trails A and B Number of Self Corrections Made
• •F-A-S Letter Fluency Total Error Score
• •F-A-S Letter Fluency Total Number of Common Nouns Named in 3 Minutes
• •Orientation to Date Score

For the analytical model, the variables available for analysis were baseline qMRI hippocampal, cortical and forebrain parenchymal volumes, ApoE genotype, CFLN, cholinesterase inhibitor and memantine duration, start and stop dates, age, gender, education, homocysteine level, ADRD diagnosis, and FAST stage severity. Other potentially confounding variables not available for this analysis were B12 level, methylfolate level, history of developmental folate supplementation, history of alcohol dependence, and lipid status.

The variables to be predicted in this model were the annual rates of cognitive decline per type of cognitive task. These rates were computed as change in cognitive task performance from baseline to study endpoint divided by the elapsed time in years. The variables used in this analysis were incorporated into the analytical model in several scales, when possible. Ratio scale variables were examined at binary, ordinal and ratio scales to determine the one with greatest statistical significance. Similarly, ordinal scale variables were examined at ordinal and binary scales. Interactions between treatment (CFLN, memantine, cholinesterase inhibitor) and severity (FAST stage), ADRD diagnosis, and presence of 1 or more ApoE4 alleles were tested for significant effects. Each variable was included in the model as a fixed effect. Each variable was also introduced into the model as a random effect to determine its impact on the model's predictive power at the individual level. If the random effect of a given variable did not significantly contribute, it was removed as a random effect.

A multi-level model with random and fixed effects was used to predict cognitive rates of change for all patients who had two or more cognitive assessments done during the study. The first model analyzed was the unconditional means model, which had no fixed effects, and included patient as the only random effect. The second model analyzed was a growth model, which used each cognitive task's baseline performance as the only fixed effect predictor, and used patient as the only random effect. Subsequent models used various combinations of the observable variables in terms of their impact on the treatment variables – CFLN, cholinesterase inhibitors, and memantine.

The final model estimated annual rate of decline of each cognitive task in relation to the primary variables of interest for this study, which were CFLN treatment, lowering of homocysteine from baseline, and interactions between CFLN treatment and other variables. For comparison with CFLN treatment effects, memantine and cholinesterase inhibitor treatment effects were also included. The other variables included in the model served as covariates to adjust these primary variables of interest.

Results

Memory Performance Index

Figure 2 shows the change in the MPI score for up to 8 years follow-up. The mean follow-up duration was approximately 2 years for all groups. There were no statistically significant group differences. The black-and-white dotted line and 95% confidence interval shows the changes in MPI score that occurred among HYH subjects when they were not taking CFLN (HYH+No-CFLN). The black dashed line and its gray shaded 95% confidence interval shows the changes in MPI score that occurred among the same HYH patients when they were taking CFLN (HYH+CFLN), and the solid black line and its 95% confidence interval shows the changes in MPI score that occurred among patients without HYH (No-HYH+No-CFLN). There is no overlap in the 95% confidence bands of the No-HYH+No-CFLN and the HYH+No-CFLN treatment groups. This complete separation implies that HYH patients decline more rapidly than No-HYH patients, independent of the effect of CFLN treatment. The partial separation of the HYH+CFLN and HYH+No-CFLN treatment groups within the same subjects implies that there may be a beneficial CFLN treatment effect in slowing cognitive decline. These unadjusted means and their confidence bands will be examined with covariate adjustment of the cognitive tasks assessed.

Figure 2.

Unadjusted Effects of Hyperhomocysteinemia and CFLN on Memory Performance Index

Table 4a shows that MPI is significantly predicted by the patient's education level, their baseline MPI score, the use and duration of memantine treatment, the duration of No-CFLN treatment, and the change in homocysteine level from baseline to end of study. A decrease in homocysteine level from its baseline value associated with decline in MPI score (-0.352 points per 1 point decrease in homocysteine). However, the duration of No-CFLN treatment associated with more rapid decline (-0.118 points/month) than duration of CFLN treatment (0 points/month), such that, after some period of CFLN treatment, the rate of decline in MPI score would be slowed. For example, a 10-point reduction in homocysteine level decreases MPI score by 3.52 points. CFLN treatment for 30 months would counteract this 3.52 point drop in MPI score, and longer CFLN treatment duration would then begin to slow its rate of decline. No other significant predictors were identified.

Table 4.

Cognitive Assessment Outcomes

(a) Predicting MCI Screen Memory Performance Index: Mixed Effects Analysis
Predictor	Coefficient	Std. Error	P>|Z|
Education Level (years)	0.393	0.228	0.085
Baseline MCI Screen MPI Score	0.744	0.050	0.000
Use of Memantine (yes/no)	-8.485	1.775	0.000
Memantine Rx Duration (years)	-0.138	0.024	0.000
No CFLN Rx Duration (years)	-0.118	0.020	0.000
Homocysteine Level: End Study minus Baseline (µmol/L)	0.352	0.171	0.040
Patient Random Intercept (constant)	11.757	4.269	0.006
(b) Predicting MCI Screen Immediate Recall During Learning Trials
Predictor	Coefficient	Std. Error	P>|Z|
Baseline Immediate Recall Total/30	0.585	0.066	0.000
Baseline Hippocampus+Amygdala Volume (cm3)	0.327	0.225	0.147
Use of Memantine (yes/no)	-1.785	0.899	0.047
Memantine Rx Duration (years)	-0.035	0.017	0.043
Months of Symptoms at Baseline	-0.013	0.005	0.009
No CFLN Rx Duration (years)	-0.038	0.010	0.000
Not Taking CFLN-FAST Stage Interaction	-1.068	0.282	0.000
Patient Random Intercept (constant)	8.431	3.003	0.005
(c) Predicting Trails B errors
Predictor	Coefficient	Std. Error	P>|Z|
Education Level (years)	-0.127	0.036	0.000
Baseline Trails B Errors	0.124	0.091	0.175
Baseline Forebrain Parenchymal Volume (cm3)	0.002	0.001	0.077
Cholinesterase Inhibitor Use (yes/no)	0.702	0.170	0.000
Cholinesterase Inhibitor Rx Duration (years)	-0.006	0.003	0.033
CFLN Rx Duration (years)	-0.015	0.006	0.016
Homocysteine Level: End Point minus Baseline (µmol/L)	-0.075	0.026	0.004
CFLN-VD Interaction (yes/no)	0.751	0.325	0.021
Patient Random Intercept (constant)	0.752	0.700	0.283
(d) Predicting errors made doing the CERAD Drawings Task
Predictor	Coefficient	Std. Error	P>|Z|
Baseline CERAD Draw Total Errors	0.735	0.091	0.000
Baseline Forebrain Parenchymal Volume (cm3)	-0.005	0.003	0.055
ADRD due to Cerebrovascular Disease (yes/no)	2.492	0.731	0.001
CFLN-AD Interaction (yes/no)	-1.590	0.593	0.007
No CFLN Rx Duration (years)	0.149	0.009	0.000
CFLN Rx Duration (years)	0.121	0.014	0.000
Homocysteine Level: End Point minus Baseline (µmol/L)	-0.167	0.067	0.013
Use of Memantine (yes/no)	3.524	0.738	0.000
Cholinesterase Inhibitor Rx Duration (years)	0.045	0.012	0.000
Patient Random Intercept (constant)	1.152	2.723	0.672

Immediate Free Recall Task During Wordlist Learning

Table 4b shows that baseline recall score, memantine use and duration, ADRD symptom duration at baseline, No-CFLN treatment duration, and increasing severity combined with No-CFLN use accelerated rate of decline in immediate recall performance. In contrast, CFLN use did not accelerate rate of decline in immediate recall. No-CFLN treatment duration, plus increased functional severity (FAST stage) combined with No-CFLN treatment, accelerated rate of decline in immediate free recall.

Sequencing and Shifting Between Two Sets: Multitasking via Trails B Task

For multi-tasking, or shifting performance and attention between alternating sequences of visual spatial tasks (Trails B), errors of sequencing and errors of failure to correctly shift between two sets of trails (number and letter trails) indicates failure of divided attention or response inhibition. These cognitive abilities are processed in the parietal and frontal lobes. Table 4c shows the predictors that significantly influence these sequencing and set-shifting errors of the Trails B task. CFLN treatment duration significantly reduced Trails B errors (-0.15 points/month). Other predictors of a slower rate of decline in Trails B error performance were years of education and cholinesterase inhibitor treatment duration. Significant predictors of a faster rate of decline in Trails B error performance were cholinesterase inhibitor use, an interaction between CFLN and VD, and a decrease in homocysteine level from baseline (-0.075 points per 1 point decrease in homocysteine). A 10-point reduction in homocysteine level would require 4 years of CFLN treatment to counteract the 10-point reduction, and further treatment would slow rate of decline in Trails B error performance.

Constructional Praxis Errors: CERAD Drawings Task

Constructional praxis involves viewing a geometric figure on a piece of paper, making a plan to reproduce the figure, and executing the plan by drawing it. The figure is always in view so the primary demands are on planning and execution, which are predominantly frontal lobe executive function tasks. Table 4d shows that, over the same duration of time, taking CFLN associated with slower rate of decline in constructional praxis errors than not taking CFLN (0.149 – 0.121 = 0.028 fewer errors/month. Also, CFLN treatment interacted with AD to further slow rate of decline in constructional praxis performance (-1.59 fewer errors). Predictors that increase constructional praxis errors included use of memantine, cholinesterase inhibitor treatment duration, a decrease in homocysteine level from baseline (-0.167 points per 1 point decrease in homocysteine), and VD diagnosis. A 10-point reduction in homocysteine level would require 5 years of CFLN treatment to counteract the 10-point reduction, and further treatment would slow rate of decline in constructional praxis performance.

Discussion

It should be noted that the comparison of HYH+CFLN vs. No-HYH+No-CFLN groups handicaps the HYH+CFLN group because HYH patients more rapidly decline than No-HYH patients. Therefore, a CFLN treatment effect in HYH patients could be missed if CFLN only slowed the rate of decline of HYH patients to the same rate as No-HYH patients. Consequently, it is remarkable that the HYH+CFLN group slowed rates of cognitive decline beyond those of the No-HYH+No-CFLN group.

The Importance of CFLN Treatment Duration in Slowing Cognitive Decline

The present study found that the HYH+CFLN treatment group had significantly slower cognitive decline than the No-HYH+No-CFLN group for the: 1) MPI score (learning, working memory, episodic memory); 2) immediate free recall task (learning, working memory); 3) sequencing and response inhibition task (visual-spatial executive function); and 4) constructional praxis task. From the example calculations done for the duration of CFLN treatment required to slow cognitive decline, this duration is roughly proportional to the reduction of homocysteine from its baseline value. This time requirement is also consistent with other studies in which 1 or more years of B vitamin treatment of HYH was needed to slow the rate of cognitive decline (24, 25). Therefore, studies of less than two years are less likely to demonstrate a treatment effect due to lowering HYH.

Interaction Effects

The following interactions occurred: 1) relatively better constructional praxis in AD with HYH+CFLN group; 2) relatively worse visual-spatial executive function in VD with HYH+CFLN group; and 3) relatively greater impairment in immediate free recall in more impaired patients (FAST stage) when not taking CFLN. The negative interaction between Trails B errors and VD in HYH+CFLN group associated with more rapid decline. However, this decline is counteracted by the duration of CFLN treatment, which associated with slower rate of decline. This interplay between these two effects of CFLN is an example of why CLFN treatment for 1, 2 or more years is needed to demonstrate a beneficial effect.

Possible Factors Affecting HYH Treatment Outcome

Given that the present study shows a significant treatment effect in the HYH+CFLN group on the cognitive course of at least AD and VD, it would be valuable to consider why some, but not all, studies using B vitamins to treat HYH have not found a similar benefit. Table 2 summarizes factors that, if not controlled for or effectively nullified by randomization, could account for conflicting findings among such B vitamin studies. The discussion below considers some factors that could account for different findings of the relation between HYH and cognitive function.

HYH Treatment Duration and Conversion Rate of MCI to Dementia

The VITA, prospective, 5-year study of 81 MCI subjects, found that conversion rate to dementia was lower in those who used folate and B12 together for at least 1-year (24). The Vita-Cog randomized, double blind, controlled study of 266 MCI subjects given B vitamins vs. placebo for two years found that B vitamins significantly slowed cognitive decline plus brain atrophy (25). Because HYH results in excessive methylation of DNA and histones, which disrupt the circadian cycle of cellular protein transcription, it would be useful to measure the effect of HYH treatment on this circadian cycle and on methylation of DNA and histones (26).

Type and Severity of Cognitive Impairment Associated with HYH Treatment Effects

HYH in MCI subjects associates with cognitive deficits in rule generalization, but not in rule learning. This finding implies that the type of cognitive task, and severity of cognitive impairment, can influence outcomes measuring effect of HYH on cognition (27). Selectivity of the type of cognitive impairment was demonstrated in a cross-sectional study of 325 subjects with normal aging, subjective memory complaints, MCI, AD or VD. An extensive neuropsychological test battery was used to evaluate cognitive performance in relation to homocysteine plasma levels. However, only impairment of visual constructional tasks associated with HYH (28).

Folate Supplementation During Childhood and Its Effect on HYH Response

The Hordaland Health Study, Norway, cross-sectionally examined plasma folate and B12 levels in relation to cognitive task performance in 2,203 elderly subjects who had not received mandatory folate supplementation during development. After covariate adjustment for possible confounding factors, the study found that plasma concentrations of low B12 and high folate, compared to the mid-range concentrations of both vitamins, associated with reduced risk of cognitive impairment (29).

These findings are the opposite of those found in subjects who had received folate supplementation during development. For example, the SALA cross-sectional study of 1,535 Latinos, 60 years and older, who had received folate supplementation during development, examined cognitive performance as a function of B12, folate and homocysteine levels (30). The group with HYH, low B12, and high folate concentrations had the poorest cognitive performance. Considering these two studies together, they suggest that folate supplementation during development de-sensitizes folate's post-developmental ability to convert homocysteine to methionine. Developmental folate supplementation can, therefore, alter cognitive outcomes in relation to homocysteine, B12 and folate levels during adulthood.

Folate-Homocysteine Interactions

Low folate levels may cause cognitive impairment but it appears to do so only when there is also HYH. In a cross-sectional, population-based study of 3,914 subjects aged 65 years and older, who underwent a battery of 5 cognitive tasks, only patients with both low folate and high HYH plasma levels showed poorer cognitive performance on all tasks (31).

Folic acid differentially interacts with HYH in a dose dependent manner, to impede neurogenesis in the dentate gyrus of the hippocampus. In a mouse model, HYH with normal folate levels inhibit hippocampal dentate gyrus and olfactory neurogenesis (32). This dose dependent, folate-HYH interaction provides another possible explanation for the conflicting results reported in human studies of HYH lowering with B vitamins.

Altered Circadian Gene Expression with Alcohol Toxicity

Another basis for conflicting study results in B vitamin trials of HYH relates to the interaction between alcohol toxicity and folate. High levels of alcohol interact with folate to disrupt conversion of homocysteine to methionine, which alters circadian gene expression (33). Such altered gene expression is associated with inappropriate timing of cellular protein synthesis, which occurs in a variety of disorders and in AD animal models, disrupts energy-related pathways, increases excitotoxicity, disturbs cell cycle signaling, generates synaptic abnormalities, and impairs cellular defense mechanisms (15).

Differences Between CFLN and B-Vitamins

There are also differences between commonly used B vitamins and CFLN, which may account for any differences in clinical trial outcomes between them. In addition to N-acetyl-cysteine (NAC), CFLN delivers the fully reduced, immediately bioavailable forms of methylfolate and B12, L-methylfolate and methylcobalamin, and is therefore not chemically or biologically equivalent to dietary supplements containing other forms of B vitamins or B vitamins alone.

Neuroprotective Effects of N-acetyl-cysteine

NAC is a precursor of the neuronal antioxidant, glutathione, and may account for beneficial effects of CFLN that would not be found in B vitamins. NAC has been primarily studied in cell culture and in animal models. In hippocampal cell cultures injured by hypoxia, NAC is neuroprotective by reducing reactive oxidation species, increasing cellular antioxidant levels, and reducing DNA strand breaks (34). In the neuronal cell line, N2a, exposed to two hits of the proteasome inhibitor, MG132, administration of NAC prevents neurodegenerative changes due to glutathione depletion and protein misfolding (35). Pre-natal rats administered lipopolysaccharide show pyramidal cell disorganization in CA3 hippocampal neurons, which is blocked by NAC (36). Oxysterols, 24-OH and 27-OH, are abnormally elevated in autopsied AD cortex. Human neuroblastoma cells with patho-physiologically elevated amounts of 27-OH and 24-OH, up-regulate amyloid precursor protein and β-secretase to increase Aβ1-42 production, which is blocked by NAC. The evolutionary anthropoid primate-specific gene locus G72/G30, encodes protein LG72 that may function as a mitochondrial protein and activator of the peroxisomal enzyme, D-amino-acid-oxidase (DAO). Transgenic mice (G72Tg) overexpress G72 and have reduced activity of mitochondrial complex I, increased reactive oxygen species, and behavioral deficits resembling psychiatric diseases. Affected neurons have impaired short-term plasticity with poorly sustained synaptic activity, and spatial memory deficits, which are prevented by treatment with NAC (37). These findings, combined with those of the present study, warrant further study of NAC treatment effects in humans with dementing disorders.

Study Limitations

One limitation of the present study is the non-random nature of the sample. Although statistically powered to achieve acceptable type I and II error rates, multiple analyses were done, such that a larger, randomly assigned sample should be studied to confirm the present findings. In this regard, it would be useful to compare placebo treatment to CFLN.

Another study limitation is that a number of the aforementioned factors that alter homocysteine conversion to methionine were not adjusted (see Table 2). A future study should collect data on these exogeneous factors (e.g., folate supplementation during childhood), and adjust them as potentially confounding covariates.

Conclusion

CFLN management of HYH significantly reduced rate of cognitive decline in measures of learning, working memory, episodic memory, constructional praxis, plus visual-spatial executive function at least in patients with AD and VD, and possibly other dementing disorders. Additionally, the present data show that a reduced rate of cognitive decline requires CFLN treatment for some duration, such as a few years. This reduced rate of cognitive decline is roughly proportional to the reduction of homocysteine from its baseline value. Milder severity HYH patients showed greater benefit of CFLN treatment in slowing cognitive decline. CFLN contains distinct bioactive forms of B vitamins in addition to N-acetyl-cysteine, which may contribute to the HYH treatment effect in ADRD patients. Discrepancies among studies of HYH-lowering with B vitamins may be due to exogeneous factors known to influence HYH metabolism that have typically not been controlled. Regardless, CFLN circumvents known metabolic polymorphisms that commonly used B vitamins do not. By passing these metabolic roadblocks may confer greater benefits in managing HYH and its downstream pathophysiology. The present study's results warrant further investigation of CFLN as a useful treatment in ADRD patients with HYH.

Conflict of Interest Statement:

Drs. Shankle and Hara are employees and shareholders of Medical Care Corporation whose assessment tool (MCI Screen) was used in this study. Mrs. Barrentine and Mrs. Curole are employees of Nestlé Health Science – Pamlab, Inc.

Author contributions

The authors had full access to all study data and had uniformly agreed to submit this paper for publication. Drs. Shankle and Hara have equally contributed to the study design and conceptualization along with Mrs. Barrentine and Mrs. Curole, the development of the manuscript for intellectual content. Dr. Shankle conducted statistical analyses for the study, and Dr. Hara confirmed the results.

Sponsor and Sponsor's Role:

This study was sponsored by Nestle Health Science - Pamlab, Inc. The sponsor contributed to the study design (Mrs. Barrentine and Mrs. Curole) but had no role in data collection, analysis or manuscript preparation.

Ethical Standards

This study was conducted in accordance with the International Conference on Harmonization Good Clinical Practice E6 standards and in compliance with all applicable federal, state, and local regulatory requirements.