Targeting the Microbiome to Improve Clinical Outcomes in Bipolar Disorder

RECRUITING AND INCENTIVIZING RESEARCH PARTICIPANTS

For the proposed study, the investigators will recruit research participants from the
Prechter Longitudinal Study of Bipolar Disorder, housed at the University of Michigan under
the direction of Dr. Melvin McInnis, mentor on this application. The Prechter study is in its
10th year with over 1200 participants, including psychiatric and healthy control individuals.
the investigators have successfully recruited from this study in the past to perform an
observational dietary study and single point microbiomic analysis described in the Previous
Work section. The investigators will leverage the infrastructure of the study with 50%
dedicated time for a research assistant to manage recruitment, scheduling, follow up and data
entry duties. The investigators will also leverage 5% dedicated time for the study
coordinator, Gloria Harrington, who oversees all administrative aspects of the Prechter
Study, including assurance of regulatory compliance.

The investigators will recruit 20 males and 20 females from the Prechter cohort, aged 25-65
with a confirmed diagnosis of Bipolar I disorder from a diagnostic interview for genetic
studies (DIGS), used as the baseline intake instrument for the Prechter study. Eligible
subjects will be consented and scheduled for an intake interview and assigned a unique
identification number for tracking with a secure laboratory information management system.
Subjects will be offered a monetary incentive of $200, paid in installments after completion
of study milestones activities.

DETERMINING HABITUAL DIETS

The investigators will determine the habitual baseline diets of all subjects using a 3 non
consecutive days over 1 week with an automated online 24-hour diet recall (ASA24 from the
National Cancer Institute). Subjects will meet with a certified nutritionist from the
Nutrition Assessment Laboratory in the Nutrition and Obesity Research Center at the
University of Michigan during the consent visit to explain how to use the recall
appropriately. After 1 week, subjects will return for to the MCRU for baseline biological
samples and dietary guidance as detailed below.

COLLECTING BIOLOGICAL SAMPLES, MORPHOMETRIC DATA AND STANDARD HEALTH MEASURES

During the baseline visit (following the week involving dietary recalls), all subjects will
donate baseline fasted blood samples for free fatty acid analysis; a urine sample for
analysis of oxidative stress markers; and have their height, weight, waist and hip
circumference, and blood pressure recorded and be asked to complete the interviewer guided
Hamilton Depression scale questionnaire, the Young Mania Rating Scale questionnaire and the
Global Physical Activity Questionnaire. Participants will also complete a brief 25-35 minute
neuropsychological test battery during their baseline and final visit. Subjects will also
return a stool sample, which will be collected at home with a supplied collection kit.
Subjects will repeat all procedures at the end of the 1st and 4th week of the dietary
protocol. These procedures as well as the dietary interviews will occur in the Michigan
Clinical Research Unit at the University of Michigan, staffed with phlebotomists, nurses,
nutritionists and technicians with expertise in all proposed procedures; and equipment needed
to process the samples. Finally, subjects will be asked about all prescription and
non-prescription medications taken over the previous week.

DIETARY INTERVENTION

For the diet intervention, subjects will be guided by a certified nutritionist to achieve a
target macronutrient composition (by total calorie contribution) of 15% protein, 45%
carbohydrate, 40% fat, with <10% saturated fat, ~15% PUFA and ~15% monounsaturated fatty
acids (MUFA). Specific targeted profiles are given in Table 4. For reference, The
investigators also show the cognate values attained from a 7-day observational study of 91
individuals with bipolar disorder described above. Comparing the target and previously
determined habitual values from the same population, the major change to the diet will be a
near doubling of PUFA intake and a small increase in MUFA intake, at the expense of reducing
carbohydrates and saturated fat, which is a realistic and achievable goal.

The investigators will achieve the diet with specific menus, considering personal preferences
gleaned from the habitual diet analysis described. The investigators will provide subjects
with specific cooking oils for addition to foods prepared at home, to achieve the oleic
(OLA), LA and α-linolenic (LNA) targets; and fish-oil supplements to achieve the DHA and EPA
targets. Subjects will also be provided a list of foods to eat and avoid (tailored from their
habitual diet records), as well as several menu items from chain restaurants to eat for meals
unavoidably eaten away from home. Subjects will be trained to report on their diets using the
ASA24. Records will be reviewed at least weekly by research staff and assessed for deviation
from the diet outside of threshold ranges (as average daily intake) of at least 12% PUFA and
no more than 50% carbohydrate. In parallel to this project the investigators will develop
python scripts to extract data to automate the process, however, the investigators will not
depend on success of that goal in order to meet the objectives of this project. Values
falling outside the set thresholds will be brought to the subjects' attention with
suggestions for correction. Importantly, since plasma LA correlates strongly with dietary
intake, objective measures of dietary adherence to PUFA intake will be available upon
completion of free fatty acid analysis from fasting blood samples, which can be leveraged as
covariates in statistical models. Finally, the investigators will avoid November - January
for the dietary interventions, as diets tend to be non-habitual during this period.

Following the 4-week dietary manipulation, subjects will be encouraged to maintain the high
PUFA diet for 6 months. Due to limited resources, The investigators will not continue to
collect stool samples following the 4-week period (although other funding will be pursued to
do so). However, the investigators will continue to encourage subjects to use the ASA24
weekly. The investigators will also continue to monitor self-report measures as a component
of the longitudinal study, requiring no additional resources from this application.

Dr.Evans (Principal Investigator) has established a good relationship with the research staff
in the Nutrition Assessment Laboratory (NAL) during past studies, described above, and has
developed the plans for the proposed dietary manipulation with Theresa Han-Markey, RD,
Bionutrition Manager of the NAL, who's services are budgeted into the proposed study. Dr.
Evans has also recently completed formal training in nutritional sciences as a component of a
K01 career development award and will be intimately involved in the dietary studies.
Furthermore, Dr. Charles Burant, Director of the Nutrition and Obesity Research Center and
previous mentor for Dr. Evans K01 award, will consult on these studies (see letter of
support).

INTERROGATING THE TAXONOMICAL COMPOSITION OF THE STOOL MICROBIOME

Home OmniGene Gut stool collection kits from DNA Genotek (Ottawa, CA) will be used to collect
samples at baseline, week 1 and week 4 of the dietary period. These kits were used
successfully to generate preliminary data describe above, and have the advantage of
maintaining sample DNA stability for 14 days without freezing. Subjects will return the
completed kits at the scheduled study visits. Methods for identifying microbial complements
in the samples have been previously described in a publication by the HMI Laboratory at the
University of Michigan. Briefly, microbial DNA will be extracted from stool samples and
processed for 16S ribosomal sequencing using a dual-indexing sequencing strategy. Sequencing
will be done on the Illumina MiSeq platform, according to the manufacturer's instructions.
Analysis of the 16S rRNA gene will be done using an SOP (www.mothur.org/wiki/MiSeq_SOP) to
process the MiSeq data. Sequence classifications will be determined by comparing sequences to
the Ribosomal Database Project and sequences will be clustered into OTUs.

COLLECTION OF SELF-REPORTED HEALTH MEASURES

Longitudinal participants currently complete self-rated questionnaires by mail every 2
months. For the current study the investigators use average responses previously collected as
baseline measures and new responses collected at the completion of the diet period as outcome
measures. These questionnaires include the PHQ-9, the PSQI, the LFQ, the Altman Self-Rating
Mania Scale (ASRM), the Short Form 12 Health Survey (SF-12), and the Generalized Anxiety
Disorder Assessment (GAD-7).

DATA ANALYSIS

Data analysis has 2 major phases. The first is generating normalized summary scores from the
various measures. The second is using statistical models to test for main effects of diet.
Summary of questionnaire scores and anthropomorphic measures are standard and not further
detailed here. Summary of microbiomic data sets will be done as described above and OTUs
output in 2 ways: 1) as a fractional representation table, where each OTU is expressed as a
fraction of the total OTUs per sample; and 2) as total sequence counts, where each OTU is
represented as total abundance. Both will be tested in statistical analyses.

For the second phase the investigators will use two main statistical models: 1) regression
analyses to test for main effects of diet on OTU level microbiome data and self-reported
burden of disease measures, adjusting for age, sex and BMI; and 2) structured equation
modeling analysis to test for mediation effects of OTUs on the relationship between diet and
clinical outcomes. The investigators will achieve the latter approach as diagrammed in Figure
2. Main effect of diet will be tested on plasma PUFA (primarily LA), microbiome OTUs, and
self-reported health measures, depicted by the dotted line. The cross-lag model, using SEM,
will test for statistically significant reciprocal effects, shown by the bold arrows. For
example, the investigators will test if increase in plasma LA accounts for the effect (fully
or partially) in changes in microbiome OTUs, and if changes in OTUs account for effects on
self-reported outcome measures.

Inclusion Criteria:

- Current participation in Prechter Longitudinal Study of Bipolar disorder at the
University of Michigan (HUM00000606)

- Confirmed diagnosis of Bipolar I or no diagnosis (controls)

Exclusion Criteria:

- Daily smoking habit

- Excessive alcohol defined as more than 2 drinks per day on average.

- An extremely poor diet determined by a nutritional interview as eating less than 1
planned meal per day.

- Food allergies or sensitivities.

- Anti-biotic use in the past 3 months.