international society for biological and environmental

International Society for Biological and

Environmental Repositories (ISBER) Regional Meeting

Minneapolis, USA

November 3-5, 2019

The RAMS Registry/Repository: Biobanking for

Microbiome Research in Women’s Reproductive

Health and Pregnancy

Gregory A. Buck

Virginia Commonwealth University, Richmond, Virginia

The Vaginal Microbiome Consortium at VCU

NIH (NHGRI/NIAID) 1UH2/UH3AI083263 (NIH Human Microbiome Project)NIH (NHGRI/NICHD) 8U54HD080784 (NIH Human Microbiome Project)

NIH(NICHD) 1R01HD092415NIH(NIMHD) 1R01MD011504

GAPPS/Gates PPM Grant 15011(With additional thanks to the Office of Research on Women’s Health, The Bill and Melinda Gates

Foundation, and the Global Alliance to Prevent Prematurity and Stillbirth)

Conflict of Interest Disclosure

Gregory A. Buck, Ph.D.

• Scientific Advisory Board for startup: Juno Bio, Ltd

– Kemp House, City Rd, Old Street, London EC1V 2NX, UK

– 2019-present

– No salary or fees

Image source: http://commonfund.nih.gov/hmp

~10 years ago: NIH Launches the Human Microbiome

Project (2 phases)

Objective: use high throughput genomic technologies to define the human microbiome:

from cradle to grave (”normal”)in health and diseaseIn all body niches



Project (2 phases)

Phase 1: split into two phases:Phase 1A: define the ‘healthy’ human microbiome.

What are the microbes that inhabit all human body niches?Phase 1B: define the ‘abnormal’ human microbiome.

What are the microbes in the human body that one finds in disease?

(Canada)

HMP 1 (2007-2013)• ~40 projects• ~20 institutions



Project (2 phases)

Phase 2: Multi OmicsObjective: apply massively multi omic data intensive strategies to dissect the impact of the microbiome on a targeted disease state

HMP 2 (2013-present)• 3 collaborative projects• Massive increase in samples• Massive increase in data

VCUGAPPS

Stanford/Jax

Harvard/Broad

NIH Creates Human Microbiome ProjectData Coordination Center (DACC)

(https://hmpdacc.org/)

NIH invests millions in the HMP Data Coordination Center (U. Maryland)

The DACC “provides a commonrepository for diverse humanmicrobiome datasets and minimumreporting standards established by theDCC, from both HMP1 and the secondphase of the project, iHMP, providingresearchers with the ability to queryand retrieve metagenomic,metatranscriptomic, human genetic,microbial culture, and many other datatypes from each project.”

But, no accommodation for the massive number of samples being collected in the large scale longitudinal studies mandated by the program.

VCU Alone generated over a quarter million samples!

1. The Vaginal Human Microbiome Project (VaHMP):

• A cross-sectional study

• Samples from 6,063 visits; ~4500 women, ~1000 pregnant

>60,000 vaginal, cervical, buccal, other samples to archive.

2. Multi Omic Microbiome Study: Pregnancy Initiative (MOMS PI):

• A longitudinal study of ~1500 pregnancies (1-10 visits each)

• ~1500+ neonates (delivery, discharge, follow-up)

• Yielded 45 Preterms that met criteria (spontaneous; <37 weeks)

• >200,000 samples (vaginal, buccal, rectal, blood, birth products)

Over 250,000 irreplaceable samples generated.

• An extremely valuable resource

• Not only keep them alive, but keep them stable

• Bacteria and Cells

• DNA and RNA

• Cytokines and metabolytes/lipids

Samples collected at VCU>250,000 collected/

~50,000 processed/analyzed

WMTS WMGS

rRNA

MetaLipids

Nostril

Cheek

Elbow

Chest

Palm

Rectal

Stool

Vaginal

Cervical

Blood

Human

DNA Cytokine Bacteria

2K

2K

2K

2K

2K

10K

2K

1K

1K

1K

1K

1K

1K

2K2K

M Bmom baby

30,00012,000 120120908,0002,000

2K

6K

Cord Bld

Amniot fl

Umb cord

Placenta

Membranes

2K

10K

Each sample /stabilate was stored in a special medium or buffer.Each was tested and validated for reproducibility after storage for extensive periods.

Human cells & 16S rRNADNA cytokine bacteria metabolites MTS MGS Mom Baby

To manage this onslaught of samples/data:

We created a Microbiome Registry/Repository at VCU: the Research Alliance for Microbiome Science Registry

(the RAMS Registry)

Heart of the RAMS Registry: Freezer Farm

Started buying freezers

When we run out of room:we purchase another

Bootstrapped a monitoring system

One ‘extra’ freezer for emergency backup

Supported by NIH through the project grant

Total now: ~12 freezers(no more space in farm!)

Initial Visit

• New OB kit – 28 samples each

• Health History Survey– 451 fields

Interim Prenatal

• Interim kit (2-10/subject) – 26 samples each

• Review of Systems Survey – 173 fields

L&D Triage

• Triage kit – 22 samples each

• Review of Systems Survey – 173 fields

Delivery

• Superkit (5 sub-kits) – 68 total samples each

• Delivery & Discharge Survey – 88 fields

Post-Partum

• Follow-Up kit – 18 samples each

• Follow-Up Survey – 143 fields

Clinical and Survey Data

is part of the RAMS Registry

Plus: Medical Record Abstraction: one for each of ~6,000 subjects

Electronic Data Capture (LIMS)

Cerner EMR

Clinical DB(Contact,Consent, IDX,

Survey)

Participant Assignment,Temp Survey

Research Coordinator Tablet

Participant Tablet

Clinical App (Django, Apache)

P App (Django, Apache)

LimeSurvey App (PHP, Apache)

Momspi.vmc.vcu.edu

Technical Staff

VCUHS IDX

Lab App (Django, Apache)

Tracking DB(Auth, Participant Mapping , Sample

Tracking,Results etc)

Moms-survey.vmc.vcu.edu

Smomspi.vmc.vcu.edu

MOMS-PI Sample/Data

Management & Tracking System

Clinical Metadata System

(Tablet Interface)

GAPPSVCU

Clinics

Sample Collection Data Analysis Results

Sample

Tracking

System

MOMS-PI Database

Query

Portal

Browse

Portal

Bulk

Download-Raw Data

-Analysis Files

Web PortalControlled Access to VMC members and NIH approved teams

16S Gene

Metagenomics

WMGS

Immunomics

Metabolomics

RDP Classifier

STIRRUPS

Metaphlyn,

HUMAnN

Microbial Taxonomic

Profiles

WMGS Community

Profiles

Meta

Transcriptomics

File Storage

Database

APIs

MOMS-PI

Data Management System

Raw Data

Interactomics

Results File

Parsers

Software

pipeline

Gene Expression

Profils

Cytokine Profiles

Lipid Profiles

Protein InteractionsInteractomics

software pipeline

Bacterial

Sequences

Velvet, CLC Bio

assembler

Bacterial Genome

Assemblies

16S, WMGS,

Meta Transcriptomics

Clinical Metadata

Interaction ResultsEBI IntactEBI IntAct

Bacterial Genome

Assemblies

External

Upload

System

Public Portal-Protocols

-Info for

participantsInput/OutputExisting

Lab

Sample

Processing

EBI IntactNCBI SRA

EBI IntactNCBI dbGap

EBI IntactNCBI Genbank

New

Fettweis JM, et al. Nature Med (2019)

Now holds over 250,000 samples from VaHMP (HMP1) and MOMS PI (HMP2)clinical survey data & medical record abstractions from over 6,000 participantsmulti omic data from ~50,000 samples

Bacterial Vaginosis (dysbiosis of the vaginal microbiome)

– ~29% (23-51%) prevalence in women (CDC).• Decreased Lactobacillus, increase in ‘unhealhy anaerobes’

• ‘Clue cells’: vaginal epithelial cells coated with bacterial biofilm

• Symptoms include thin, grayish discharge and vaginal odor

– Often recurrent and non-responsive to treatment

– Increased risk of STI, HIV, trichomoniasis, etc.

– Increased risk of preterm birth

Preterm birth:

– 1 in 10 births in the USA are preterm (<37 weeks gestation)• As high as 1 in 6 in some populations

– Annually over 12 million preterm births world wide (2005 WHO)• Mortality rate as high as 42%.

• Survivors have short and long term health issues

– Over $26 billion/yr in US for prematurity associated healthcare.

– 30-40% caused by poorly defined bacterial agents.• Microbial etiology is not clear (which bacteria, mechanism, host factors…)

What have we learned?

Why are we interested in the

vaginal microbiome?--a clear role in women’s urogenital health --

Healthy

Bacterial vaginosis

Adapted from: Jim Litch, GAPPS<10%

10-15%

>15%

Not available

Not applicable

Preterm rate, 2010

A global challenge… ~1 of 10 neonates worldwide.

Over 12 million annually

Preterm birth rates worldwide(ranges from <10% to over 15% of live births)

Gallo Images

Coral reef….

Gut microbiome

Oral Microbiome high diversity is healthy

Skin microbiome

Vaginal microbiome low diversity is healthy

Complex ecological environment

In contrast to gut and oral microbiomes, health

of the vaginal microbiome is defined differently.

Wheatfield

monoculture (low diversity)

Dogma: Lactobacillus dominated vaginal microbiome is healthy;Diverse vaginal microbiome considered less healthy.

Population matters…

HMP Consortium Healthy Cohort (143 healthy young women)

143 samples

Mostly Lactobacillus:Very healthy (carefully screened)

Young (college age)Mostly of European ancestry (Caucasian)

Lactobacillus sp.

Impacts of the Vaginal Microbiome in Disease and PregnancyBernice Huang, Victoria Pokhilko, Gregory Buck, and

The Vaginal Microbiome Consortium

Analysis of Vaginal Microbiome Profiles of 4,891 Women

Departments of Microbiology & Immunology, Statistical Sciences and Operations Research and Center for the Study of Biological Complexity.

Virginia Commonwealth University, Richmond, VA

Figure 1 Demographics of 4891 women in the Vahmp Cohort.

Figure 2 To understand the broad system level dynamics of the

vaginal microbiota over a woman’s life, we characterized the

alpha diversity of 2,930 women; 1,851 healthy, 376 with BV, 580

pregnant healthy, 49 pregnant with BV, and 52 postmenopausal

receiving hormone replacement therapy (HRT). Women were

considered ‘healthy’ at the time of the visit if the purpose of the

visit was for an annual examination, they received no diagnosis

and were asymptomatic. BV testing by Amsel’s criteria was

performed only when indicated.

Contact: [email protected]

Participant Demographics

• The vaginal microbiome is not homogeneous.

• Vaginal microbiome profiles cluster reproducibly (vagitypes).

• A homogeneous microbiome profile is not always “healthy”.

• The vaginal microbiome is less complex in pregnancy.

• Diversity remains stable in reproductive aged women.

results summary

The Multi-Omic Microbiome Study-Pregnancy Initiative (MOMS-

PI) is a collaborative project with the Global Alliance to Prevent

Prematurity and Stillbirth (GAPPS) based at Seattle Children's

and the Research Alliance for Microbiome Science (RAMS)

Registry at VCU. The research is funded by the NIH Roadmap

Human Microbiome Project (HMP) Phase 2: The Integrative

H u m a n M i c r o b i o m e P r o j e c t ( i H M P ) , N I H g r a n t

(NICHD)8U54HD080784, with support from NICHD, NHGRI,

ORWH, and NCCIH. We acknowledge the other members of the

Vaginal Microbiome Consortium who contributed to this work.

Vaginal complaints are one of the most common

reasons for women seeking medical attention

worldwide and cl in ical symptoms are often

misdiagnosed. New insights into the diversity of

human microbial flora, including vaginal flora, have

been made possible through DNA sequencing via

culture-independent approaches. As part of the

Vaginal Human Microbiome Project at Virginia

Commonwealth University (Vahmp), we are sampling

thousands of women from the clinical setting to

explore the relationship of the vaginal microbiome with

various physiological and infectious conditions. Here,

we report an analysis of vaginal microbiome profiles of

4,891 women using deep sequencing of 16S rDNA

partial sequences. Species-level analysis of vaginal

microbiome profiles reveals how the samples further

cluster into distinct ‘vagitypes,’ driven by the

predominant bacterial species in the sample. The

vagitypes include several Lactobacillus types (L.

crispatus, L. iners, L. gasseri, L. jensenii, L.

delbruecki), a Gardnerella vaginalis type, a type

dominated by bacterial vaginosis-associated bacterium

(BVAB1), and more than fifty additional rare and minor

vagitypes. A complete species-level analysis reveals

more than 200 species-level taxa in the vaginal

microbiome including ~35 novel or uncharacterized

taxa.

26%

Caucasian

62%

African

American

9%

Other

2%

Multiple

1%

Asian

Figure 3 16S rRNA gene sequencing and

species-level taxonomic resolution shows

an unexpectedly diverse spectrum of

vaginal profiles. These profiles were

grouped into “vagitypes” according to

predominating bacterial species, where a

predominating species was defined as

>30%. The major vagitypes include those

predominated by L. crispatus, L. jensenii, L.

gasseri, L. iners, Gardnerella vaginalis or

bacterial vaginosis-associated bacterium-1

(BVAB1). A number of minor vagitypes are

observed and are described in Table 1.

Vagitypes are consistent with hierarchical

clustering results but cluster more

reproducibly. In the absence of a large

sample size, hierarchical clustering fails to

identify minor vagitypes of importance.

Vagitype %

L. iners 33.3

G. vaginalis 20.3

L. crispatus 17.02

BVAB1 10.37

S. amnii 2.4

L. gasseri 2.9

Prevotella cluster2 1.23

L. jensenii 1.8

P. bivia 1

Atopobium vaginae 0.94

S. agalactiae 0.8

M. girerdii 0.79

Prevotella amnii 0.61

S. anginosus 0.47

TM7 OUT-H1 0.71

M. hominis 0.5

No predominant Taxon 4.83

Majo

r V

ag

ity

pe

Min

or

Vag

ity

pe

Taxonomic Identification: 16S rRNA vs WMGS

Figure 4 A. Comparison of 16S rRNA V1-V3 sequencing (top) and whole-genome shotgun

sequencing (WMGS, bottom) taxonomic identification for 45 participants with clinically

diagnosed BV. Bacterial species were identified with the MetaphlAn2 package. Note that the

MetaPhlAn reference database used for bacterial classification of WMGS reads, does not

include S. amnii, BVAB1, or M. girerdii and as a result cannot assign those genera. B. We

measured the similarity of 16S and WMGS taxa profiles by calculating pearson correlations

across the relative abundance of taxa detected by both technologies. Among the 42 samples

with at least 4 taxa detected by both technologies, 86% exhibited 16S vs WMGS profile

correlations exceeding 0.8. See Table 1 for taxa color legend.

Study Design: This was a study to characterize the

role of the vaginal microbiota in health and disease.

Women were recruited from outpatient clinics at the

Virginia Commonwealth University (VCU) Medical

Center and the Virginia Department of Health from

2009-2013 following written, informed consent.

Inclusion criteria included women age 18 and older.

This study was approved by the Institutional Review

Board at VCU.

Sampling and Sample Processing: During a

speculum examination, multiple cervical, mid-vaginal,

introital, perianal and buccal samples were collected

using CultureSwab EZ (Becton Dickinson). DNA was

extracted from the swabs within 4 h of collection using

the Powersoil kit (MoBio). The swabs were swirled

directly in the Powerbead tubes and processed

according to the manufacturer’s instructions.

Bacterial 16S rRNA Sequencing: Extracted DNA was

amplified with the 16S rRNA gene V1-V3 primers. PCR

products were sequenced using the Roche 454 GS

FLEX Titanium platform.

Whole Metagenome Shotgun Sequencing: Shotgun

DNA libraries were made from total nucleic acid

isolated from 45 vaginal swab samples isolated from

women with clinically diagnosed BV. DNA libraries

were prepared with the KAPA Library Prep Kit and

sequenced on an Illumina HiSeq2500. Bacterial

species were identified with the MetaPhlAn2 package.

Table 1. Vagitypes

BA

Alpha Diversity Across Age

abstract

materials & methods

1

1

2

2 3

3

pro

port

ion

Populations matter!!!

More realistic…

Jacques Ravel et al. PNAS 2011

Five community state types

G.

vaginalis

BVAB1L. jensenii.

L.gasseri

L.inersL.crispatus

Impacts of the Vaginal Microbiome in Disease and PregnancyBernice Huang, Victoria Pokhilko, Gregory Buck, and

The Vaginal Microbiome Consortium

Analysis of Vaginal Microbiome Profiles of 4,891 Women

Departments of Microbiology & Immunology, Statistical Sciences and Operations Research and Center for the Study of Biological Complexity.

Virginia Commonwealth University, Richmond, VA

Figure 1 Demographics of 4891 women in the Vahmp Cohort.

Figure 2 To understand the broad system level dynamics of the

vaginal microbiota over a woman’s life, we characterized the

alpha diversity of 2,930 women; 1,851 healthy, 376 with BV, 580

pregnant healthy, 49 pregnant with BV, and 52 postmenopausal

receiving hormone replacement therapy (HRT). Women were

considered ‘healthy’ at the time of the visit if the purpose of the

visit was for an annual examination, they received no diagnosis

and were asymptomatic. BV testing by Amsel’s criteria was

performed only when indicated.

Contact: [email protected]

Participant Demographics

• The vaginal microbiome is not homogeneous.

• Vaginal microbiome profiles cluster reproducibly (vagitypes).

• A homogeneous microbiome profile is not always “healthy”.

• The vaginal microbiome is less complex in pregnancy.

• Diversity remains stable in reproductive aged women.

results summary

The Multi-Omic Microbiome Study-Pregnancy Initiative (MOMS-

PI) is a collaborative project with the Global Alliance to Prevent

Prematurity and Stillbirth (GAPPS) based at Seattle Children's

and the Research Alliance for Microbiome Science (RAMS)

Registry at VCU. The research is funded by the NIH Roadmap

Human Microbiome Project (HMP) Phase 2: The Integrative

H u m a n M i c r o b i o m e P r o j e c t ( i H M P ) , N I H g r a n t

(NICHD)8U54HD080784, with support from NICHD, NHGRI,

ORWH, and NCCIH. We acknowledge the other members of the

Vaginal Microbiome Consortium who contributed to this work.

Vaginal complaints are one of the most common

reasons for women seeking medical attention

worldwide and cl in ical symptoms are often

misdiagnosed. New insights into the diversity of

human microbial flora, including vaginal flora, have

been made possible through DNA sequencing via

culture-independent approaches. As part of the

Vaginal Human Microbiome Project at Virginia

Commonwealth University (Vahmp), we are sampling

thousands of women from the clinical setting to

explore the relationship of the vaginal microbiome with

various physiological and infectious conditions. Here,

we report an analysis of vaginal microbiome profiles of

4,891 women using deep sequencing of 16S rDNA

partial sequences. Species-level analysis of vaginal

microbiome profiles reveals how the samples further

cluster into distinct ‘vagitypes,’ driven by the

predominant bacterial species in the sample. The

vagitypes include several Lactobacillus types (L.

crispatus, L. iners, L. gasseri, L. jensenii, L.

delbruecki), a Gardnerella vaginalis type, a type

dominated by bacterial vaginosis-associated bacterium

(BVAB1), and more than fifty additional rare and minor

vagitypes. A complete species-level analysis reveals

more than 200 species-level taxa in the vaginal

microbiome including ~35 novel or uncharacterized

taxa.

26%

Caucasian

62%

African

American

9%

Other

2%

Multiple

1%

Asian

Figure 3 16S rRNA gene sequencing and

species-level taxonomic resolution shows

an unexpectedly diverse spectrum of

vaginal profiles. These profiles were

grouped into “vagitypes” according to

predominating bacterial species, where a

predominating species was defined as

>30%. The major vagitypes include those

predominated by L. crispatus, L. jensenii, L.

gasseri, L. iners, Gardnerella vaginalis or

bacterial vaginosis-associated bacterium-1

(BVAB1). A number of minor vagitypes are

observed and are described in Table 1.

Vagitypes are consistent with hierarchical

clustering results but cluster more

reproducibly. In the absence of a large

sample size, hierarchical clustering fails to

identify minor vagitypes of importance.

Vagitype %

L. iners 33.3

G. vaginalis 20.3

L. crispatus 17.02

BVAB1 10.37

S. amnii 2.4

L. gasseri 2.9

Prevotella cluster2 1.23

L. jensenii 1.8

P. bivia 1

Atopobium vaginae 0.94

S. agalactiae 0.8

M. girerdii 0.79

Prevotella amnii 0.61

S. anginosus 0.47

TM7 OUT-H1 0.71

M. hominis 0.5

No predominant Taxon 4.83

Majo

r V

ag

ity

pe

Min

or

Vag

ity

pe

Taxonomic Identification: 16S rRNA vs WMGS

Figure 4 A. Comparison of 16S rRNA V1-V3 sequencing (top) and whole-genome shotgun

sequencing (WMGS, bottom) taxonomic identification for 45 participants with clinically

diagnosed BV. Bacterial species were identified with the MetaphlAn2 package. Note that the

MetaPhlAn reference database used for bacterial classification of WMGS reads, does not

include S. amnii, BVAB1, or M. girerdii and as a result cannot assign those genera. B. We

measured the similarity of 16S and WMGS taxa profiles by calculating pearson correlations

across the relative abundance of taxa detected by both technologies. Among the 42 samples

with at least 4 taxa detected by both technologies, 86% exhibited 16S vs WMGS profile

correlations exceeding 0.8. See Table 1 for taxa color legend.

Study Design: This was a study to characterize the

role of the vaginal microbiota in health and disease.

Women were recruited from outpatient clinics at the

Virginia Commonwealth University (VCU) Medical

Center and the Virginia Department of Health from

2009-2013 following written, informed consent.

Inclusion criteria included women age 18 and older.

This study was approved by the Institutional Review

Board at VCU.

Sampling and Sample Processing: During a

speculum examination, multiple cervical, mid-vaginal,

introital, perianal and buccal samples were collected

using CultureSwab EZ (Becton Dickinson). DNA was

extracted from the swabs within 4 h of collection using

the Powersoil kit (MoBio). The swabs were swirled

directly in the Powerbead tubes and processed

according to the manufacturer’s instructions.

Bacterial 16S rRNA Sequencing: Extracted DNA was

amplified with the 16S rRNA gene V1-V3 primers. PCR

products were sequenced using the Roche 454 GS

FLEX Titanium platform.

Whole Metagenome Shotgun Sequencing: Shotgun

DNA libraries were made from total nucleic acid

isolated from 45 vaginal swab samples isolated from

women with clinically diagnosed BV. DNA libraries

were prepared with the KAPA Library Prep Kit and

sequenced on an Illumina HiSeq2500. Bacterial

species were identified with the MetaPhlAn2 package.

Table 1. Vagitypes

BA

Alpha Diversity Across Age

abstract

materials & methods

1

1

2

2 3

3

More recently:• analysis of more samples (e.g., VaHMP)

• more community state types (vagitypes)

• ~10 major, more minor

Surprisingly, there is not a very good

correlation with health:

• Some complex vagitypes

in apparently healthy women

• Some homogeneous vagitypes

in women with symptoms

What happens to the vaginal

microbiome during pregnancy?

Non-Pregnant

~2,935 women

VaHMPCohort

Pregnant

~880 women

• increase in prevalence of Lactobacilli (p < 0.05) (mostly L. iners)

• decrease in prevalence of G. vaginalis (p < 0.05) and other vagitypes (e.g., Sneathia, Atopobium, etc.)

• No decrease in BVAB1

Pregnancy drives a ‘healthier’ (more Lactobacillus dominant microbiome?

Lactobacillus clusters

Aagaard et al. Plos One (2011)

Walther-Antonio et al. Plos One (2014)

Romero et al. Microbiome (2014)

MacIntyre et al. Sci Rep (2015)

Freitas et al. Sci Rep (2017)

Lower numbers (less significance)

Focused cohort (e.g., Caucasian only)

Targeted age group (college age)

This cohort: take all comers (no case-matching, preterm birth, etc., not excluded)

Lactobacillus clusters

Gardnerella vaginalis

G. vaginalis

BVAB1

BVAB1

What specific taxa are altered in

abundance in pregnancy?

~10 taxa affected (q<0.05)

Increased:

L. iners

At the expense of (decreased):

A. vaginalis

G. vaginalis

Sneathia sp.

Prevotella sp.

Dialister sp

BVAB2

Note: BVAB1 not decreased

Mann-Whitney U

Serrano, Parikh et al. Nat Med (2019)

156

Pregnant

Women

African

American

156

Nonpregnant

Women

61

Pregnant

Women

Non-African

American

61

Nonpregnant

Women

Changing pregnancy microbiome is driven

by changes in Pregnant African- Americans

Non-African-American:

• proportion with Lactobacillus dominated microbiome unchanged.

• Decreased L. crispatus, increased L. iners (p<0.05)

African-Americans:

• increased Lactobacillus (mostly L. iners) in pregnancy (p<0.05).

• Less G. vaginalis and other vagitypes (p<0.05).

In terms of total Lactobacillus, pregnant African American women:

profiles begin to resemble non-African profiles

Vaginal microbiome: differences based on populations

• Racial, socioeconomic, other environmental factors


African American Other

49

Longitudinal

Full Term

Pregnancies

African-American

These results show:

1. The vaginal microbiome transitions to a less complex, Lactobacillus-dominated profile during pregnancy.

2. The bulk of these changes occur early in pregnancy and are generally complete by the 2nd trimester.

3. The changes are more prevalent in our cohort of women of African ancestry (not shown).

Vagitype transitions tend to occur relatively early in pregnancy.


What about Preterm Birth?

45 preterm births (from ~1000 pregnancies)Only spontaneous preterm births included90 case matched (age, race, SES) controls

Clear differences in PTB samples (q < 0.05):less: L. crispatusmore: BVAB1/2,

S. amnii/sanguinegins, Dialister spPrevotella sp

No diff: Gardnerella vaginalisFettweis et al. Nat Med (2019)

Term Birth

Preterm Birth

Va

gin

al 1

6S

rR

NA

ab

un

da

nce

BVA

B1

Pcl2

Mty1

Sam

n

Dcl51

Pam

n

BVA

B2

TM7-

H1

Dm

ic

P14

2

0

10-2

10-1

1

Gesta

tio

na

l ag

e a

t d

eliv

ery

(da

ys)

PTB predictive score

140

160

180

200

220

240

260

280

300

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5


Gesta

tion

al a

ge

at d

eliv

ery

(da

ys)

0.0 0.1 0.2 0.3 0.4 0.5270

280

290

300

310

320

330

340

350

a b

c

Term Birth

Preterm Birth

Va

gin

al 1

6S

rR

NA

ab

un

da

nce

BVA

B1

Pcl2

Mty1

Sam

n

Dcl51

Pam

n

BVA

B2

TM7-

H1

Dm

ic

P14

2

0

10-2

10-1

1

Gesta

tion

al a

ge

at d

eliv

ery

(da

ys)


140

160

180

200

220

240

260

280

300

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5


Ge

sta

tion

al a

ge

at

de

live

ry (

days)

0.0 0.1 0.2 0.3 0.4 0.5270

280

290

300

310

320

330

340

350

a b

c False positives: those falsely

predicted to experience PTB

False negatives: those falsely predicted to not experience PTB

(need to minimize)

Model using microbiome data alone

predicts risk of preterm birth

Improve with:

Clinical

Demographics (race, SES)

Microbiome

Cytokine

Proteome

Metabolome

Lipidome

~77% sensitivity and ~76% specificity (slightly better than current clinical parameters)

Promise for better predictive measures to permit early intervention.

a b

Term Birth Cohort Preterm Birth Cohort

Strong negative correlation between L. crispatus and pro-inflammatory cytokines and taxa associated with dysbiosis and PTB.

L. iners associated with IP10 (induces chemotaxis of immune cells and is considered to be pro-inflammatory). (Jespers et al. 2017)

In women who would experience PTB, dysbiotic taxa and pro-inflammatory cytokines form a tighter cluster.

IP10 and L. iners no longer correlated

Preterm Birth group shows tight clustering of dysbiotic taxa and proinflammatory cytokines.

What about the host immune response?Integrative sparse canonical correlation analysis

(cytokines vs taxa)

What do we believe now?

1. Pregnancy tends to homogenize the vaginal microbiome toward a moreLactobacillus-dominant composition. This is more evident in women ofcertain racioethnic groups. Precision microbiome: populationdependent.

2. This homogenization reduces the prevalence of a finite panel of bacterialtaxa that are often associated with dysbiosis.

3. This homogenization occurs relatively early in pregnancy.

4. An overlapping panel of anaerobic bacteria are associated with (andpossibly causative of) preterm birth. Again, this is more evident inwomen of certain racioethnic background.

5. Prevalence of these taxa generally decrease as pregnancy progresseseven in women who will experience preterm birth.

6. Early in pregnancy, pro-inflammatory cytokine expression is correlatedwith bacteria that are associated with preterm birth.

Can we improve on our accuracy of prediction of risk of preterm birth?Can we predict risk of preterm birth even prior to pregnancy?

Early in pregnancy, potentially pathogenic

bacteria are present in the vaginal microbiome

Model for how the vaginal microbiome could cause preterm birth

Complex pro-inflammatory microbiome early in pregnancyAs pregnancy progresses, Lactobacillus taxa are favoredHowever, some pro-inflammatory bacteria persist, or ascend.Infection or inflammation causes labor and premature birth.

1. RAMS Registry: an important asset to our efforts:

• Incredible investment of time and effort.

• Establishment, validation, digital investment, longevity

• Many places to ‘go wrong’.

• Limited financial support (what happens when grants end?).

In summary:

2. That said, it has largely worked for us:

• Over 250k samples stored in validated protocols

• ~50,000 samples processed

• Digital tracking of sample processing and storage works

• Published many manuscripts

• Succeeded in new funding to continue our work.

Thank you!

The Vaginal Microbiome Consortium at VCU

NIH

Lita Proctor (NHGRI)Hagit David (NIAID)John Ilekis (NICHD) Mary Perry (NIH Common Fund)

VCU Team

Kim JeffersonJerry StraussJen FettweisMyrna SerranoJ. Paul BrooksDavid EdwardsTom ArodzPhilippe GirerdK. Hendriks-MuñozBernice HuangSophonie JeanAndrey MatveyevAbbie GlascockVishal KopardeHardik ParikhNihar ShethStephanie VivadelliOthers…..

GAPPS Team

Mike GravettAmber SextonCraig RubensOthers….

NIH (NHGRI/NIAID) UH3AI083263

(NHGRI/NICHD) U54HD080784

The Gates Foundation

Global Alliance to Prevent Prematurity & Stillbirth

The ~8,000 women who have generously enrolled in our studies

international society for biological and environmental

Documents