Donor-recipient compatibility in HSCT

(Hematopoietic stem cell transplantation)

Guillaume DELEVOYE

Lab Meeting 15/01/19

PharmD thesis project

PH CABRERA Quentin & PH MOARS Catherine

You are unique

Are human parts interchangeable ?

Blood 1667 | 1903

Lungs 1963

Liver 1963

Kidneys 1954

Heart 1967

Intestine 1988

Pancreas 1966

Cornea 1905

Tendons 1910

Valves 1960

Veins 1869

Skin 1869

Bone marrow 1956

>> Yes !...

What are HSC in the bone marrow ?

(Hematopoietic Stem Cells)

 

One cell rebuilds them all...

The whole immune system and all blood cells are built from HSC

When  serious deregulation occurs

Death often follows

The immune system has almost no protection mechanism against itself: it's the last wall

HSCT

The main idea

1 ) Almost kill your patient by removing the immune system

 

2) Put back a new +/- similar immune system, hope it will work

 

3) Prey

 

 

While patient.is_alive or patient.is_sick:

The difference between HSCT and other transplantations

Every transplantation except HSCT

HSCT

Host Immune System (IS)

Attacks

Graft

Host' IS

Grafted IS

Host' organs

All weapons out

Immunosuppressant drugs

GVHD

The whole process in one picture

Our project

We would like to build predictive models of compatibility for clinicians or researchers

 

And THIS is our PROJEEEEEEEEECT

DELEVOYE Guillaume

PH CABRERA Quentin & PH MOARS Catherine

What already exists

Relevant and numerous ML tools exist for almost all forms of grafts

 

Except in HSCT >> Too complex

Just the 100 day overall mortality

Really a lot of censored data

Quick overview of what could be interesting to look at

 

 

 Different outcomes for different indications

• GRFS 100
• Long terme relapse-free for tumoral pathologies

Quantify the impact of protocols

More than a hundreds of protocols:

• Total ablations
• Partial removals
• Sometimes patients already had 1 or 2 other transplants

Main idea:

It's barbarous

And it's a mess !!

Prediction of opportunistic infections

allo-HSCT requires to:

• Exterminate the previous immune system
• Use massive doses of immunosuppressants so that the graft doesn't attack the host

--> Patients are cured... And then often die from infections

Prediction of  Graft versus host disease (GVHD)

Acute forms are lethal

... And chronic forms are also lethal

Dramatic incident

Really frequent

Prediction of Graft versus Tumor effect (GVTE)

Allografts: HLA-system

~25% of chances of full compatibility between siblings

• More possible combinations for a single human than there are humans beeins (~ 18 billions)
• Two people taken randomly = 1/10⁶ chances
• Strongly dependant of ethnies
• Strong biais in the databases

The main ideas

• Some 9/10 compatibility grafts show incredible results
• Some 10/10 compatibility grafts show poor results

 

We will be the first to ever work on HLA data

--> Determine if there are suboptimal and optimal HLA mismatches

 

The goal is to find good balances between:

 

• HLA Match VS Optimal mismatch
• GVHD vs GVT
• Immunosuppression VS infection risk
• Ablation VS Relapse risk

Oila oila

Thx :)

What makes you unique ?

Blood type

Genetic variants

DNA

Measurable attributes

Personal history

Personnality

Tastes

Health status

Almost everything is unique

Many unique things are still interchangeable

Autologous graft

With the patient's cells

Even this simple case often fails

Immune system (IS)

• The immune tolerance is an ACTIVE phenomena

 

• Every cells shows "self-antigens" to proof it belongs to the self

 

Anything that cannot express those self-antigens shall be shot down

 

 

• Auto-reactive cells get destroyed during the "training" process

Auto-immunity and tolerance