EMBRYO SELECTION IS VITAL FOR SUCCESSFUL IVF WHY TIME-LAPSE - PowerPoint PPT Presentation

EMBRYO SELECTION IS VITAL FOR SUCCESSFUL IVF – WHY TIME-LAPSE IMAGING IS PROVING IMPORTANT Prof. Fishel Simon

THE EMBRYO – WHY THE FOCUS? • Largest single cause of IVF failure. Live Birth • Single main cause of early 45 miscarriage. 40 ~ 35% of embryos 35 make babies 30 % 25 20 15 10 5 0 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 Female age

THE EMBRYO – A NEED TO SELECT THE MOST VIABLE Time Lapse Imaging (non-invasive) • PGT-A (preimplantation Genetic Screening-Aneuploidy)

SELECTING EMBRYOS – TIME LAPSE IMAGING • There’s Time -Lapse Imaging … (using it as a closed microscope) • Predicting blastulation - and there’s • CAREmaps!

CAREMAPS – COMPUTING MORPHOKINETICS • Time to cell stages. • Durations. morphokinetic • Dynamics e.g. • Pronuclei. • Reverse cleavage. • Compaction. • Blastulation. • Strings.

TIME-LAPSE IMAGING – CONSTANT VISUAL OF MORPHODYNAMICS Intracytoplasmic Extracytoplasmic • Granulation. • Polar body. • Refractile bodies. • Perivitelline space. • Vacuolation. • Zona Pellucida. • Smooth Endoplasmic Reticulum • Fragments. Clustures. • Multi Nuclei. • Pronuclei. Imaging - Single v 300/day!

CAREMAPS ATLAS BMA Annual Book Awards 2016

BLASTOCOEL COLLAPSE IS NOT A GOOD THING Difficult to detect without TLI . ‘Embryos that exhibit collapse are as likely to hatch as those that do not, but are less likely to implant and should not be replaced if alternatives are available’. (Meseguer et al 2015, Human Reprod) 715 transferred blastocysts

TIME LAPSE ALLOWS NOVEL OBSERVATIONS • Which may be used to enhance embryo selection. CARE Fertility example: • Angle between polar bodies.

AN ACTIVE ACTIVATED EGG & PIV!

SIGNIFICANTLY  CPR WHERE ANGLE OF PB2 EXTRUSION WAS >45 ° FROM PB1. • Retrospective Logistic Clinical Pregnancy regression analysis of 2,367 ICSI 40 transferred blastocysts, all ages – known clinical outcome (KID) 32 24 % 16 8 0 Angle between PBs < 45 > 45 (p<0.001, n=2,367)

TIME-LAPSE TO PREDICT BLASTULATION • A high percentage of blastocyts never make babies. Of the 9, 7 predicted • A significant percentage of blastocysts result in to be a blastocyst. They did. miscarriage. • Some blastocysts make unhealthy babies. All 7 were • A case(s) in point… chromosomally abnormal!

‘ CAREMAPS ’ M.A.P.S M orphokinetic A lgorithms to P redict S uccess

TLI – COMPUTING MORPHOKINETICS VP= t1 - tPNf Compaction= tM - tSC CC1=t2 - t2PB Blastulation= tHN – tSB CC2= t4 - t2 Collapse= tBCend(n)- tBCi(n) CC3= t8 - t4 CC4= t16 – t8 S2= t4 - t3 S3= t8 - t5

PRIBENSZKY ET AL 2017: TIME-LAPSE META-ANALYSIS • Meta-analysis of RCT of morphokinetic algorithms v single time point morphology for embryo selection. • Outcome measures: • Clinical pregnancy, Live birth, EPL, Stillbirth. • 5 RCTs (n=1637) • LB: 44.2% v 31.3% (OR= 1.668; P<0.009) • EPL: 15.3% v 21.3% (OR= 0.662; P<0.019) • Stillbirth: NS

PRIBENSZKY ET AL 2017: TIME-LAPSE META-ANALYSIS “time -lapse is shown to significantly improve overall clinical outcome” Favours control   Favours time-lapse

TLI – COMPUTING MORPHOKINETICS : ‘ CAREMAPS ’ VP= t1 - tPNf Compaction= tM - tSC CC1=t2 - t2PB Blastulation= tHN – tSB CC2= t4 - t2 Collapse= tBCend(n)- tBCi(n) CC3= t8 - t4 CC4= t16 – t8 S2= t4 - t3 S3= t8 - t5

TLI - QA

QA – ATTAINMENT SCORES

IMPROVED OBJECTIVITY: QUALITY ASSURANCE Interobserver and intraobserver agreement for assessed parameters

CREATING A TIME LAPSE ALGORITHM • An algorithm is a detailed step-by-step instruction or formula for solving a problem or completing a task. • Objective and based on data. • Ranks embryos according to their likelihood of: Asking: “Which embryo should I transfer?”

A CHOICE OF ALGORITHMS: Integrated software Published In house derived • Easy to use. • Many to choose • Takes time and from. expertise. • Derived from large • Can retro-test on • Data and experience heterogeneous data. own data. grow quickly. • Supporting • Little evidence to • Can be as simply or evidence. support complex. suggesting better transferability. • Reassurance that than morphology • Barrie et al *2017 built and validated alone. Fertil Steril. in house. * ‘These results highlight the need for the development of in -house ESAs that are specific to the patient, treatment, and environment.’

WHAT MAKES A GOOD ALGORITHM? • Based on large data and tested independently. • A high predictability score (e.g. Area Under ROC Curve). • Incorporates variables which are reliably annotated (high IOC). • Scores are weighted according to importance. • Allows relative ranking. • Simple to decode and sense check.

USING TIME-LAPSE DATA ONLY WHEN OUTCOME KNOWN (USING LIVE BIRTH DATA ONLY) SET DET SET DET KID positive KID negative KID negative (KID Positive) No KID** 1 x implantations lost Pregnancy 2 x implantations loss lost No LB KID Two Live Birth One Live Birth No Live Birth 2 x LB KID KID positive KID positive KID negative negative > 2,500 LIVE BIRTHS!

“BREAKTHROUGH SCIENCE” Nominated for the Rbmonline Robert Edwards 2013 Award

TIME-LAPSE IMAGING – CONSTANT VISUAL OF MORPHODYNAMICS Intracytoplasmic Extracytoplasmic • Granulation. • Polar body. • Refractile bodies. • Perivitelline space. • Vacuolation. • Zona Pellucida. • Smooth Endoplasmic Reticulum • Fragments. Clustures. • Multi Nuclei. • Pronuclei. Imaging - Single v 300/day!

CARE FERTILITY GROUP – 1000 LIVE BIRTHS Mean ♀ Age: CAREmaps = 36.9 v SI = 35.3 Live births = 973 deliveries

INCLUSIVE CONFOUNDING VARIABLES • Embryoscope (y/n) • Total # ectopic • Patient age (<38/38+) • BMI • Day of embryo transfer • AMH • No. embryos transferred • AFC • Patient type • Gonadotropin type • Donor age • Gonadotropin dosing days • Total previous cycles • Gonadotropin total dose • Total no. previous live births • # Eggs collected • ICSI (y/n) • # M2 eggs • Total no. miscarriages • Ratio M2 eggs to total • Intralipid (y/n) • # M2 eggs fertilised • Duration of infertility

CONFOUNDING VARIABLES INCLUDED IN THE MODEL • Embryoscope (y/n) • Total # ectopic • Patient age (<38/38+) • BMI • Day of embryo transfer • AMH • No. embryos transferred • AFC • Patient type • Gonadotropin type • Donor age • Gonadotropin dosing days • Total previous cycles • Gonadotropin total dose • Total no. previous live births • # Eggs collected • ICSI (y/n) • # M2 eggs • Total no. miscarriages • Ratio M2 eggs to total • Intralipid (y/n) • # M2 eggs fertilised • Duration of infertility

CARE FERTILITY DATA (LIVE BIRTH OUTCOME ONLY) 24,000 records of treatment • 21,379 Standard incubation treatments • 2,527 Embryoscope treatments • 14,878 unique patients Statistical Analysis: *multiple variable logistic regression models were fit to assess the effects of embryo rank on each potential confounder * Akaike information criterion (AIC) penalty fit for number of parameters and stepwise selection

LIVE BIRTHS (DELIVERY EVENTS AGE <38) – SIG POINTS TLI SI 60,0 * P<0.0001 50,0 * P<0.0001 * P<0.0001 40,0 * P<0.0001 * P<0.001 30,0 20,0 10,0 0,0 # pos Beta # with >=1FH Biochem Loss # Implantations # Clin # Live Birth babies/EmbTrd miscarriage Event

LIVE BIRTHS (DELIVERY EVENTS) Significance diluted in >37 age group * P<0.0001 CAREmaps SI * * 45 * * 40 35 30 25 20 25% 19% 15 10 5 0 Overall Single Blast Uplift

CAREMAPS – HIERARCHICAL RANKING OF EMBRYOS 2 nd Study in the ‘Trilogy’ – Rbmonline – 2018 Time-Lapse Imaging Algorithms Rank Human Preimplantation Embryos According to their Probability to Result in a Live Birth. Simon Fishel 1a* , Alison Campbell a , Sue Montgomery b , Rachel Smith c , Lynne Nice d , Samantha Duffy b , Lucy Jenner e , Kathryn Berrisford, Louise Kellam e , Rob Smith f , Fiona Foad g , Ashley Beccles a

HIERARCHICAL SELECTION OF EMBRYOS: Single Blast Transfer P < 0.001 60,0 tSB or relSBIVF ≤ 93.1h A tSB or relSBIVF >93.1 h B 50,0 dB ≤ 12.5 h 47.7% tSB or relSBIVF >93.1 h C 40,0 dB >12.5 h Unable to be annotated D 30,0 % 20,0 # ET LBR Miscarriage (%) A 373 51.7 24.0 10,0 B 297 35.0 32.5 C 93 31.2 32.6 0,0 D 80 13.8 35.3 A B C X % LBR Miscarriage (%)

RANKED EMBRYO LB OUTCOME • D << A (OR = 0.3046; P<0.010) Strong evidence of an • D << B (OR = 0.428; P<0.01) A has a 233% effect of embryo rank • B < A (OR = 0.7114; P<0.01) chance of LB on the odds of live • C <A (OR = 0.6501; P< 0.01) compared to D births. • B > C (OR = 1.09; P<0.01) • C >> D (OR=2.135; p<0.01)

CAREMAPS V MORPHOLOGY: Grade 2:2 highest LB!

CAREMAPS MORPHOKINETICS V MORPHOLOGY 1373 Blast SET cycles 3 rd Study in the ‘Trilogy’ . 679 (50%) Live Births Live Births (%) v MK Grade Live Births (%) v Morphology 70 70 60 60 50 50 40 40 % % 30 30 20 20 10 10 0 1:1 1:2 2:1 2:2 2:3 3:2 3:3 0 Live Births (%) 49,4 50,6 60 53,3 28,9 15 20 A B C D Morphology Grade