The 100,000 Genomes Project

Cancer Sample Quality J Louise Jones

Clinical Lead for Molecular Pathology Genomics England

Consultant Pathologist Barts Health

‘Standard’ Tissue Handling

Operative procedure “Warm ischaemia”

Whole sample in fixative

Delivery to Pathology

Sample ‘cut-up’

Type of operation:

Cut vascular supply

Size of sample Volume of

fixative Type of fixative

Time since procedure:

2hrs – 48hrs

Size of blocks: Standard vs ‘mega’ block

Sample processing

Different machines Different programmes

Weekend programmes: additional fixation 12hrs

• Standard fixation is ‘Formalin’

– 40% formaldehyde in 100w/v distilled water = 100% formalin

– 10% formalin is standard

– 10% neutral buffered formalin = phosphate buffered saline

• Impact of formalin on nucleic acid:

– Fragmentation

– Base modification

– Cross-linking within DNA and between DNA and proteins

Effect of fixation on nucleic acid

Optimal preservation of nuclei acid afforded by snap-freezing samples: Liquid nitrogen (-176oC) or Isopentane cooled in LN

Fresh Frozen (FF) samples optimal

Effect of formalin fixation on WGS

“Choppiness” prevents reliable CNV and SV calling

Data courtesy of J Becq, Illumina

‘Optimal’ Tissue Handling

Operative procedure “Warm ischaemia”

Whole sample collected fresh

Delivery to Pathology < 2hrs

Sample ‘cut-up’

Type of operation:

Cut vascular supply

Snap-frozen

Reduces variability Limits DNA damage

2

• Early pilot data: real-world FFPE

–High rate of fails pre-sequencing

–Patterns of severe coverage bias impede CNV calling • Development of series of experiments?improvement

–controlled side-by-side replicates, varying only single parameters

• Largely around DNA extraction conditions (temperature, salt)

–Similar experiments performed across a number of centres

• Different and evolving background conditions for molecular pathology • Massive drive on sample-handling data

–Fully annotate pilot samples

–Fully annotate experimental samples Massive integrated dataset

• Multi-centre

• Multiple tissue types

• Included variation in conditions around experimental parameters

• Pre-sequencing QC and sequencing QC metrics

Development of insights

FFPE - 19 + 17 = 36 FF - 6 + 10 = 16 Blood - 2 + 10 = 12

Tissue type e.g

lung

Cold Ischaemia

time

Tissue Block Size

Fixative PH of Fixation

Fixation time

Processing Machine

Clearing Agent

Processing Protocol

Time in Fixative on processor

Time in hot wax

Maximum temp

during processing

FFPE sample Type

Number of scrolls, sections or punches

Thick- ness or dia- meter

Sample disrupted prior to de-wax

De- wax Agent

De-wax Protocol E.g. tube or slide

Re- Hydrati on

56 degree lysis inc. time

56 degree lysis inc. agitation

Reverse x-linking temp

Reverse x-linking inc. time

RNA removal

DNA Purification kit

Manual or automated purification

No. of

elution's

Elution buffer

Elution volume

13 FFPE DNA extraction protocols from 10 centres

reviewed

Tissue Handling Variables

Data courtesy of Shirley Henderson & Pauline Robb, Oxford

Data courtesy of Dr Shirley Henderson & Pauline Robb, Oxford

Tissue Handling Variables

Tissue type

Cold Ischaemia time Tissue Block Size Fixative Fixation pH pH checked fixation time

Tissue processing machine man Tissue processing machine model Processing xylene free Processing Protocol Time in fixative on the processor Time in hot wax on the processor min Maximum temp during processing degrees

tumour type tumour sample cellularity

percent neoplastic Thicknness/diameter Necrosis

Sample Type Number of scrolls sections or punches

De Paraffinisation Agent De Paraffinisation Protocol Re Hydration 56 degree lysis incubation time 56 degree lysis incubation agitation Reverse x.linking temperature Reverse x.linking incubation time minutes Reverse x.linking additions RNA removal DNA purification kit DNA Purification manual or automated

Percent aligned Percent chimeric pairs Median insert size Diversity

GC dropout AT dropout

DeltaCq measured by Illumina

Responses: DNA QC Metadata collected for Cambridge, ICL, KCL, Oxford, UCLH Responses: WGS QC

2 multivariate analysis performed (where n > 10 for each factor of each covariate): • Metadata with DNA QC • Metadata with WGS QC

86

Multivariate Regression

Orientation of arrow shows improvement of the metric compared to the mean (dashed line). Number is the sample size. The length of the bar is proportional to the effect of this factor on deltaCq

Observations:

• Block size: standard improves; mega worsens deltaCq

• Time in hot wax: 140 to 360 min improves; 490 – 540

worsens deltaCq Fixation time: >48h worsens deltaCq

• Time in fixative on processor: < 60 min improves deltaCq

• Samples type: Scrolls worsens deltaCq

• RNA removal: if yes improves deltaCq; if no worsens

deltaCq 87

Multivariate Analysis on deltaCq

Observations: • Block size: small improves AT and insert size but worsen GC; mega worsens % chimeric pairs, % aligned

• Time in fixative on processor: 180 min improves % aligned, diversity, insert size; 2900 min worsens % chimeric pairs, %

aligned, diversity and insert size

• Sample type: Sections improves % chimeric pairs; Scrolls improves AT worsens GC; cores worsens % aligned, diversity and

insert size

• RNA removal: if yes improves GC and insert size

• Fixative: Neutral buffered formalin improves AT, % chimeric pairs, % aligned; 10% formal saline worsens AT, chimeric pairs

and % aligned but improves GC

• Fixation time: 24-48h improves % chimeric pairs, % aligned, diversity and insert size; >48h improves GC but worsens

chimeric pairs, % aligned, diversity and insert size 88

Block - standard Block - mega

Block - small

AT Dropout GC Dropout % chimeric pairs % aligned diversity m. Insert size ation on processor 2900 min

ixation on processor 180 min

ixation on processor <60 min

Sample - sections

Sample - scrolls

Sample - cores

RNA removal - yes

RNA removal - no

Neutral bufferd form

nbuffered 10% formal salin ixation time 24-48 h

Fixation time >48h

Multivariate Analysis on Sequence QC

‘Optimal’ Tissue Handling

Operative procedure “Warm ischaemia”

Whole sample collected fresh

Delivery to Pathology < 2hrs

Sample ‘cut-up’

Type of operation:

Cut vascular supply

Snap-frozen

Operative procedure “Warm ischaemia”

Whole sample collected fresh

Delivery to Pathology < 2hrs

Sample ‘cut-up’

Type of operation:

Cut vascular supply

Small FFPE ‘Genomic Block’

10%NBF <36hrs fixation

Fresh Frozen samples give best chance of WGS success For FFPE: need ‘optimised’ tissue handling

Impact of Optimised Tissue Handling

• Communication: – Timely contact from theatre

• Labour intensive: – Requires ‘on-call’ specimen collection – Requires ‘on-call’ pathologist cut-up

• Against current workflow

– Push versus pull

• Space issues

– Access to cut-up bench according to sample arrival

• Facilties – Availability of liquid nitrogen for snap-freezing

Challenge of Optimised Tissue Handling

• Cooling sample to inhibit enzyme activity – Observational data from samples with prolonged refrigeration prior

to sampling

– Samples (lung excisions) refrigeration post-surgery for 12-72 hrs

– Piece taken for genomic analysis – into FFPE

– Remaining sample fixed for routine cut-up

– DNA extracted using Qiagen protocol

– Submitted to Genomics England

Approaches to Mitigate Cold Ischaemia

22 hrs

72 hrs

Sample refrigeration: variable delta Cq: not

related to ischaemic time

Good quality sequence QC

Data courtesy of Illumina

Fresh specimen (<1hr post Sx) Refrigerate 4oC

Vacuum Pack RT

(7) 24hrs 48hrs 72hrs

Frozen FFPE Vacuum Pack 4oC

FF + FFPE FF + FFPE FF + FFPE

FF tissue: DNA extraction (Qiagen) RNA extraction Morphology FFPE: Morphology Immunohistochemistry 8 samples (2 bilateral), all conditions in duplicate

Assessment of Vacuum-Packing

Yield is excellent and variable regardless of conditions

Diversity and % aligned are excellent. Variations are among expected noise

Insert sizes are not reduced in the vacuum pack samples compared to the 0h control AT and GC coverage are in the expected range for FF

samples

Patient 3: QC data – all VP, RT versus 4oC

The analysis showed very good quality libraries and no signs of degradation regardless of the experimental condition.

Data courtesy of Illumina

Tumour FF Time =0 Tumour FF VP 72hrs

Normal FFPE Time =0 Normal FFPE VP 72 hrs

Morphology Maintained

Excellent RNA integrity to 48 hrs

Implementation of vacuum packing into

Pathology

Vacuum Pack in Theatre

Into fridge or coolbox

Transfer in coolbox to Pathology

Refrigerate until cut-up

3

• Molecular Pathology handling is absolutely key

–Size of block, processor time, fixation time, fixative

• FFPE remains suboptimal compared to FF

–Tight molecular pathology will improve sequence quality

–CNV calling from ‘good’ FFPE is still sub-optimal (10-fold increase in CNVs)

–Ongoing work required….

• Fresh Frozen tissue remains superiorand gold standard

–Time can be ‘suspended’ using vacuum-packing and/or refrigeration

–Implementation of vacuum packing/refrigeration will take time and incur (modest) cost

• DNA extraction conditions

–Likely important influences of scrolls/sections, proteinase K lysis +others

–Modification of RXL temp/salt can significantly improve sequence from tissue highly compromised from upstream mol path handling

–Caution: not universal panacea; ??negative impact on well-handled tissue

Key Determinants of Tissue Quality

Annals Oncology 2016; 27: 532

• Higher frequency of 90bp from UMFIX vs NBF

• Higher frequency of 129bp and 305bp fragments from UNFIX (SF significantly higher)

• Greater specificity in SNV calls in UNFIX and SF vs NBF

SNV Calls

UMFIX superior to NBF

Annals Oncology 2016; 27: 532

Alternative Freezing Strategies

GEL Code Treatment (Time 0) Time = 4hrs

1 Liquid Nitrogen – onto dry ice -80oC

2 Isopentane on dry ice -80oC

3 Cryospray (indirect) – onto dry ice -80oC

4 Dry ice -80oC

5 Wet ice -80oC

6 Saline – into fridge (4oC) -80oC

7 Cryospray (direct) -80oC

DNA Quality (Tapestation)

6A 1A 5A 2A 3A 4A 1B 5B 3B 2B 4B 6B 2C 1C 6C

X 8.5 8.3 8.7 9.0 8.7 8.9 7.4 9.1 8.8 8.7 X 8.6 8.8 6.2

Sample

DIN

RNA Quality

• Optimal tissue handling is essential for WGS

• Fresh-frozen samples yield highest quality whole genomes

• Attention to detail can significantly improve fixed samples: – ‘Optimised FFPE’

• Need to consider alternative strategies: – Refrigeration

– Vacuum-packing

– Alternative fixatives

– Alternative freezing strategies

Summary

Oxford BRC

• Prof Anna Schuh

• Dr Shirley Henderson

• Ms Pauline Robb

Illumina

• Mark Ross

• Jennifer Becq

• David Bentley

Genomics England

• Clare Turnbull

• Jo Mason

• Kay Lawson

Experimental Leads

• Gerry Thomas

• Adrienne Flannagan

• Clare Verrill

• Emily Shaw

Acknowledgements