Markers of biobank sample quality Rainer Lehmann

Symposium 4
Biospecimen Science
Markers of biobank sample quality
Rainer Lehmann
Div. of Clinical Chemistry and Pathobiochemistry
University Hospital Tuebingen
and
Institute for Diabetes Research and Metabolic Diseases
German Center for Diabetes Research (DZD)
Germany
Rainer Lehmann, University Hospital Tübingen, Germany
The general opinion is, that essential pre-requisites for the
success of research projects are
- good experimental design
- precise, robust and reproducible analytical techniques (wet lab)
- sophisticated data analysis (dry lab) and interpretation
Rainer Lehmann, University Hospital Tübingen, Germany
1
Biobank sample quality
„….the quality of biospecimen is of paramount importance
for the reliability and validity of research results and all
downstream applications.“ Fay Betsou
(ISBER President 2013-2014)
Î pre-analytical phase!
Rainer Lehmann, University Hospital Tübingen, Germany
Biobanking / Blood collection
blood
collection
clinical
study
transport
centrifugation
freezer
plasma /
serum
Unfortunately, in general the interest of many investigators
for the pre-analytical phase is only minor
In complex clinical studies blood collection is often entitled
as the easy part
Rainer Lehmann, University Hospital Tübingen, Germany
2
Biobanking / Blood collection
blood
collection
transport
centrifugation
freezer
plasma /
serum
clinical
study
Typical questions of investigators (frequently asked after the end of their
studies) are:
The centrifuge of our ward was highly frequented, consequently the processing of
some blood samples was delayed. Is this a problem?
One third of the blood of our study was drawn by inexperienced medical students.
Unfortunately this resulted in several more or less hemolytic samples. Can we still use
the sample set?
Several biobank samples of my study have been thawed one or two times.
Are they still suitable e.g. for metabolomics investigations?
The centrifuge was in another building. We could not process the blood samples
before 2 h, is this a problem?
...
.
Rainer Lehmann, University Hospital Tübingen, Germany
Biobanking / Blood collection
blood
collection
clinical
study
transport
centrifugation
freezer
plasma /
serum
Typical questions of investigators (frequently after the end of their studies)
are:
Variable
The
centrifuge
and prolongated
of our wardexposure
was highly
time
frequented,
of blood tothat
room
is way
temperature
the processing
before of
some samples in
centrifugation
was
occasional
delayed. instances
Hemolytic
sera/plasma
are
instudy
the sample
set by medical students, which unfortunately
One third of
the blood of
the
was drawn
resulted in several more or less hemolytic samples. Can we use the samples?
Repetitive
freezesamples
and thawofcycles
had have
been been
performed
with
samples
Several biobank
my study
thawed
onesome
or two
times.
Are they still suitable e.g. for metabolomics investigations?
Systematic
prolongation
of thebuilding.
exposureI could
time of
blood
before
The centrifuge
was in another
not
process
theprocessing
blood samples
before 2 h, is this a problem?
...
.
Rainer Lehmann, University Hospital Tübingen, Germany
3
Markers of biobank sample quality
Î we applied metabolomics to investigated these preanalytical
steps
Blood collection
hemolysis
Transportation
delayed processing
Storage
repetitive
freeze/thaw cycles
Rainer Lehmann, University Hospital Tübingen, Germany
Metabolomics
TARGETED Metabolomics
NON-TARGETED Metabolomics
analysis of selected, well-known
metabolites
non-hypothesis driven,
comprehensive approach
20 - 300 metabolites per sample
> 2.000 metabolite-ion masses per
sample
automation (commercial kits available)
quantitative
Î identification of most important
metabolites
time consuming and labor intensive
semi-quantitative
Î population based studies
> 500 samples / week possible
2 weeks until final result
Î well selected and characterized
samples
≥ 4 months until final result
Rainer Lehmann, University Hospital Tübingen, Germany
4
Metabolomics data processing
primary note for the coming slides: Scheme of data evalution
Rainer Lehmann, University Hospital Tübingen, Germany
Markers of biobank sample quality
blood
collection
transport
centrifugation
analysis
plasma /
serum
clinical
study
pre-analytical process
biobank
sample
preparation
data
evaluation /
bioinformatics
valid results
Analytical specialists (wet lab)
Æ blood collection / hemolysis
Æ transportation
Æ storage
Hemolytic specimens occur frequently in clinical laboratories
Æ as high as 3.3% of all of the routine samples,
that means e.g. around 100 – 130 samples / day in our lab
Æ they account for up to 40%–70% of all unsuitable specimens identified
Rainer Lehmann, University Hospital Tübingen, Germany
G. Lippi et al. CCLM (2008)
5
Blood drawing
Hemolysis
blood
drawing
non-targeted metabolomics
classical parameters
n = 10
Rainer Lehmann, University Hospital Tübingen, Germany
Blood drawing
Hemolysis
PCA scores plot
High
Intensity
no
hemolysis
moderate
hemolysis
strong
hemolysis
18% of all metabolite ion masses were significantly altered (p<0.05, FDR<0.1)
Rainer Lehmann, University Hospital Tübingen, Germany
6
Crucial point in non-targeted metabolomics
nontargeted Metabolomics
Bioinformatics Æ canditate list of features / ion masses
Elucidation of the metabolite identity
Data base searches / MSn / confirmation by a standard compound
Rainer Lehmann, University Hospital Tübingen, Germany
Blood drawing
Hemolysis
Identified metabolites (sign. changed):
Lyso-PC 16:0
Lyso-PC 18:0
Lyso-PC 18:1
Tryptophan
Sphingosine 1-phosphate (S-1-P)
N-acetylornithine
C8-carnitine
Suggestions:
- hemolytic samples should be excluded, at least from high resolution -omics
research projects
- S-1-P, Trp, LPC C16:0,…. should only carefully be nominated as biomarker
candidates in research studies and must be thoroughly validated for robustness
- All biobank samples should be tested for hemolysis before storage
Î free Hb can easily be measured by a two wavelenght method
Rainer Lehmann, University Hospital Tübingen, Germany
Golf-SW et al. J Clin Chem Clin Biochem (1985)
7
Markers of biobank sample quality
blood
collection
transport
centrifugation
clinical
study
analysis
plasma /
serum
biobank
sample
preparation
data
evaluation /
bioinformatics
valid results
Æ blood collection
Æ transportation / processing of blood
Æ storage
Rainer Lehmann, University Hospital Tübingen, Germany
Markers of biobank sample quality
© Universitätsklinikum
Ulm
A 9 ml sample of blood from an adult subject contains
5 × 1010 red blood cells
3 × 109 platelets
„liquid tissue“
8 × 107 leukocytes
Rainer Lehmann, University Hospital Tübingen, Germany
8
Transportation / processing of blood
Experimental design
- 2h to 4 h exposure of blood to room temperature or ice water
Î clinic internal handling / transportation
- 8h and 24 h exposure to room temperature Î external transportation
n = 10
n = 10
2h, 4h, 8h and 24h
2h and 4h
non-targeted metabolomics by LC/MS
Rainer Lehmann, University Hospital Tübingen, Germany
Transportation / processing of blood
Transport of whole blood
(EDTA-blood)
room temperature
Peiyuan Yin
ice water
80
control
20
2h
4h
10
70
1
60
50
40
30
5
20
10
6
4
10
0
0
-10
-30
-10
2
8
7
-20
9
-40
3
-50
-60
-20
-70
-40
-30
-20
-10
0
10
20SIMCA-P 11 - 2012/1/10
30 11:29:47
40
n = 10
Rainer Lehmann, University Hospital Tübingen, Germany
-100
-80
-60
-40
-20
0
20
40
60
80
100
n = 10
Yin P, ….Xu G., Lehmann R.: Clin Chem 2013; 59:833-45
9
Transportation / processing of blood
Exposure of EDTA blood samples to room temperature
24h
external transports
8h
1.0
4h
2h
time needed for
clinical internal
handling / transport
0h
0.1
fold change of peak area 10
(m/z)
Peiyuan Yin
Rainer Lehmann, University Hospital Tübingen, Germany
Yin P, ….Xu G., Lehmann R.: Clin Chem 2013; 59:833-45
10
(m/z)
Transportation / processing of blood
fold change of peak area 0.1
1.0
24h
external transports
8h
4h
2h
0h
time needed for
clinical internal
handling / transport
Identified metabolites
C2-carnitine
C8-carnitine
C12-carnitine
N-acetylornithine
Hypoxanthine
Proline betaine
Palmitic amide
Sphingosine 1-phosphate
Indole
Tryptophan
LPC 16:0
LPC 18:0
LPC 18:1
LPC C20:3
Oleamide
L-Methionine
Biliverdine
Rainer Lehmann, University Hospital Tübingen, Germany
10
10
(m/z)
Transportation / processing of blood
external transports
8h
peak area 0.1
1.0
fold change of 24h
4h
2h
time needed for
clinical internal
handling / transport
0h
A targeted metabolomics approach which included
several of these metabolites confirmed the findings
B. Kamlage, et al. Clin. Chem. 2014
Identified metabolites
C2-carnitine
C8-carnitine
C12-carnitine
N-acetylornithine
Hypoxanthine
Proline betaine
Palmitic amide
Sphingosine 1-phosphate
Indole
Tryptophan
LPC 16:0
LPC 18:0
LPC 18:1
LPC C20:3
Oleamide
L-Methionine
Biliverdine
Rainer Lehmann, University Hospital Tübingen, Germany
Transportation / processing of blood
New (ongoing) studie
Goal:
- define confidence intervals for good and poor quality of blood
samples in ca. 100 samples (samples were selected at random in a
metabolic ward in our outpatient clinic)
- receive an impression of the number of changed metabolites
Æ modern, more sensitive mass spec (triple TOF-MS), new analytical strategy
Analytical approaches
Î non-targeted metabolomics
Î targeted metabolomics
Rainer Lehmann, University Hospital Tübingen, Germany
11
NON-targeted metabolomics
positive mode
detected ion masses:3137
negative mode
detected ion masses:2910
Significant changed metabolite ion masses
Xinyu Liu
Î 262 metabolite ion masses were significantly altered
2h
ESI +
ESI -
4h
room temp. 0.64 %
3.6 %
ice water
0.13 %
0.29 %
room temp. 1.48 %
5.09 %
ice water
0.82 %
0.14 %
114 metabolite ions
148 metabolite ions
n = 30
p < 0.05, FDR 0.05
unpublished data
Rainer Lehmann, University Hospital Tübingen, Germany
NON-targeted metabolomics
Potential markers of biobank sample quality
Xinyu Liu
metabolite ion mass A
metabolite ion mass D
50000
40000
30000
20000
10000
0
0h
Ice 2h
Ice 4h
RT 2h
RT 4h
0h
Ice 2h
Ice 4h
RT 2h
RT 4h
n = 30
Rainer Lehmann, University Hospital Tübingen, Germany
unpublished data
12
NON-targeted metabolomics
?
blood processing delayed by 4 h
metabolite (B)
metabolite (D)
metabolite (C)
Lactate
%-Sensitivity
%-Specificity
metabolite ion mass (A)
= lactic acid
98.3
76.7
metabolite ion mass (B)
91.7
98.9
metabolite ion mass (C)
96.7
100
metabolite ion mass (D)
100
100
n = 30
unpublished data
Rainer Lehmann, University Hospital Tübingen, Germany
Markers of biobank sample quality
blood
collection
transport
centrifugation
analysis
plasma /
serum
clinical
study
biobank
sample
preparation
data
evaluation /
bioinformatics
valid results
Æ blood collection
Æ transportation / processing of blood
Æ storage
Æ repetitive freeze and thaw cycles
Rainer Lehmann, University Hospital Tübingen, Germany
13
Biobanking / stored samples
Problem: the number of aliquots is limited
Î leads unavoidable to repetitive freeze and thaw
cycles aiming to save sample material
we investigated EDTA plasma by non-targeted metabolomics:
controls (metabolomics sample pretreatment was performed at once after drawing blood)
1x frozen (= common biobank sample)
2x thawed / frozen
4 x thawed / frozen
Rainer Lehmann, University Hospital Tübingen, Germany
Biobanking / stored samples
Î no significant alterations of metabolite ion masses between
fresh controls and 1x frozen plasma samples were detected
Î unexpectedly, only 4 masses changed significantly after 2x and 4x freeze/thaw
cycles!
Î we detected individual differences in the sample
stability!
instabilities of the metabolite pattern were detected in plasma
samples of 2 out of 10 individuals, even in 1x frozen aliquots!
Rainer Lehmann, University Hospital Tübingen, Germany
14
Planning phase of studies
and another aspect…. relevant for
biobank samples intended for MSdriven –omics analyses
Additives included in the sample collectors may affect the ionization
process during an LC-MS run thereby suppressing the ionization of
metabolites or introduce interfering compounds.
Rainer Lehmann, University Hospital Tübingen, Germany
Planning phase of studies
different S-Monovetten (Sarstedt)
were investigated
Î consequently, the first important step in the planning and preparation of a study is to check
Yinbe
P, used
….Häring
H.-U.,
Lehmann R.:ofClin
Chem 2013;59:833-45
to
for the
generation
biobank
samples
specific
sample
planned
Rainereach
Lehmann,
University
Hospitalcollector
Tübingen, Germany
15
Summary (1)
Blood collection/storage is the easy part of complex clinical studies,
BUT
may heavily affect the quality of biobank samples and consequently the
outcome and success of research projects
SOPs
strictly regulate the exposure time of whole blood to room temperature or
cooling (as well as the further processing),
BUT
this exposure time may vary from SOP to SOP, e.g. based on logistical
problems and resulting compromises in the SOP
and the exposure time may be prolonged in occasional instances
Rainer Lehmann, University Hospital Tübingen, Germany
Summary (2)
The discovered pre-analytical biomarkers may help to identify
a) systematic inaccuracies that arise during processing of whole blood that
affect the quality of biobank samples
b) random errors leading to particular outliers in the sample set
and
Investigators should be careful in nominating metabolites identified to be
sensitive to preanalytical alterations as biomarker candidates in their research
studies
P. Yin, et al. Clin. Chem. 2013
B. Kamlage, et al. Clin. Chem. 2014
Rainer Lehmann, University Hospital Tübingen, Germany
16
Summary (3)
- Recommendations 1.) the suitability of sample collectors should be tested before starting
sample collection
2.) hemolytic samples must be excluded
3.) place blood immediately in ice water after drawing until further processing
(for a fixed time; ideally not longer than 2 hours)
Î we prefer (EDTA-)plasma instead of serum as the favorable sample material
Î processing: centrifugation at 4°C, for 10 min, at 2000 × g
4.) the quality of biobank samples should be tested by using EDTA plasma aliquots
before the samples are used in expensive and time consuming analytical projects
5.) handling of biobank samples:
- samples should be thawed in ice water
- non refrozen plasma aliquots of biobank samples are recommended
- mix-up of samples exposed to different freeze and thaw cycles should be
avoided
Rainer Lehmann, University Hospital Tübingen, Germany
Summary (4)
- Recommendations Poor sample quality does not necessarily mean that the biobank sample has to be discarded,
but these samples should not be used for high resolution analyses, like metabolomics analysis
Stability in whole blood at room temperature
Stability in serum / plasma at room temperature
Albumin
Alk. Phosph.
Ammonia
Rainer Lehmann, University Hospital Tübingen, Germany
W. Guder et al., J. Lab. Med. 2002
17
Hans-Ulrich Häring
Andreas Fritsche
Norbert Stefan
Robert Wagner
Erwin Schleicher
Andreas Peter
Ann Kathrin Pohl
Heike Runge
Mareike Walenta
Miriam Hoene
Sabine Neukamm
Magnus Wolf
Christian Klingler
Cora Weigert
current
Bente K. Pedersen
Peter Plomgaard
Jakob Hansen
Oliver Kohlbacher
Erhan Kenar
Holger Franken
Lars Rosenbaum
Andreas Zell
Dalian University of
Technology
Xiaohui Lin
Hai Wei
Dalian Institute
of Chemical Physics
- CAS -
Guowang Xu
Xinjie Zhao
Perry Yin
Xinyu Liu
Shili Chen
Zhongda Zeng
Jia Li
NSFC
funding:
Rainer Lehmann, University Hospital Tübingen, Germany
18