How do we define an adequate

response to synacthen? Issues over

the standardisation of current
immunoassays – what is the truth?
Dr Les Perry
Consultant Clinical Scientist
Croydon University Hospital NHS Trust
London, UK
April 2016
 Introduction
• Standardisation
• Calibration
• Physiology of cortisol assay
• Application of cortisol assay in investigation
of pathology
• Analysis
– Historical
– Immunoassay
– Mass assays
• Current problems
 Immunoassay standardisation &
calibration
• Definitions

– Standardisation
‘process of ensuring that all methods for
determining the concentration of a particular analyte
give the same results’

– Calibration
‘process of assigning values to an unknown
samples using a standard’
 Immunoassays

• Not direct measurement (eg ruler).

• They estimate analyte concn in sample by comparing
signal strength (from labelled reagent) with that
from a similarly treated ‘standard’ sample

• Problem
– for most analytes quantified by immunoassay
there are no reference methods with which to
calibrate standards
 Importance of standardisation
• Usual practice to report result and give ref range for
that method next to result
• Once ref range established difficult to avoid it being
used to subconciosly being used to interpret results

• Serious implication for patient management if:

– lab changes the method and the ref interval is
significantly different
– Manufacturer changes calibration without
change in reference range
 Lack of standardisation

• Hinders communication and scientific progress

• eg reduce value of sex- and age-related reference
intervals in literature which could otherwise be
used to improve diagnostic capability of different
manufacturers tests for the same analyte

– Some publications state that LH:FSH ratio >3:1 diagnostic

of PCOS. Newer LH immunometric assays give much
lower values than previous methods used to derive the
ratio so this diagnostic ratio is no longer appropriate
 Standardisation: EQA schemes
• Regular EQA schemes initiated improvement in
standardisation of same analyte by different
immunoassays.
• Distribution of pooled samples; results returned to
scheme coordinator; generation of ALTM; feedback to
labs and manufacturers. (ALTM ≠ ‘true value’)
• Wide variation improved over time
• Some methods consistent BIAS. Recovery experiments
help pinpoint inappropriately standardised methods
• Responsibility of participant in EQAS to notify
manufacturer of sig BIAS of their method from ALTM
 Standardisation: International Standards
• To achieve agreement across methods , a single
recognised standard is needed.
(eg cortisol 98.9 ±0.2% purity NIST)

• Management by International body for credibility

• Most IS for immunoassay analytes prepared in UK
(NIBSC)
• Prepn of IS is a specialised activity and validation of
each step is required with support from several
independent labs
 Standardisation: International Standards
• To be designated an IS preparation has to be
authenticated by the expert Committee on
Biological Standardisation (ECBS) of the WHO.
• An IS is a preparation to which an International Unit
(IU) has been assigned based on international
collaboration involving different assay systems in
different global labs.
• IRP no longer used for newer materials.
• IS designation has helped to resolve potential
differences between US and European attempts to
standardise locally.
 Standardisation
• Many organizations involved in standardisation
– Institute for Reference Materials & Measurement (IRMM) -
Brussels
– Centres for Disease Control (CDC) – USA
– College of American Pathologists (CAP) – USA
– Clinical Laboratory & Standards Institute (CLSI) – USA
– National Institute of Standards & Technology (NIST) – USA
– National Institute of Health (NIH) – USA

All have subtle differences but now IFCC active in encouraging

standardisation globally
 Definitive & Reference methods
• Definitive method - accurate precise thoroughly
investigated & evaluated for inaccuracy; used
accurate primary reference materials.
• Suitable for calibration of all methods & Standards in
USA

• Reference method – may be a subsidiary method,

calibrated from the definitive method and used to
calibrate field methods
• For some analytes reference method the highest
category as experts consider a definitive method
unlikely to become available
 Method-related causes of
Standardisation differences

• Differences between methods even when ‘pure’

analyte available for preparation of standards
• Variability in molecular structure of the analyte
between ref std and patient sample
• Variability of antibody in assays standardised using
same ref material
• Differences in ‘non-analyte’ constituents of
standard and sample (Matrix Effect)
 ‘Matrix’ effect: Matrix

Constant bias in analysis between 2 sample types

eg serum v plasma
or
serum & charcoal stripped serum

Matrix preparation of calibration curve versus test

sample
‘Matrix’ effect : Sample interference

Action of a known substance that can bias an

otherwise unaffected sample

Eg complement in sample may bind to Ab thereby

affect Ab binding to analyte
or
urine sample measurements tend to be sig biased
if used serum calibrants but the amount of
interference vary significantly in individual samples
depending on the salt concentration of the urine
 ‘Matrix’ effect : Cross-reactivity
This is where interfering molecule binds to analyte-
binding site of Ab due to structural similarity this can
lead to false elevation of test result
Testosterone
 Summary
• Homogeneous assays do not have a wash step so
their design requires meticulous attention to
minimise matrix effects

• Select antibody with the highest affinity and best

specificity
• Ensure appropriate choice of matrix for standards
and calibrators
• will significantly reduce the method-related bias
and sample interferences

• Ref: The Immunoassay Handbook. Ed David Wild, 4th Edition 2013.

Elsevier Publications
 Cortisol assays and the synacthen test

• Physiology
• Pathophysiology
• Synacthen test
• Cortisol immunoassays
• Interpretation of the short synacthen test
 Physiology
• Steroid hormone: produced
in z. fasiculata of adrenal • Lipophilic. Transported bound
cortex to CBG (82%), albumin (10%).

• Secretion directly controlled • ‘Free’ fraction (8%) mediates

by anterior pituitary ACTH metabolic effects
release which is regulated
by hypothalamic CRH • Essential role in health & well
being:
– Gluconeogenesis
• Exerts –ve feedback on both
– H2O & electrolyte balance, BP
ACTH and CRH; stress
control
directly stimulates
hypothalamic CRH release
 Pathophysiology

• Acute cortisol deficiency (Addisonian crisis)

catastrophic dehydration & Na loss – potentially fatal
so early diagnosis essential.

• Mild cortisol deficiency is more difficult and requires

an alert clinician to pick up on symptoms (fatigue,
lethergy, weakness)

• Symptoms of cortisol excess (Cushing’s syndrome) eg

wt, BP or glucose often attributable to other causes.
 Pathophysiology
• Widely used random cortisol of little value when
assessing adrenal status
– (eg Centaur 0800-1100h, healthy volunteers, cortisol 96 – 720 nmol/L)*
– Rarely diagnostic however if <100nM ?adrenal insuff or >450-550 nM
hypoadrenalism unlikely.
– Any cortisol result between these two extremes needs further
investigation
– Early am serum cortisol and concurrent plasma ACTH can help determine
if adrenal insuff is 1o or 2o **

When adrenal disease suspected most useful tests are those

that stimulate or suppress the HPA-axis, using cortisol as the
marker to quantify the adrenal response

*Grimspoon S & Biller B. JCEM 79; 923-31, 1994

**Oelkers W; Diederich & Bahr V. JCEM 75; 259-64, 1992
 Pathophysiology - history
• 1965
SST introduced –rapidly assess adrenal function
• Mid 1980’s
SST used to assess the integrity of the HPA axis
Prior to this used the ITT – hazardous!
• 1988 – definition of a ‘pass’ for SST not consistent
amongst UK endocrinologists
• Mid 1990’s
Discrepancies reported between SST and ITT based on
the pass/fail cut-off values used for each test; SST
based on Mattingly method
 Cortisol analysis : Historical

• Pre- 1962
Research labs only. Req’d several labour –intense steps

• 1962 – Mattingly method

Simple sulphuric acid–induced fluorescence of
11-hydroxycorticoids (cortisol + corticosterone)
Used 2ml plasma 6 samples/90min

• 1973: 1st published RIA

 Cortisol assays : Immunoassays
RIA became method of choice for cortisol analysis in
many general chemistry labs despite its limitations
- many Ab’s with large variation in specificity
- method used to liberate steroid molecule from
its carrier protein (plethora of publications)

• Mid 1990’s
1st automated immunoassay analyser for cortisol 1992
By mid-1990’s automated immunoassay analyser
expansion and most cortisol analyses done on an
automated platform
 Cortisol assay : Immunoassay
• 1998
Defining the normal response to the Syncthen test: implications
for the investigation of hypothalamic-pituitary disorders
Penny Clark et al Clin Endo 49: 287-92.

Examined cortisol concentration at 0 & 30 min post Synacthen

n=100; 67% female
30’ mean (5-95th centile) 0-30’ Increment
TDX (Abbott) 811 (626-1431) 488 (289-776)
ACS-180 (Chiron) 741 (569-1078) 399 (208-593)
DELFIA (Pharmacia) 707 (510-1088) 409 (222-641)
Coat-a-Count (DPC) 866 (590-1548) 494 (222-762)
 Cortisol assays

• Demonstrated assay dependence of cortisol cut-off

• Effect of gender on the cortisol response to SST

varied across methods for females

• Not prompt widespread application of method

specific cortisol cut-offs for interpreting the SST

• Most labs continued to quote the ‘historical’ 550

nmol/L cut-off
 National Audit of the SST*
• Evidence from Welsh Regional audit 2005
• UK National audit 2010 (89/ 244) labs gave response
to web-based questionnaire). Wide difference in
protocols for performing and interpreting SST.
• Majority of labs consider >550 nmol/L – pass
• Increment >200 nmol/L – pass
• Despite all labs knowing their method NEQAS BIAS
only 19% consider this when interpreting SST results
• Using ID-GCMS targeted sample (550nmol/L)
routine IA analysers gave results from 435-609
nmol/L
* Chatha, Middle & Kilpatrick. Ann Clin Biochem 2010; 47 : 158-164
 Cortisol assay: mass assays

• 1975
1st GCMS – derivatisation & 3H for procedural
losses

• 2013
Owen et al. Developed rapid assay for analysis &
implementation into routine service lab
Ann Clin Biochem 50: 345-52
 Clinical Chemistry
Candidate Reference Method
Procedure for the
Quantification of Total Serum
Cortisol with LC-MS/MS.

Hawley JM, Owen LJ,

MacKenzie F, Mussell C,
Cowen S, Keevil BG.

Clin Chem. 2016

Jan; 62(1) : 262-269.
28
Current problems?
 Cortisol assays: Gender & Matrix effects

• The effect of serum matrix and gender

on cortisol measurement by commonly
used immunoassays.

Dodd et al; Ann Clin Biochem 2014, 51: 379–385

Gender differences

31
Gender differences
The nature of the
beast is that since we
want to use exciting
endogenous levels in
the Scheme, the male
donations we get in
through the door get
used for Cortisol (and
now General
chemistry) so the
effect that has always
been there was largely
over-looked/ignored
although JGM and
some participants
were aware of the
potential problem
32
 EQA sample from the UK NEQAS
scheme June 2015

Female pool and a clear bimodal distribution (reprinted with

permission form Dr Finlay Mackenzie, UK NEQAS scheme
organiser).
Youden
Plots

Reprodu
cibility
on
repeats

34
Pool Stability

35
No recent change in calibration

36
 UKNEQAS

• Although the coefficients of variation for individual

methods are typically very tight, when all methods
are accounted for the overall variation can be
considerably wider.

• This frequently produces a wide dispersion of

results with a gender-specific bimodal distribution
 Cortisol assay instability

• Several recent studies* with current IA platforms

continue to highlight :
- matrix & gender differences across platforms
- ‘pass’ concn of SST should be method related
- although helpful in the short-term, changes to
reagent lots, calibrators & assay reformulation
may increase result uncertainty
(new Roche GenII assay has caused concern)
- increase in % labs with NEQAS BIAS score >10%
limit. So NEQAS working on alternative to ALTM

* El-Farhan et al Clin Endo (Oxf) 2013; 78: 673-80

 UKNEQAS : Cortisol assay performance
2011 - 2015
Analyte B Score (%) 2011 2012 2013 2014 2015

Cortisol ALTM ±10% 26.4 36.4 61.2 64.2 69.7

UKNEQAS : Cortisol bias v ALTM
Recovery

41
Female Male

42
 Cortisol assays: Binding proteins

Current assays measure total cortisol

Binders are CBG (hi affinity/low capacity) and albumin
Conditions that affect CBG and albumin affect cortisol
- pregnancy or OCP over-estimate cortisol
- nephrotic disease or liver impairment – under
estimate cortisol concentration

Early RIA’s displaced steroid from its binding protein; not

all current methods do:
Roche – Danazol.
Abbott- Ab affinity. adequate if normal CBG; ?not if ^CBG
 Cortisol assay: Standardisation

• Differences in Ab specificity
• Lack of a single reference material
– At least 6 certified reference materials for cortisol
(current platforms traceable to some of these)
• Lack of single reference method
– Several Mass Spec methods : ID-GCMS initially but
more recently using LC-MS/MS as this technology
has improved
– Current platforms confirming the validity of their
platform against some of these LC-MS/MS
methods
Gordon.avery@abbott.com; HHarper@beckman.com; Robert.Banfield@Roche.Com;
Allan.at.Thompson@Siemens.com; Lisa.Wright@siemens.com

Dear Diagnostic Company Colleagues

I am writing to the manufacturers of all 5 major Cortisol methods whose products cover around 98% of
participants in the UK NEQAS Scheme.

The five Methods are Abbott Architect, Roche Cobas/Modular, Siemens ADVIA Centaur, Siemens Immulaite
2000/2500 and Beckman Access/Dxi
Letter to the
I am asking if it would it be possible for you to share with me a ‘dossier’ of your most up-to-date validation
data (including recoveries, cross reactants, reference method values, between-method comparisons etc
etc.)?
Diagnostics
I know that when speaking informally to you all before, each of you believes your own assay to be valid and
accurate. Nevertheless, there is a wide spread of biases seen between-methods on our minimally
Industry
manipulated serum pools. From my perspective this is not something I can ignore.

For my part, I can confirm that our material is human serum and is minimally manipulated. Pools are made
from male-only or female-only donations. We mix only like-for-like concentrations. The Pools are filtered to
November
0.5 um and do not contain preservative. I have not tested my material for possible cross-reactants etc etc
but although my working hypothesis is that they are representative and that they are commutable, I would be
happy to investigate both aspects further. The majority of our Pools are endogenous, but we do perform
2013
recovery experiments with exogenous cortisol. Different cortisol powders have been tested in the past and
the results were completely super-imposable between compounds.

I can also confirm that the recovery of added cortisol is not always in step with the method bias. We observe
that there are methods which over-recover and have a high positive bias. We have methods which under-
I took it upon
recover and have a large negative bias. However, there are methods which over-recover, but are essentially
unbiased in the Scheme. The ALTM, our current target value, recovers ~ 104% of added cortisol.
myself to talk to
For operational reasons, the majority of our specimens for Cortisol are from male donors; the female the major
donations being preferentially reserved for our E2 and Progesterone programmes. We do send out a good
number of female specimens and the between-method differences characteristics are different to that of
male serum.
manufacturers.
Please see the attached pdfs which show where we are in terms of B score versus C score Penalty Box I wanted to avoid
Plots and some representative histograms.

Please feel free to give me a call if you require any further background.
the marketing and
I look forward to your replies. get to the brass
Kind regards
tacks
Finlay

Finlay MacKenzie, Scheme Organiser UK NEQAS for Steroid Hormones

th
Tuesday 5 November 2013.
45
Cortisol method bias to CR MS method

Abbott
Architect
Cortisol method bias to CR MS method

Roche E170
Modular
Cortisol method bias to CR MS method

Beckman
Access
Cortisol method bias to CR MS method

Siemens
Advia
Cortisol method bias to CR MS method
Cortisol method bias to CR MS method
 UKNEQAS – Dec 2015

• 2015
UKNEQAS announced in annual report
that from April 2016 will abolish ALTM
and use candidate reference LC-MS/MS
target for BIAS calculation
What-if Mass Spec Target

53
February 2016

54
 Roche Gen II cortisol assay

• Released July 2015

• New Ab and new calibration

Male samples v cLCMS/MS
 Female samples v cLCMS/MS

A D

B E

C F
 Pregnancy v cLCMS/MS

D
A

B E

C F
Cortisol v Cortisol Binding globulin
 Prednisolone v cLCMS/MS

A D
1600

Roche Gen II - Target

1400
1200

(nmol/L)
1000
800
600
400
200
0
-200
0 20 40 60 80 100
B E Certified Target (nmol/L)

C F
 Metyrapone v cLCMS/MS

A D

B E

C F
 The Way Forward...
• NEQAS gave participants a ‘What-if’ – Mass Spec as Target
report on their own Cortisol performance at Distribution 422
using data from Distributions 401 through to 422.
Intend to use this reference method on all Cortisol pools used in
the Scheme
• Thrust to move towards fully validating the target values used
by all schemes.
• Periodic, retrospective use, of reference values may well be the
practical way forward as batching pools for reference analysis is
a much more cost effective approach than ‘real-time’ reference
values. ‘Reference methodS’ can still have batch-to-batch
‘wobble’, though this is less of a consideration.
• If we validate the ‘field method’ MSMS mean, we could use
this as a real-time target. This is becoming more of a possibility
as we now have 10 MSMS users
 The way forward …..
All analytes
• For all analytes we are continuing investigating replacing
the ALTM with a better estimate of the truth as target.
• This follows on from use of ‘Restricted ALTMs’ or MSMS
means for some analytes and (because of the number of
results we were dealing with for some analytes) for
17OHP we moved straight to a MSMS target.
• A validated ‘field method’ MSMS target, like we have for
Female Testosterone, is a good and reliable target and
after April 2016 we may be using this ‘field method’
MSMS mean this as the target for all methods for Female
Testosterone.
• Immunoassay users may find this a bit of a challenge.
LC-MS/MS wider application in larger clinical
labs

With greater specificity of LC-MS/MS, the

cortisol response to SST and LDDST may need to
be redefined
 Estimated lower limit normal for 30min
post synacthen cortisol by method

Assay Males Females* M+F Female#

(non-OCP) (OCP)

GCMS 418 421 420 643

Siemens Centaur 448 446 446 619

Abbott Architect 430 416 NA 577

Roche E170 Gen1 574 524 NA 791

Siemens Immulite (2K) 469 478 474 688

Beckman Access 459 455 NA 604

 Summary
• Routine serum cortisol immunoassay performance
remains variable
• Increasing awareness of the impact of assay-specific
bias on result interpretation it is unrealistic to
expect all users (clinicans) to be cognisant of their
limitations
• Case for assay & gender specific cut-offs been made
but with reagent, calibration and assay
reformulation changes this may increase the
uncertainty; so welcome PATH (Partnership accurate
testing of hormones)