Journal of Huntington’s Disease 6 (2017) 349–361 349

DOI 10.3233/JHD-170269
IOS Press

Research Report

Validation of Ultrasensitive Mutant

Huntingtin Detection in Human
Cerebrospinal Fluid by Single Molecule
Counting Immunoassay
Valentina Fodalea,b,# , Roberto Boggiob,1,# , Manuel Daldina , Cristina Cariuloa ,
Maria Carolina Spieziaa , Lauren M. Byrnec , Blair R. Leavittd , Edward J. Wildc ,
Douglas Macdonalde,∗ , Andreas Weissb,2,# and Alberto Bresciania,#
a IRBM Science Park, Pomezia, Rome, Italy
b IRBM Promidis, Pomezia, Rome, Italy
c University College London (UCL) Institute of Neurology, National Hospital for Neurology

and Neurosurgery, London, UK

d Department of Medical Genetics, Centre for Molecular Medicine and Therapeutics,

University of British Columbia, Vancouver, BC, Canada

e CHDI Management/CHDI Foundation, Los Angeles, CA, USA

Abstract.
Background: The measurement of disease-relevant biomarkers has become a major component of clinical trial design,
but in the absence of rigorous clinical and analytical validation of detection methodology, interpretation of results may
be misleading. In Huntington’s disease (HD), measurement of the concentration of mutant huntingtin protein (mHTT) in
cerebrospinal fluid (CSF) of patients may serve as both a disease progression biomarker and a pharmacodynamic readout
for HTT-lowering therapeutic approaches. We recently published the quantification of mHTT levels in HD patient CSF by a
novel ultrasensitive immunoassay-based technology and here analytically validate it for use.
Objective: This work aims to analytically and clinically validate our ultrasensitive assay for mHTT measurement in human
HD CSF, for application as a pharmacodynamic biomarker of CNS mHTT lowering in clinical trials.
Methods: The single molecule counting (SMC) assay is an ultrasensitive bead-based immunoassay where upon specific
recognition, dye-labeled antibodies are excited by a confocal laser and emit fluorescent light as a readout. The detection
of mHTT by this technology was clinically validated following established Food and Drug Administration and European
Medicine Agency guidelines.
Results: The SMC assay was demonstrated to be accurate, precise, specific, and reproducible. While no matrix influence
was detected, a list of interfering substances was compiled as a guideline for proper collection and storage of patient CSF
samples. In addition, a set of recommendations on result interpretation is provided.

1 Current address: Merck Serono at RBM S.p.A., Istituto di # Theseauthors contributed equally to this work.
Ricerche Biomediche A. Marxer, Colleretto Giacosa (TO), Italy. ∗ Correspondence to: Douglas Macdonald, 6080 Center Drive,
2 Current address: Evotec AG, Manfred Eigen Campus, Suite 100, Los Angeles, CA 90024, USA. Tel.: +1 310 342 5508;
Hamburg, Germany. E-mail: douglas.macdonald@chdifoundation.org.

ISSN 1879-6397/17/$35.00 © 2017 – IOS Press and the authors. All rights reserved
This article is published online with Open Access and distributed under the terms of the Creative Commons Attribution Non-Commercial License (CC BY-NC 4.0).
350 V. Fodale et al. / Ultrasensitive Mutant HTT Detection in Human CSF

Conclusions: This SMC assay is a robust and ultrasensitive method for the relative quantification of mHTT in human
CSF.

Keywords: Assay validation, biomarker, cerebrospinal fluid, huntingtin protein, Huntington’s disease, immunoassay, mutant
huntingtin, polyglutamine, ultrasensitive assay

INTRODUCTION protein over the wild type protein (wtHTT) by using

recombinant protein standards.
The availability of biomarkers can greatly facili- The present work aims at providing a detailed ana-
tate the interpretation of clinical trial results, but the lytical clinical validation of the SMC assay based
use of poorly validated assays to detect them has on the 2B7-MW1 antibody pair in order to provide
resulted in a profound lack of useful biomarkers. a strong data foundation for its use in future clin-
The biomarker pipeline was recently described as ical trials. To this end, the 2B7-MW1 assay was
“plagued by problems” and “too prone to failures” preliminarily tested for its specificity for detection of
by Ioannidis and Bossuyt [1]. In the present work mHTT over wtHTT recombinant protein. Hereafter,
we provide a rigorous analytical clinical validation the Food and Drug Administration (FDA) and the
of an assay we previously demonstrated to be able European Medicines Agency (EMA) bioanalytical
to detect mutant huntingtin (mHTT) in human cere- method guidelines [10, 11] were applied to tech-
brospinal fluid (CSF) of Huntington’s disease (HD) nically validate the 2B7-MW1 assay by SMC in
mutation carriers [2]. human CSF samples. This assay, enabled by the Sin-
HD is a neurodegenerative condition caused by gulex Erenna platform, is relative quantitative and
the expansion of a CAG nucleotide repeat domain in the mHTT amount present in biological samples is
the HTT gene [3]. This results in the expression of a calculated against a purified recombinant mHTT pro-
polyglutamine expanded huntingtin protein (mHTT) tein standard curve [2]. Our validation of the assay
that ultimately causes neuronal death [4]. This fact, comprised the evaluation of calibration curve perfor-
together with HD being a penetrant monogenic mance, accuracy, precision, stability, matrix effects,
disease, strengthens the concept of decreasing mHTT selectivity, specificity, and reproducibility.
levels as the most proximal therapeutic strategy for
disease modification [5]. To this aim, the quantifica- MATERIALS AND METHODS
tion of mHTT in CSF, an accessible central nervous
system (CNS) related body fluid, may be informa- Human CSF and blood samples
tive not only as a biomarker for patient stratification,
but also as a target engagement pharmacodynamic Human CSF and blood samples were collected
measure for mHTT-lowering therapeutics, such as from healthy and HD patients, at University Col-
RNAi modalities [6, 7]. We recently described a lege London (UCL) by Dr E. Wild and human
novel single molecule counting (SMC) method capa- CSF samples were collected from healthy and HD
ble of detecting and quantifying mHTT in human patients at the Centre for Molecular Medicine and
and animal model CSF [2]; subsequently another Therapeutics of Vancouver (BC, Canada) by Dr
method using microbeads-based immunoprecipita- B.R. Leavitt as previously described [2]. All work
tion followed by flow cytometry (IP-FCM) [8] was involving human volunteers was performed in accor-
reported. Both techniques are immunoassays based dance with the Declaration of Helsinki of 1975 and
on antibody pairs; in each method, one detects total approved by the Central London Research Ethics
HTT and the other preferentially detects mHTT medi- Committee and the University of British Columbia
ated by the expanded polyglutamine recognition. The Clinical Research Ethics Board. All participants pro-
polyglutamine directed antibody is MW1 [9] in both vided written informed consent.
the SMC and IP-FCM assays, whereas the HTT spe- In order to perform the assay, the samples and ref-
cific antibody 2B7 is used in the SMC assay, and erence proteins were diluted in artificial CSF (aCSF):
HDB4E10 is used in the IP-FCM assay. To date, PBS, 300 mM NaCl, 6 mM KCl, 2.8 mM CaCl2 ,
the extent of validation that has been carried out for 1.6 mM MgCl2 .
these assays has been focused on the demonstration Human blood, serum, and plasma with EDTA K2,
of selective and specific recognition of the mHTT EDTA K3, Na-citrate, Na-Heparin and Na-EDTA
 V. Fodale et al. / Ultrasensitive Mutant HTT Detection in Human CSF 351

were purchased from Seralab in order to test solution of 1X Erenna System Buffer (from Singulex-
the assay selectivity. Hemoglobin A quantification Millipore) 750 mM NaCl, 6% BSA, 0.8% Triton-X
was performed using a commercial ELISA (Bethyl and protease inhibitor cocktail (Complete tablets
Laboratories) according to the manufacturer’s speci- from Roche). Washes were performed on a magnetic
fication. rack, using an Elx washer (from Biotek), in 1X Erenna
Wash Buffer (from Singulex-Millipore). Afterwards
HTT silencing in human HD fibroblasts MPs were incubated with MW1 detection antibody
for 1 h. Washes were performed again on a magnetic
Three primary fibroblast cells, collected from an rack, in an Elx washer, in 1X Erenna Wash Buffer.
HD, JHD, and a compound heterozygous HD patient, The MPs were then transferred into a new Axygen 96
having 45/23, 176/23, and 50/40 glutamines respec- well plate to eliminate nonspecific binding to the plas-
tively, were obtained from The Coriell Institute for tic. After buffer aspiration the elution buffer (acidic
Medical Research, Camden, NJ, USA (cat. num. glycine solution, 0.1 M, pH 2.7) was added to the plate
GM09197, GM01085, GM04857). HTT silencing for 5 min of incubation under shaking. The eluted
was carried out using HTT-specific siRNA from detection antibody was transferred to a Nunc 384-
SIGMA (cat. num. SASI HS01 00241076) and Lipo- well analysis plate and neutralized with neutralization
fectamine 2000 (Invitrogen) as transfecting reagent. buffer (Tris, 1 M, pH 9). The plate was spun down,
Transfected cells were collected 48 h after treatment sealed and subsequently analyzed with the Erenna
and lysed. Lysis was done in PBS, 0.4% Triton-X Immunoassay System (Singulex-Millipore).
and Protease Inhibitor cocktail tablets (complete from
Roche). Data analysis

Antibodies and recombinant proteins Data analysis was performed using GraphPad
Prism software in order to obtain standard curve
The MW1 antibody was developed by the late fitting and back-calculations on fitting models. The
Dr. Paul Patterson [9]. 2B7 antibody generation and standard curve was obtained without associating any
characterization were as previously described [12]. weight to each standard concentration.
The 2B7 antibody was conjugated to magnetic par-
ticles (MPs), to a final concentration of 25 ␮g/mg of
MPs, and the MW1 antibody was labeled, to a final RESULTS
concentration of 1 ␮g/␮l, using the Erenna capture
and detection reagent labeling kits from Singulex- Assay technology, antibodies, and detected
Millipore, following the manufacturer’s instructions. signals
Conjugated/labeled antibodies were diluted in Assay
Buffer (from Singulex-Millipore), prior to perform- Singulex Erenna technology provides high sensi-
ing the assay, to 1:400 and 1:1000 respectively. tivity with a broad dynamic range [15] in a 384-well
Purified recombinant proteins were obtained from plate format. The signal detected is a train of events
CHDI Foundation and produced as previously pub- generated when dye-labeled molecules are excited
lished. The large fragment amino terminal human by the confocal laser of the instrument and emit fluo-
HTT proteins HTT (1-573) Q23, HTT (1-573) Q45, rescent light. The instrument measures three outputs
and HTT (1-573) Q73 were produced according to based on photon detection. The output best suited
Macdonald et al. [13], and the full length human HTT to the generation of dilution curves described by
proteins HTT (1-3144) Q17 and HTT (1-3144) Q46 the standard protein is event photons (EP), which is
were produced according to Huang et al. [14]. the average photon count in all detected events. To
perform the immunoassay, two anti-HTT antibod-
Immunoassay ies were used. The 2B7 antibody binds to the N17
region of HTT [12] and was conjugated to magnetic
The immunoassay was performed as previously particles for use as the capture antibody. The MW1
described [2]. Briefly, the assay starts with a capture antibody recognizes the polyglutamine expansion of
step, where the 2B7 conjugated-MPs are incubated mHTT [9] and was labeled with a fluorophore and
with samples for 1 h, in an Axygen polypropylene used as the detection antibody. Both antibodies are
V-bottom 96 well plate, previously coated with a well-established in the detection of mHTT by ELISA
352 V. Fodale et al. / Ultrasensitive Mutant HTT Detection in Human CSF

or time-resolved Förster resonance energy transfer pathological polyglutamine expansions of similar

(TR-FRET) in biological samples where the mHTT length (Q46 and Q45, respectively). The dynamic
protein is relatively abundant, such as cellular lysates range is defined as lying between the lower and upper
or tissue homogenates [12, 13, 16]. limits of quantification (LLoQ and ULoQ): the lowest
and highest points of the curve for which the per-
Calibration curve performance centage relative error (systematic error or bias from
nominal value – % RE) and the random error (impre-
Calibration standards were prepared by spiking cision or percentage coefficient of variation – % CV)
known amounts of purified human recombinant are ≤25% [11, 17]. The determined LLoQ and ULoQ
HTT proteins in aCSF, a CSF surrogate matrix that were 6.5 and 8000 fM for the large fragment, and 16.5
matches the physiological electrolyte concentrations and 8000 fM for the full-length protein (Fig. 1B).
of human CSF. Five forms of human recombinant It has to be noted that, the same concentration of
HTT protein were tested: three large fragment pro- the full length and the large fragment protein, bear-
teins comprising the 1-573 N-terminal amino acids ing almost the same polyglutamine stretch (Q46 and
(N573) and bearing polyglutamine expansions of 23, Q45, respectively), were detected with one log differ-
45 and 73 glutamines in Exon 1; and two full length ence in intensity with the large fragment producing
proteins comprising the 1-3144 amino acids (FL) pro- a higher signal than the full length (Fig. 1B). This
teins with 17 and 46 glutamines in Exon 1. Signals observation suggests that HTT detection is not only
obtained by serially diluting the N573 fragments and influenced by the length of polyglutamine expansion,
FL proteins, starting from 4 pM and 25 pM respec- but also by the size of protein.
tively, demonstrated the specificity of the assay in In order to identify a proper fitting function, data
detecting the mutant forms of HTT and its polyglu- points relative to the N573 Q45 protein were fit-
tamine dependency. In fact, the mHTT forms (N573 ted with linear regression, four-parameter logistic
Q45, Q73 and FL Q73) were each well-detected in the (4PL) and five-parameter logistic (5PL) models [18]
assayed range of concentrations, with detection per- (Fig. 2A-C). The values calculated by the three fitting
formance improving with polyglutamine length (i.e., models were compared with the ones measured by the
N573 Q73 >N573 Q45). On the contrary, the detec- assay determining the % RE and the % CV. The sum
tion of wild-type HTT forms (N573 Q23 and FL Q17) of the % RE and the % CV is defined as the percentage
is ineffective or just above the limit of detection (LoD) total error (% TE). The model that best fits the protein
at the highest tested concentrations (Fig. 1A). detection is the one where the majority of the dilution
The calibration curve performance was carried out points (at least four of six points) do not exceed the
by using either FL or N573 HTT proteins bearing 30% of TE; this rule is known as the 4-6-30% rule

Fig. 1. Calibration curve performance and definition. Detection of human recombinant (hr) full length (with 17 and 46 glutamines in the
polyglutamine stretch) and N573 (with 23, 45, and 73 glutamines in the polyglutamine stretch) HTT protein by the 2B7-MW1 SMC assay.
A) Polyglutamine and protein length-dependency in HTT detection were demonstrated analyzing the five proteins. B) A longer series of
protein dilutions was analyzed for both the full length Q46 and the N573 Q45 in order to define the LLoQ and ULoQ of the assay. These two
values were calculated as the lowest (above the LoD) and the highest points of the standard curve for which % CV and % RE are <25 [19].
They are 6.5 and 10000 fM for the large fragment and 16.5 and 10000 fM for the full length protein. The X and Y axes have logarithmic
scales and the curves were fitted with a parameter logistic (5PL) model. Each point is the mean of 3 replicates. Bars represent standard
deviation. LoD was calculated as the mean of the blank EP values plus three times the standard deviation.
 V. Fodale et al. / Ultrasensitive Mutant HTT Detection in Human CSF 353

Fig. 2. Standard curve fitting model evaluation. EP values obtained from the 2B7-MW1 SMC analysis of serial dilutions of N573 Q45 HTT
protein were fitted using (A) linear regression, (B) four-parameter logistic (4PL), and (C) 5PL models. Each point is the mean of 3 replicates.
Bars represent standard deviation. (D) % TE and % RE were evaluated for each dilution point for the three fitting models. The 4PL and
5PL models achieved less than 30% error for points within the quantification range of the assay, as required by the EMA assay validation
guideline. For the 5PL model, all assayed dilutions including the one below the LLoQ, showed error <30%.

originally described by DeSilva et al. and adopted by Specificity for human endogenous HTT protein
the EMA assay validation guidelines [11, 18, 19]. In
our study, the linear regression model was demon- We initially proceeded to investigate whether the
strated to be inadequate since using this model, 50% assay is suitable for specifically detecting endoge-
of the analyzed points show a % TE >30 (Fig. 2A, D). nous mHTT protein by conducting an HTT detection
The 4PL model instead was shown to be compliant specificity study in cells. To this aim we used pri-
with the above-mentioned rule for the dilution points mary fibroblasts from one juvenile HD (JHD) and two
within the quantification range of the assay (Fig. 2B, adult-onset HD patients expressing mHTT containing
D); the 5PL model was compliant with the rule for all 176/23, 45/23, 50/40 glutamines, respectively. These
the dilution points (including those under the LLoQ) cells were transfected with an HTT-specific siRNA
and was thus used in the following validation steps and a scrambled control siRNA, then lysed and col-
(Fig. 2C, D). lected after 48 h. HTT silencing was achieved to a
level between 50–60% of the scrambled transfected
cells, as determined by quantitative RT-PCR mea-
Accuracy and precision of the standard curve
suring HTT mRNA (data not shown). When mHTT
protein levels were quantified using the 2B7-MW1
The accuracy and precision of the standard
SMC assay on silenced and control cells a decrease
curve for detection of mHTT were investigated by
of the target protein, similar to the observed mRNA
within-run (intrabatch) and between-run (interbatch)
decrease, was detected. This provides a preliminary
assessment. Within-run evaluation was carried out
demonstration of the assay’s validity in complex
by testing three technical replicates of each stan-
matrices (Fig. 3).
dard concentration in a single run, while between-run
evaluation was performed by at least six indepen-
dent standard curves assayed on different days. The Analyte stability
obtained results show that % RE (Table 1A) and %
CV (Table 1B) were less than 20% in all the batch To demonstrate that the 2B7-MW1 SMC assay is
runs for mHTT concentrations between the LLoQ suitable for the study of biological samples, accu-
and ULoQ. These results are compliant with the EMA racy and precision have to be assessed on validation
guidelines for bioassay validation [11]. For CV and samples (VS). VS are real biological samples pre-
RE calculation, intrabatch precision was estimated by pared once and then tested, with the same standard
the pooled intrabatch standard deviation of the mean operating procedure as all investigational samples,
of the calculated concentrations, while interbatch pre- in every single analyte quantification study to val-
cision (total random error) was estimated by analysis idate the performance of the assay. At least five
of variance (ANOVA). Statistical methods and for- VS should be used to assess accuracy, precision
mulas used were as described in detail by DeSilva and the total error of the method: an anticipated
et al. [18]. ULoQ control (VS1); a high control (VS2); a mid-
354 V. Fodale et al. / Ultrasensitive Mutant HTT Detection in Human CSF

Table 1
Accuracy and precision. Accuracy and precision of the 2B7-MW1 SMC mHTT (N573 Q45) detection assay were investigated within-run
(intrabatch), by testing three technical replicates of each standard concentration, and between-run (interbatch), by testing six independent
standard curves. % RE (A) and % CV (B) measured for each standard concentration in each run are reported. Both values were demonstrated
to be less than 20% for all the concentrations between LLoQ and ULoQ for each run. Intrabatch and Interbatch % CV and % RE were
estimated using the statistical methods described by DeSilva et al. [18]. Both interbatch and intrabatch precision and accuracy in %CV and
%RE calculation were demonstrated to be <20%
A. Percent Relative Errors (%RE) of Back-Calculated Standard Concentration
Nominal Concentration (fM)
Batch Run 4000 1600 640 256 102 41 16
1 –1.15 2.01 5.07 –10.50 7.20 4.41 9.31
2 1.85 –3.63 0.61 –1.89 9.97 –6.72 1.16
3 –0.47 4.40 2.87 –7.77 4.25 4.66 –6.32
4 2.80 –4.29 10.67 –11.82 10.09 1.11 –3.66
5 7.58 –10.10 4.35 –0.04 7.44 –6.58 1.83
6 0.72 –4.50 8.86 –7.64 4.21 1.74 –1.93
Intrabatch (within-run) Statistics (Pooled):
1.89 –2.39 5.09 –6.61 7.20 –0.23 –0.08
Interbatch (between-run) Statistics (ANOVA):
1.89 –2.68 5.40 –6.61 7.20 –0.23 0.06
ULOQ LLOQ
B. Percent Coefficient of Variation (%CV) of Back-Calculated Standard Concentration
Nominal Concentration (fM)
Batch Run 4000 1600 640 256 102 41 16
1 4.80 8.22 11.26 14.32 9.89 13.18 29.71
2 3.59 6.53 17.70 12.72 8.54 10.97 5.27
3 15.98 16.42 16.01 12.97 5.24 9.10 2.58
4 18.42 1.83 2.22 11.65 9.65 4.39 5.08
5 9.61 14.89 10.41 13.05 7.53 16.06 8.81
6 8.81 3.38 16.54 16.67 8.40 18.17 3.61
Intrabatch (within-run) Statistics (Pooled):
11.55 11.43 14.02 13.66 8.35 12.81 11.92
Interbatch (between-run) Statistics (ANOVA):
12.96 12.31 15.91 15.05 9.29 13.86 13.28
ULOQ LLOQ

control (VS3); a control that is more than three times in VS after two freeze-thaw cycles to a % RE between
the LLoQ (VS4); and an anticipated LLoQ control 80 and 120% (Fig. 4C). As a consequence, the addi-
(VS5) [11]. These VS were prepared by pooling CSF tion of 1% Tween-20 is recommended to preserve
from healthy and HD donors and, finally, spiking the integrity of mHTT by preventing possible ex vivo
recombinant mHTT protein in order to obtain the precipitation, oligomerization, or aggregation.
five desired mHTT concentrations as presented in
Fig. 4A. Parallelism, dilution linearity, and spike recovery
VS were also evaluated for their stability after
two freeze-thaw cycles by measuring their variation The parallelism test is used to demonstrate that the
(% RE) from the initial mHTT concentration. The endogenous analyte in its biological matrix behaves
obtained recovery rate was less than 80% after the in a similar immunochemical manner to the standard
first freeze/thaw cycle and decreased even more than protein in the same matrix or in a substituted one
50% after the second cycle (Fig. 4B). One possible [19]. In the present work, parallelism was assessed
determinant of this loss may be the propensity of between the calibration standard curve and serially-
mHTT to form aggregates by hydrophobic interac- diluted HD CSF samples in order to exclude possible
tions [20]. For this reason, the addition of a detergent off-target affinities for other matrix resident analytes
was evaluated to prevent mHTT aggregation. Indeed, and to validate the use of aCSF as substituted matrix
1% Tween-20 stabilized the concentration of mHTT for the standard curve [19]. A preliminary evaluation
 V. Fodale et al. / Ultrasensitive Mutant HTT Detection in Human CSF 355

of parallelism was made using the similarity between dilution series does not exceed 30% [11]. Results are
the curve shape of the CSF dilutions, where at least represented in Fig. 5B, where the calculated concen-
three points fell within the quantification range, and trations (observed concentration × dilution factor)
the standard mHTT curve (Fig. 5A). Parallelism were divided by the mean of the concentrations and
was assessed verifying that the variation (% RE) plotted against the inverse of the dilution factor [19].
between back-calculated sample concentrations in a The calculated ratio was not affected by sample dilu-
tion for the four samples which were detected within
the quantification range of the assay. Back-calculated
mHTT levels were found to not exceed 20% RE in
those samples, thus validating the dilution linearity of
the assay and consequently its parallelism. Further-
more, the two CSF samples from premanifest HD
patients showed a % RE within the 30%, even if they
were below the LLoQ, demonstrating that a matrix
effect is not observed in real CSF samples using the
2B7-MW1 SMC assay.
mHTT spike recovery was carried out to evaluate
the influence of the CSF matrix in mHTT detection
by the assay [19]. Recombinant mHTT protein N573
Q45 was spiked and diluted in aCSF and in five real
Fig. 3. Specificity for human endogenous HTT protein. The 2B7-
MW1 SMC assay specificity was demonstrated by analyzing HTT- CSFs collected from HD patients. The spike recov-
silenced fibroblasts derived from one JHD and two adult-onset HD ery rate percentage was calculated by comparing the
subjects. Cells were transfected with an HTT-specific siRNA and a assay-determined mHTT amount to a nominal one,
scrambled control. The decrease in mHTT protein levels, as back-
calculated on the N573 Q45 standard protein, detected in siRNA
after subtracting the sample basal mHTT levels [19,
treated cell lines, with respect to the scramble treated ones, was 21]. All mHTT concentrations spiked in the six matri-
found to be coherent with the decrease in HTT mRNA (data not ces were recovered within the acceptable range of
shown). 80–120 % (Fig. 6).

Fig. 4. Stability evaluation. Signals obtained from five validation samples (VS) with mHTT levels distributed across the quantification range
and their stability. (A) N573 Q45 standard dilutions (black dots) are interpolated with a 5PL model. VS1 anticipates ULoQ, VS2 is a high
control, VS3 is a mid-control, VS4 is less than three times the LLOQ and VS5 anticipates LLoQ [11]. Each point is the average of 3 replicates.
Bars represent standard deviation. (B, C) Variations in the detection of mHTT in VS detected by the 2B7-MW1 SMC assay on fresh samples
after storage at –80◦ C (first cycle of freeze/thaw), or after 6 h at room temperature (RT), or after the second cycle of freeze/thaw, (B) without
preservative or (C) with the addition of 1% Tween. The addition of 1% Tween greatly improved the stability of HTT in all tested conditions
(the accuracy % RE remained between 80 and 120 %).
356 V. Fodale et al. / Ultrasensitive Mutant HTT Detection in Human CSF

Fig. 5. Parallelism evaluation. Six CSF samples were used: four CSF from advanced HD patients and two CSF from pre-manifest HD
patients. (A) N573 Q45 standard dilutions (black dots) are interpolated with a 5PL model and shown in each plot together with the signal
obtained from 2B7-MW1 SMC analysis of the CSF. X and Y axes have logarithmic scales and the curves were fitted with a 5PL model. Each
point is the average of 3 replicates. Bars represent standard deviation. (B) Parallelism was demonstrated by dilution linearity. Calculated
mHTT concentrations (observed concentration × dilution factor) were divided by the mean of the concentrations and plotted against the
inverse of the dilution factor [19]. The %RE among the dilution points was <20% for all the samples detected over the LLoQ and <30% for
the two samples detected under the LLoQ.

Fig. 6. Spike recovery evaluation. Seven dilutions of the N573 Q45 recombinant protein were spiked in aCSF and in five CSF samples
collected from HD patients. (A) The obtained curves were fitted with a 5PL model. The X and Y axes have logarithmic scales. Each point
is the mean of 3 replicates. Bars represent standard deviations. (B) The spike recovery rate of five HTT dilutions was between 80 and 120
% of the nominal concentration in all the tested matrices.

Selectivity is dependent on accidental CSF contamination dur-

ing sampling by other tissues such as blood. Both
The selectivity of the 2B7-MW1 SMC assay was the FDA and the EMA guidelines for bioanalyti-
evaluated in order to demonstrate the suitability of cal method validation suggest performing selectivity
the assay to detect mHTT unequivocally even in the tests in real biological matrices; nonetheless the vol-
presence of additives and/or contaminants in the tar- ume of human CSF required to complete the study
get matrix [10, 19]. The former deals with chemicals with all the selected contaminants was incompatible
added to CSF samples to preserve them; the latter with the availability of HD CSF. To overcome this
 V. Fodale et al. / Ultrasensitive Mutant HTT Detection in Human CSF 357

Fig. 7. Selectivity evaluation. The selectivity of the 2B7-MW1 SMC assay was assessed against multiple concentrations of potential
interfering substances. Different concentrations of eight components were spiked into artificial CSF: (A) plasma with EDTA K2, EDTA
K3, sodium citrate, sodium heparin and sodium EDTA, serum and (B) blood collected from six HD subjects and one healthy control. EP
signals are reported with respect to the % spike in aCSF in A, or against human hemoglobin concentration in B. Each point is the mean of 3
replicates. Bars represent standard deviations. Numeric results are summarized in Table 2.

limitation, aCSF was used instead of HD CSF. Our While CSF is normally centrifuged after sampling
findings for parallelism, dilution linearity and spike in order to remove blood cells, soluble components of
recovery, described above, demonstrate the suitabil- blood and the contents of already-hemolyzed blood
ity of aCSF as a matrix surrogate in the determination cells cannot be removed. The presence of hemolyzed
of the assay’s selectivity for mHTT. blood was found to impact the assay process by
Selectivity was assessed by analyzing multiple causing the magnetic particles to clump. In addi-
concentrations of the following potential interfering tion to this general hemolyzed blood interference,
substances: human plasma pooled from healthy donor blood contamination may alter mHTT levels in CSF
with EDTA K2, EDTA K3, sodium citrate, sodium by specifically adding mHTT to the CSF. To assess
heparin and sodium EDTA (which are commonly this possibility, blood from six HD patients was seri-
used as anti-coagulants and sample preservatives), ally diluted in aCSF, starting from a concentration
serum and whole blood (which can be present in CSF of 0.15%, and analyzed by the 2B7-MW1 Singulex
due to sampling contamination). assay. Hemoglobin (HbA) levels for all blood dilu-
The obtained results demonstrated that up to 10% tions were measured and employed as an independent
plasma with sodium heparin or serum did not affect means of quantifying blood contamination that could
the assay signal. In contrast, plasma with EDTA then be used to establish a hemoglobin threshold
K2, EDTA K3, sodium citrate, and sodium EDTA below which contamination of CSF by mHTT from
concentrations higher than 2, 0.4, 0.3, and 0.08% blood is expected to be negligible.
respectively resulted in signals higher that the LoD The correlation between mHTT and hemoglobin
(Fig. 7A), suggesting that they should be avoided levels is reported in Fig. 7B. An appreciable signal
as CSF preservatives. A summary of all the additive was detected at blood spikes greater than 0.002%,
thresholds is reported in Table 2. corresponding to 2 ␮g/mL of hemoglobin which, as
a consequence, was fixed as threshold for mHTT
Table 2 quantification in HD CSF.
Summary of selectivity test results. Contaminant interference in All the thresholds of the tested contaminants,
the detection of mHTT and the corresponding threshold concen-
trations for each tested substance
which may interfere with mHTT detection by the
2B7-MW1 SMC assay, are summarized in Table 2.
Presevative/Contaminant mHTT detection Acceptance
interference criteria
Plasma w EDTA K2 Yes ≤2%
Reproducibility
Plasma w EDTA K3 Yes ≤0.40%
Plasma w Na-citrate Yes ≤0.30%
Plasma w Na-Heparin No up to 10% The reproducibility of an assay is determined by
Plasma w Na-EDTA Yes ≤0.08% the incurred sample reanalysis (ISR), which is the
Serum No Up to 10% reanalysis of a portion of subject samples to deter-
Hemoglobin Yes ≤2 ␮g/ml
mine whether the original results are reproducible
358 V. Fodale et al. / Ultrasensitive Mutant HTT Detection in Human CSF

[10, 11]. The ISR was carried out on 24 HD CSF sam- perform the reference standard is detected in a similar
ples (i.e., 10% of a putative 240-participant study) biochemical manner to endogenous mHTT protein.
analyzed in triplicate. Three healthy CSF samples The assay was demonstrated to be specific for mHTT
were included as negative controls. The time between with a lower limit of detection of 6.5 fM for the N573
the first and the second analysis was set to one week, Q45 and 16.5 fM for the FL Q46. It has to be noted
whereas five weeks elapsed between the first and that the assay better detects the shorter form of mutant
the third analysis. After collection, samples were HTT (N573) than the full length one (FL) even though
aliquoted in single-use vials to avoid multiple freeze- the two were almost identical in terms of polyglu-
thaw cycles among the different runs. The three tamine expansion. This fact may be relevant when
control sample results were lower than the LLoQ, interpreting results, as there are reports showing that
confirming the assay specificity. One HD CSF sample endogenous mHTT is present in various truncated
result was lower than the LLoQ; thus it was excluded forms in cells resulting from proteolytic cleavage as
from the ISR study. Hemoglobin levels in the remain- well as proteosomal degradation [24–26].
ing samples were below the 2 ␮g/mL threshold with CSF matrix composition was demonstrated not to
the exception of one sample, which was excluded interfere with the assay throughout a wide range of
from the ISR study (see Table 3). The RE for each dilutions. Still, we recommend the addition of 1%
run was calculated with respect to the average mHTT Tween-20 to CSF, at the time of sampling or after the
concentration. 77.3% of tested samples showed a RE first thawing, to retain the integrity of the biosample
within 30% (Table 3), which is compliant with the by avoiding possible loss of mHTT due to protein
assay validation guidelines for large molecules, being precipitation, oligomerization, or aggregation. When
greater than the two-thirds (67%) of the incurred sam- various additives were tested for their interference,
ples. [10, 11]. plasma with EDTA K2, EDTA K3, Na-citrate, or NA-
EDTA were found to impact on mHTT quantification;
thus their use should be avoided. With regard to
DISCUSSION contaminant influence, hemolyzed blood was demon-
strated to be a potential major issue when hemoglobin
One of the more promising therapeutic approaches exceeds 2 ␮g/ml due to both matrix effect and serum
to mitigate or modify the course of HD is based mHTT contamination: thus we recommend that CSF
on various techniques to lower the expression of samples containing hemoglobin levels above this
mHTT protein. In order to enable such therapeutic threshold be avoided. Finally, the assay was demon-
approaches, biomarkers of HTT lowering must be strated to be reliable as mHTT levels in 77.3% of a
developed to demonstrate that delivery of a HTT low- 22 pre-/early manifest group were consistently quan-
ering agent does indeed lower the amount of HTT tified over three independent runs.
protein in the brain of an HD patient. Since the Despite our demonstration that the 2B7-MW1
brain cannot be non-invasively used for a pharmaco- SMC assay is accurate and precise, there are some
dynamic examination, the analysis of brain-derived recommendations that should be taken into account
proteins enriched in patient CSF can provide a “win- when interpreting data produced by the present pro-
dow into the brain” [6, 22, 23] and represents the most cedure. First, it has to be considered that the output of
accessible opportunity to biochemically sample the mHTT quantification is dependent, to some extent, on
CNS milieu. two important factors: the poly-glutamine expansion
The aim of the present work was to clinically val- size and the protein size. Therefore, in absence of a set
idate the approach proposed by Wild and Boggio of standard recombinant mHTT proteins with every
et al. that is based on the use of the 2B7-MW1 SMC single possible combination of polyglutamine expan-
assay as a biomarker for HTT lowering therapeu- sions and protein fragments, the calculated mHTT
tic approaches. The validation process was carried concentration must be considered as a best estimate
out following the recommendations of both the FDA rather than an absolute value; as a consequence the
and EMA [10, 11]. One partial limitation complicat- assay can be defined as relatively quantitative [17]. A
ing this validation study was the limited availability consequence is that the most meaningful results will
of HD patient CSF. Nonetheless, the parallelism, likely be obtained in paired analyses where a single
dilution linearity, and spike recovery studies demon- subject is monitored over time – such as within the
strated that patient CSF matrix can be substituted by setting of a clinical trial. A further consideration is
aCSF and that the recombinant mHTT protein used to the choice between N573 and FL standard mHTT.
 Table 3
Reproducibility evaluation. Reproducibility of the 2B7-MW1 SMC mHTT detection assay was successfully investigated by testing 22 CSF samples, collected from HD patients, in three independent
runs. The mean of mHTT concentrations obtained in each of the three runs is reported, together with CV % and RE %. The RE % was calculated with respect to the average of concentration
obtained from the three runs (indicated as AVE). Asterisks indicate samples which failed to meet one of the following criteria: HbA <2 ␮g/ml, CV <30% or RE <30%
Sample ID Hemoglobin Average HTT fM AVE SD CV% Note Recovery % w.r.t. AVE RE %
␮g/ml

V. Fodale et al. / Ultrasensitive Mutant HTT Detection in Human CSF

run #1 run #2 run #3 run #1 run #2 run #3 run #1 run #2 run #3
HD 1 0.47 109.09 147.59 90.22 115.63 29.24 25.29 94.35 127.63 78.02 –5.65 27.63 –21.98
HD 2 2.95∗ 98.03 340.09 75.36 171.16 146.74 85.73∗ Hb >2 ␮g/ml excluded from reproducibility analysis
HD 3 0.22 61.02 97.73 63.21 73.98 20.59 27.83 82.48 132.09 85.43 –17.52 32.09∗ –14.57
HD 4 0.33 11.10 14.80 10.63 12.18 2.29 18.76 91.18 121.55 87.27 –8.82 21.55 –12.73
HD 5 0.14 75.88 86.95 78.55 80.46 5.78 7.19 94.30 108.07 97.63 –5.70 8.07 –2.37
HD 6 0.23 116.81 156.66 73.77 115.75 41.46 35.82∗ 100.92 135.35 63.73 0.92 35.35∗ –36.27∗
HD 7 0.34 35.55 48.13 43.15 42.28 6.34 14.99 84.08 113.85 102.07 –15.92 13.85 2.07
HD 8 <LoD 28.55 32.15 35.42 32.04 3.44 10.73 89.10 100.35 110.55 –10.90 0.35 10.55
HD 9 0.09 41.35 43.25 42.50 42.37 0.96 2.26 97.60 102.09 100.30 –2.40 2.09 0.30
HD 10 0.43 32.63 35.75 22.70 30.36 6.81 22.44 107.48 117.74 74.78 7.48 17.74 –25.22
HD 11 0.15 15.23 20.00 233.15 89.46 124.46 139.12∗ 17.03 22.36 260.62 –82.97∗ –77.64∗ 160.62∗
HD 12 0.40 73.27 96.72 86.82 85.60 11.77 13.75 85.59 112.98 101.42 –14.41 12.98 1.42
HD 13 0.09 59.93 69.33 58.37 62.54 5.93 9.48 95.82 110.85 93.33 –4.18 10.85 –6.67
HD 14 0.33 427.62 365.50 220.00 337.71 106.57 31.56∗ 126.63 108.23 65.15 26.63 8.23 –34.85∗
HD 15 0.34 58.15 75.93 64.50 66.19 9.01 13.61 87.85 114.71 97.44 –12.15 14.71 –2.56
HD 16 0.06 22.20 20.53 20.00 20.91 1.15 5.49 106.17 98.19 95.64 6.17 –1.81 –4.36
HD 17 0.18 14.55 20.37 26.57 20.50 6.01 29.32 71.00 99.37 129.63 –29.00 –0.63 29.63
HD 18 0.04 238.00 109.52 57.64 135.05 92.85 68.75∗ 176.22 81.10 42.68 76.22∗ –18.90 –57.32∗
HD 19 0.10 121.61 171.77 116.60 136.66 30.51 22.33 88.99 125.69 85.32 –11.01 25.69 –14.68
HD 20 0.07 3.50 1.06 1.23 1.93 1.36 70.43∗ <LLoQ excluded from reproducibility analysis
HD 21 0.40 82.20 102.18 58.50 80.96 21.87 27.01 101.53 126.22 72.25 1.53 26.22 –27.75
HD 22 <LoD 48.13 56.75 59.30 54.73 5.85 10.70 87.95 103.70 108.35 –12.05 3.70 8.35
HD 23 <LoD 149.29 186.44 149.90 161.88 21.28 13.14 92.23 115.18 92.60 –7.77 15.18 –7.40
HD 24 <LoD 51.57 76.65 64.65 64.29 12.55 19.51 80.21 119.23 100.56 –19.79 19.23 0.56
Healthy 1 <LoD 3.08 4.62 3.25 3.65 0.84 23.12 <LLoQ excluded from reproducibility analysis
Healthy 2 <LoD 3.05 4.79 2.09 3.31 1.37 41.34∗ <LLoQ excluded from reproducibility analysis
Healthy 3 <LoD 3.30 6.02 1.13 3.48 2.45 70.20∗ <LLoQ excluded from reproducibility analysis

359
360 V. Fodale et al. / Ultrasensitive Mutant HTT Detection in Human CSF

Because the performance of these two standards is advisory boards were paid through UCL Consul-
different, mHTT levels can be compared only when tants Ltd, a wholly owned subsidiary of UCL. His
a single standard is used across different runs. More- host clinical institution, University College London
over, in consideration of the propensity of mHTT to Hospitals NHS Foundation Trust, receives funds as
aggregate, the quality of the standard proteins must compensation for conducting clinical trials for Ionis
be periodically controlled by independent methods Pharmaceuticals, Pfizer and Teva Pharmaceuticals.
such as native PAGE. BRL has participated in scientific advisory boards
The present work demonstrated that the presence with Hoffmann-La Roche Ltd, Ionis, Raptor, Teva,
of hemolyzed blood is highly detrimental to the uniQure, and LifeMax. Research funding has been
assay performance. When the contamination is evi- provided to his institution by Teva, uniQure, and
dent it impairs the assay performance, while at minor Lifemax, and he receives funds as compensation for
contamination (>2 ␮g/ml of hemoglobin) errors in conducting clinical trials for Teva and Ionis Pharma-
mHTT determination were observed. Therefore, a ceuticals.
third recommendation would be to always quantify There are no patents, products in development, or
hemoglobin levels in the CSF prior to running the marketed products to declare.
mHTT quantification in order to exclude contami-
nated samples from the study.
REFERENCES
Finally, to better interpret the mHTT concentration
levels, ancillary biomarkers such as the total pro- [1] Ioannidis JPA, Bossuyt PMM. Waste, leaks, and failures
tein amount in the CSF should be used that provide in the biomarker pipeline. Clin Chem. 2017;63:963-72.
an additional characterization of the testing matrix doi:10.1373/clinchem.2016.254649
[27, 28]. [2] Wild EJ, Boggio R, Langbehn D, Robertson N, Haider
S, Miller JRC, et al. Quantification of mutant huntingtin
In conclusion, the present work, guided by FDA protein in cerebrospinal fluid from Huntington’s disease
and EMA principles, demonstrates that the 2B7- patients. J Clin Invest. 2015;125:1-8. doi:10.1172/JCI8074
MW1 assay is a valid, robust, reproducible, and 3DS1
ultrasensitive method for the relative quantification [3] A novel gene containing a trinucleotide repeat that is
expanded and unstable on Huntington’s disease chromo-
of mHTT in human CSF. somes. The Huntington’s Disease Collaborative Research
Group. Cell. 1993;72:971-83.
[4] Leavitt BR, Wellington CL, Hayden MR. Recent insights
ACKNOWLEDGMENTS into the molecular pathogenesis of Huntington disease.
Semin Neurol. 1999;19:385-95.
[5] Aronin N, DiFiglia M. Huntingtin-lowering strategies in
We thank the late Dr. Paul Patterson (CalTech,
Huntington’s disease: Antisense oligonucleotides, small
USA) for the MW1 antibody. RNAs, and gene editing. Mov Disord. 2014;29:1455-61.
This work is dedicated to Dr. Sergio Altamura who [6] Ross CA, Aylward EH, Wild EJ, Langbehn DR, Long
prematurely passed way at the age of 59. At IRBM, JD, Warner JH, et al. Huntington disease: Natural history,
biomarkers and prospects for therapeutics. Nat Rev Neurol.
Sergio was that leader who feels rewarded by mak- 2014;10:204-16.
ing things possible without asking for more. A rare [7] Wild EJ, Tabrizi SJ. Targets for future clinical trials in
person who we miss. Huntington’s disease: What’s in the pipeline? Mov Disord.
EJW is funded by the Medical Research Council 2014;29:1434-45.
[8] Southwell AL, Smith SEP, Davis TR, Caron NS, Vil-
(UK), CHDI Foundation Inc and European Hunting- lanueva EB, Xie Y, et al. Ultrasensitive measurement
ton’s Disease Network. This work was supported in of huntingtin protein in cerebrospinal fluid demonstrates
part by the National Institute for Health Research increase with Huntington disease stage and decrease follow-
University College London Hospitals Biomedical ing brain huntingtin suppression. Sci Rep. 2015;5:12166.
doi:10.1038/srep12166
Research Centre, the UCL Leonard Wolfson Experi- [9] Ko J, Ou S, Patterson PH. New anti-huntingtin mono-
mental Neurology Centre. clonal antibodies: Implications for huntingtin conformation
and its binding proteins. Brain Res Bull. 2001;56:319-29.
doi:10.1016/S0361-9230(01)00599-8
CONFLICT OF INTEREST [10] Food and Drug Administration (2013) Guidance for Indus-
try: Bioanalytical method validation. http://www.labcompli
ance.de/documents/FDA/FDA-Others/Laboratory/f-507-bio
EJW has participated in scientific advisory boards
analytical-4252fnl.pdf.
with Hoffmann-La Roche Ltd, Ionis, Shire, GSK [11] European Medicines Agency, Committee for Medicinal
and Wave Life Sciences. All honoraria for these Products for Human Use. Guideline on Bioanalytical
 V. Fodale et al. / Ultrasensitive Mutant HTT Detection in Human CSF 361

Method Validation. European Medicines Agency, London, [20] Legleiter J, Mitchell E, Lotz GP, Sapp E, Ng C, DiFiglia
UK (2011). http://www.ema.europa.eu/docs/en GB/docu M, et al. Mutant huntingtin fragments form oligomers in
ment library/Scientific guideline/2011/08/WC500109686. a polyglutamine length-dependent manner in vitro and in
pdf vivo. J Biol Chem. 2010;285:14777-90. doi:10.1074/jbc.
[12] Weiss A, Abramowski D, Bibel M, Bodner R, Chopra V, M109.093708
DiFiglia M, et al. Single-step detection of mutant huntingtin [21] Andreasson U, Perret-Liaudet A, van Waalwijk van Doorn
in animal and human tissues: A bioassay for Huntington’s LJC, Blennow K, Chiasserini D, Engelborghs S, et al. A
disease. Anal Biochem. 2009;395:8-15. doi:10.1016/j.ab. practical guide to immunoassay method validation. Front
2009.08.001 Neurol. 2015;6:1-8. doi:10.3389/fneur.2015.00179
[13] Macdonald D, Tessari MA, Boogaard I, Smith M, Pulli [22] Reiber H. Proteins in cerebrospinal fluid and blood: Barriers,
K, Szynol A, et al. Quantification assays for total and CSF flow rate and source-related dynamics. Restor Neurol
polyglutamine-expanded huntingtin proteins. PLoS One. Neurosci. 2003;21:79-96.
2014;9:e96854. doi:10.1371/journal.pone.0096854 [23] Blennow K, Hampel H, Weiner M, Zetterberg H. Cere-
[14] Huang B, Lucas T, Kueppers C, Dong X, Krause M, brospinal fluid and plasma biomarkers in Alzheimer disease.
Bepperling A, et al. Scalable production in human cells Nat Rev Neurol. 2010;6:131-44.
and biochemical characterization of full-length normal and [24] Landles C, Sathasivam K, Weiss A, Woodman B, Moffitt
mutant huntingtin. PLoS One. 2015;10:1-23. doi:10.1371/ H, Finkbeiner S, et al. Proteolysis of mutant huntingtin pro-
journal.pone.0121055 duces an exon 1 fragment that accumulates as an aggregated
[15] Todd J, Freese B, Lu A, Held D, Morey J, Livingston R, et protein in neuronal nuclei in Huntington disease. J Biol
al. Ultrasensitive flow-based immunoassays using single- Chem. 2010;285:8808-23. doi:10.1074/jbc.M109.075028
molecule counting. Clin Chem. 2007;53:1990-5. doi:10. [25] DiFiglia M, Sapp E, Chase KO, Davies SW, Bates GP,
1373/clinchem.2007.091181 Vonsattel JP, et al. Aggregation of huntingtin in neuronal
[16] Fodale V, Kegulian NC, Verani M, Cariulo C, Azzollini L, intranuclear inclusions and dystrophic neurites in brain.
Petricca L, et al. Polyglutamine- and temperature-dependent Science. 1997;277:1990-3. doi:10.1126/science.277.5334.
conformational rigidity in mutant huntingtin revealed by 1990
immunoassays and circular dichroism spectroscopy. PLoS [26] Vonsattel JPG, Keller C, Cortes Ramirez EP. Hunting-
One. 2014;9:e112262. doi:10.1371/journal.pone.0112262 ton’s disease – neuropathology. H Handb Clin Neu-
[17] Lee JW, Devanarayan V, Barrett YC, Weiner R, Allinson J, rol. 2011;100:83-100. doi: 10.1016/B978-0-444-52014-2.
Fountain S, et al. Fit-for-purpose method development and 00004-5
validation for successful biomarker measurement. Pharm [27] Hühmer AF, Biringer RG, Amato H, Fonteh AN, Harrington
Res. 2006;23:312-28. doi:10.1007/s11095-005-9045-3 MG. Protein Analysis in Human Cerebrospinal Fluid: Phys-
[18] Desilva B, Smith W, Weiner R, Kelley M, Smolec J, Lee B, iological Aspects, Current Progress and Future Challenges.
et al. Recommendations for the bioanalytical method vali- Dis Markers. 2006;22:3-26. doi:10.1155/2006/158797
dation of ligand-binding assays to support pharmacokinetic [28] Schutzer SE, Liu T, Natelson BH, Angel TE, Schepmoes
assessments of macromolecules. Pharm Res. 2003;20:1885- AA, Purvine SO, et al. Establishing the proteome of nor-
900. doi:10.1023/B:PHAM.0000003390.51761.3d mal human cerebrospinal fluid. PLoS One. 2010;5:e10980.
[19] Lee JW. Method validation and application of protein doi:10.1371/journal.pone.0010980
biomarkers: Basic similarities and differences from biother-
apeutics. Bioanalysis. 2009;1:1461-74. doi:10.4155/bio.09.
130