Pa allelism in Neu odegene a i e Bioma ke Tes s: Hidden
E o s and he Risk o Misconduc
Axel Pe zold∗and Joachim Pum†and Da id P. C abb‡
No embe 19, 2025
Abs ac
Bioma ke s a e c i ical ools in he diagnosis and
moni o ing o neu odegene a i e diseases. Reliable
quan i ica ion depends on assay alidi y, especially
he demons a ion o pa allelism be ween dilu ed
biological samples and he assay’s s anda d cu e.
Inadequa e pa allelism can lead o biased concen-
a ion es ima es, jeopa dizing bo h clinical and e-
sea ch applica ions. He e we sys ema ically e iew
he e idence o analy ical pa allelism in body luid
(se um, plasma, ce eb ospinal luid) bioma ke as-
says o neu odegene a ion and e alua e he ex en ,
ep oducibili y, and epo ing quali y o pa ial pa -
allelism.
This sys ema ic e iew was egis e ed on PROS-
PERO (CRD42024568766) and conduc ed in ac-
co dance wi h PRISMA guidelines. We included
s udies published be ween Decembe 2010 o July
2024 wi hou language es ic ions. Eligible s ud-
ies included o iginal esea ch assessing bioma ke
concen a ions in body luids wi h da a sui able
o e alua ing se ial dilu ion and s anda d cu e
pa allelism. The da a ex ac ion o in e oga ing
pa allelism included dilu ion s eps, measu ed con-
cen a ions, and sample ypes. Fo each s udy we
gene a ed pa allelism plo s in a uni o m and com-
pa able way. These g aphs we e used o come o
a balanced decision on whe he pa allelism o pa -
ial pa allelism we e p esen . The isk o bias was
assessed based on sample p epa a ion, bu e con-
sis ency, and me hodological anspa ency.
O 44 eligible s udies, 19 p o ided su icien da a
o gene a ing 49 pa ial pa allelism plo s. O hese
plo s, only 7 (14%) demons a ed clea pa ial pa -
allelism. Pa ial pa allelism was ypically achie ed
o e a na ow dilu ion ange o abou h ee dou-
bling s eps. Mos assays de ia ed om pa allelism,
isking o e - o unde es ima ion o bioma ke le -
els i de e mined a di e en dilu ion s eps. A high
∗Uni e si y College London, Queen Squa e Ins i u e o
Neu ology, London, WC1N 3BG, Uni ed Kingdom.
Email: a.p[email p o ec ed]
†LABanaly ics GmbH, An on-B uckne -Weg 22, Jena,
Ge many
‡Ci y S . Geo ge’s, Uni e si y o London, No hamp on
Squa e, London, EC1V 0HB, Uni ed Kingdom
isk o bias was iden i ied in 9 s udies using spiked
o a i icial samples, inconsis en dilu ion bu e s,
o incomple e epo ing. Se e al s udies assessed
sample- o-sample pa allelism a he han sample-
o-s anda d, con a y o guidelines by egula o y
au ho i ies.
In conclusion, pa ial pa allelism was in e-
quen ly obse ed and inconsis en ly epo ed in
mos bioma ke assays o neu odegene a ion. Na -
ow dilu ion anges and a iable me hodologies
limi gene alizabili y. T anspa en epo ing o di-
lu ion p o ocols and adhe ence o es ablished an-
aly ical alida ion guidelines is needed. This sys-
ema ic e iew has p ac ical implica ions o clini-
cal ial design, egula o y app o al p ocesses, and
he eliabili y o bioma ke -based diagnos ics.
Keywo ds: Assay alida ion; pa allelism; ma ix
e ec s; hook e ec ; p o ein agg ega ion.
1
Pa allelism Page 2
Con en s
1 In oduc ion 2
2 Te minology and concep 4
2.1 Pa allelism and s anda d cu es . . . 4
2.2 De e mina ion o pa allelism . . . . . 5
3 Clinical con ex 6
4 Me hods 7
4.1 Eligibili y C i e ia . . . . . . . . . . 7
4.2 In o ma ion Sou ces . . . . . . . . . 7
4.3 Sea ch S a egy . . . . . . . . . . . . 7
4.4 S udy Selec ion . . . . . . . . . . . . 8
4.5 Da a Collec ion P ocess and Da a
i ems.................. 8
4.6 Risk o Bias in Indi idual S udies . . 8
4.7 Summa y Measu es . . . . . . . . . . 8
4.8 Syn hesis o Resul s . . . . . . . . . 8
4.9 Risk o Bias Ac oss S udies . . . . . 8
5 Resul s 9
5.1 S udy Selec ion . . . . . . . . . . . . 9
5.2 S udy Cha ac e is ics . . . . . . . . 9
5.3 Risk o Bias Wi hin S udies . . . . . 9
5.4 Resul s o Indi idual S udies . . . . 9
5.4.1 Amyloid β.......... 9
5.4.2 α-Synuclein . . . . . . . . . . 11
5.4.3 DJ-1.............. 12
5.4.4 Tau P o ein . . . . . . . . . . 12
5.4.5 Apolipop o ein E . . . . . . . 14
5.4.6 Neu o ilamen p o eins . . . . 14
5.4.7 Glial ib illa y acidic p o ein 14
5.4.8 Dipep ide epea p o eins . . 14
5.5 Syn hesis o Resul s . . . . . . . . . 15
5.6 Risk o Bias Ac oss S udies . . . . . 15
6 Discussion 17
7 Conclusion 23
7.1 PRISMA 2020 Checklis . . . . . . . 31
1 In oduc ion
Bioma ke assays ha e become cen al o ad anc-
ing esea ch and clinical applica ions in neu ode-
gene a ion. Fo ins ance, he as ocy ic bioma ke
glial ib illa y acidic p o ein (GFAP) has been ap-
p o ed by he U.S. Food and D ug Adminis a ion
(FDA) as pa o a panel o guide decisions on
b ain imaging a e auma [1]. Simila ly, he FDA
g an ed apid app o al o no el ea men s in mul-
iple scle osis (MS) [2] and amyo ophic la e al scle-
osis (ALS) [3] on he basis o neu o ilamen (N )
bioma ke s [4]. The accu a e quan i ica ion o such
bioma ke s elies on obus assay pe o mance. To
ensu e his, he FDA and o he egula o y au ho i-
ies ha e es ablished comp ehensi e amewo ks o
bioma ke alida ion, co e ing all s ages om sam-
ple collec ion and p ocessing o analy ical alida-
ion, clinical applica ion, quali y con ol, and ac-
c edi a ion. Howe e , he b ead h o his ame-
wo k can also c ea e ambigui y. Fo example, he
e ms “sensi i i y” and “speci ici y” a e used di e -
en ly in analy ical e sus clinical con ex s (Table 1).
This dis inc ion is illus a ed by neu o ilamen ligh
chain (N L), which has a epo ed analy ical speci-
ici y o 99.3% [5] due o minimal c oss- eac i i y
wi h o he N iso o ms, ye demons a es low clin-
ical speci ici y because blood N L le els ise ac oss
a wide ange o neu ological diseases [2, 3] and in
many condi ions ha comp omise he in eg i y and
unc ion o neu ons and hei connec ions [4,6].
The amewo k o labo a o y es e alua ion
p esen ed in Table 1 equi es addi ional cla i ica-
ion because ce ain e ms a e equen ly misin-
e p e ed in he li e a u e [8]. A no able exam-
ple conce ns he dis inc ion be ween “pa allelism”
and “dilu ion linea i y” Guidance documen s such
as ISO/IEC 17025 and EURACHEM e e o pa al-
lelism when illus a ing me hod alida ion, bu he
concep la gely o igina es om applied labo a o y
p ac ice. In p ac ical use, pa allelism and dilu ion
linea i y a e some imes con la ed because bo h in-
ol e se ial dilu ions. The key dis inc ion lies in
he sample ype: dilu ion linea i y expe imen s use
samples spiked wi h analy e a concen a ions de-
signed o minimize ma ix e ec s, whe eas spiked
samples a e no pe missible o assessing pa al-
lelism [8, 9]. Al hough his di e ence may appea
sub le, as will be a gued in he p esen c i ical e-
iew, i has impo an implica ions o assay ali-
da ion and in e p e a ion.
Beyond hese e minological dis inc ions, co e
analy ical pa ame e s such as p ecision, accu acy,
linea i y, and he e alua ion o ma ix e ec s e-
main cen al o assay alida ion [7]. P ecision and
accu acy ensu e ha esul s a e bo h ep oducible
and co ec , while linea i y con i ms ha mea-
su ed alues emain p opo ional o analy e con-
cen a ions ac oss he assay’s dynamic ange [7].
These pa ame e s can be comp omised by ma ix
e ec s, which in oduce sys ema ic bias [8,10]. This
challenge is pa icula ly p onounced when analyz-
ing complex biological samples such as se um and
plasma [7, 8, 10], especially in mass spec ome y-
based assays [11–13]. To mi iga e such issues, ha -
moniza ion ac oss labo a o ies and analy ical pla -
o ms has become a majo ocus in bioma ke al-
ida ion s udies [14–17]. The goal is o ensu e
compa abili y o esul s and eliabili y o clinical
decision-making [8]. As emphasized ea lie , one
o he hidden sou ces o e o is he lack o pa -
allelism [18, 19]. Fo his eason, egula o y au-
ho i ies equi e demons a ion o pa allelism be-
ween se ially dilu ed samples and he assay cali-
C i ical Re iews in Clinical Labo a o y Sciences
Pa allelism Page 3
Table 1: F amewo k o labo a o y es alida ion as p o ided by egula o y au ho i ies.
Abb e ia ions: U.S. Food and D ug Adminis a ion = FDA, Con o mi é Eu opéenne = CE, College
o Ame ican Pa hologis s = CAP, Clinical and Labo a o y S anda ds Ins i u e = CLSI, In e na ional
O ganiza ion o S anda diza ion = ISO, analy ical measu emen ange = AMR, limi o he blank =
LoB, limi o de ec ion = LoD, limi o quan i a ion = LoQ. Table adap ed om e e ence [7] wi h a
ocus on quan i a i e labo a o y es s.
Labo a o y es ISO 15189 / 17025 CAP
FDA app o ed / CE
ce i ied
In-house & modi ied
FDA app o ed / CE
ce i ied
FDA app o ed / CE
ce i ied
In-house & modi ied
FDA app o ed / CE
ce i ied
P ecision & Bias Ve i y Es ablish Ve i y Es ablish
CLSI EP15-A3 CLSI EP15-A3 CLSI EP15-A3 CLSI EP15-A3
Me hod Compa ison Expe i-
men
Compa e wi h cu -
en me hod
Compa e wi h e e -
ence me hod
Compa e wi h cu -
en me hod
Compa e wi h e e -
ence me hod
CLSI EP09-A3 Re-
g ession Analysis,
Di e ence Plo
CLSI EP09-A3 Re-
g ession Analysis,
Di e ence Plo
CLSI EP09-A3 Re-
g ession Analysis,
Di e ence Plo
CLSI EP09-A3 Re-
g ession Analysis,
Di e ence Plo
Analy ical Sensi i i y — Es ablish Ve i y Es ablish
— CLSI EP17-A2 LoB,
LoD, LoQ, P ecision
P o ile App oach,
P obi Analysis
Documen a ion
om manu ac u e
o li e a u e, CLSI
EP17-A2 Ve i ica-
ion o LoB, LoD,
LoQ
CLSI EP17-A2 LoB,
LoD, LoQ, P ecision
P o ile App oach,
P obi Analysis
Analy ical Speci ici y — Es ablish Ve i y Es ablish
— CLSI EP07-A Tes
o hemolysis,
ic e us and lipemia,
po en ial c oss-
eac i i ies
Documen a ion
om manu ac u e
o li e a u e
CLSI EP07-A Tes
o hemolysis,
ic e us and lipemia,
po en ial c oss-
eac i i ies
Diagnos ic Sensi i i y — — — —
— — — —
Diagnos ic Speci ici y — — — —
— — —
Linea i y Ve i y Es ablish Ve i y Es ablish
E alua ion o lin-
ea i y, Calib a ion
Ve i ica ion, Ve i i-
ca ion o AMR
E alua ion o linea -
i y, Calib a ion Ve -
i ica ion, Es ablish
AMR
E alua ion o lin-
ea i y, Calib a ion
Ve i ica ion, Ve i i-
ca ion o AMR
E alua ion o linea -
i y, Calib a ion Ve -
i ica ion, Es ablish
AMR
Ca yo e Ve i y Es ablish Ve i y Es ablish
S anda d p o ocol
o Sho p o ocol
S anda d p o ocol S anda d p o ocol
o Sho p o ocol
S anda d p o ocol
Measu emen Unce ain y Es ablish Es ablish Es ablish Es ablish
ISO 11352 /
NORDTEST
Me hod
ISO 11352 /
NORDTEST
Me hod
ISO 11352 /
NORDTEST
Me hod
ISO 11352 /
NORDTEST
Me hod
Re e ence Range / Cu -o Ve i y Es ablish Ve i y Es ablish
Documen a ion
om manu ac u e
o li e a u e
CLSI EP28-A3c Use
di ec o indi ec
me hod o da a
collec ion
CLSI EP28-A3c
(Use di ec o in-
di ec me hod o
da a collec ion)
OR Documen a ion
om manu ac u e
o li e a u e
CLSI EP28-A3c
(Use di ec o in-
di ec me hod o
da a collec ion)
C i ical Re iews in Clinical Labo a o y Sciences
Pa allelism Page 4
b a ion cu e [20, 21]. Demons a ing pa allelism
ensu es ha measu ed concen a ions emain con-
sis en and accu a e ac oss he ele an dilu ion
ange. Ye , he ex en o which guidelines [20,21]
o es ing pa allelism a e sys ema ically applied in
bioma ke alida ion s udies emains unknown.
The e o e, we sys ema ically e iewed he
bioma ke li e a u e on neu odegene a ion o e -
idence o pa allelism es ing, beginning wi h he
i s epo o i s absence in he neu o ilamen hea y
chain (N H) ELISA, which was a ibu ed o p o-
ein agg ega e o ma ion [18]. P o ein agg ega-
ion is a hallma k o many neu odegene a i e dis-
eases [22], meaning ha he lack o pa allelism
obse ed in N H assays has implica ions a be-
yond a single bioma ke [18]. Building on his
obse a ion [18], we examined subsequen s udies
o de e mine whe he o he bioma ke s a e sim-
ila ly a ec ed. De ia ions om pa allelism [23]
can lead o sys ema ic o e - o unde es ima ion o
bioma ke concen a ions, c ea ing isks o mis-
in e p e a ion ha a e pa icula ly consequen ial
in he con ex o clinical ials [2, 3] and egula-
o y submissions [1,20,21]. Finally, ou c i ical e-
iew unde sco es he limi a ions o cu en epo -
ing p ac ices in bioma ke alida ion s udies, which
o en include inconsis en dilu ion p o ocols, in-
comple e epo ing o dilu ion anges, and eliance
on non- ep esen a i e spiked o a i icial samples.
Wi h ega d o e minology, we begin by cla -
i ying he concep o pa allelism, a e m o igi-
nally oo ed in geome y. Al hough s a is ical ap-
p oaches o es ing pa allelism a e a ailable, de-
ailed discussion o hese me hods is p o ided in
he supplemen a y ma e ials. In he main ex ,
we ins ead ocus on isual me hods, as hey a e
mo e accessible o eade s wi hou a s ong ma h-
ema ical backg ound. This app oach is suppo ed
by illus a i e examples o ensu e cla i y. We hen
p esen he me hodology o ou sys ema ic e iew,
conduc ed in s ic acco dance wi h he PRISMA
2020 guidelines. The esul ing indings a e in e -
p e ed in he con ex o clinical labo a o y science
and ex ended o add ess b oade me hodological
and egula o y issues. Finally, we highligh p ac i-
cal ecommenda ions o labo a o ies, manu ac u -
e s, and egula o s, wi h he aim o ensu ing appli-
cabili y beyond he esea ch communi y.
2 Te minology and concep
The o mal concep o pa allelism can be aced
o he G eek ma hema ician, geome e , and logi-
cian Euclid (Εὐκλείδης), who, a ound 300BCE,
au ho ed he seminal ea ise Elemen s. In Book
I, De ini ion 23, Euclid de ined pa allel lines as
“s aigh lines which, being in he same plane and
being p oduced inde ini ely in bo h di ec ions, do
no mee one ano he in ei he di ec ion.” How-
e e , i is Euclid’s i h pos ula e, known as he pa -
allel pos ula e, ha has his o ically a ac ed he
mos sc u iny. O e he cen u ies, ma hema icians
sough o p o e he i h pos ula e using Euclid’s
o he axioms. These e o s pe sis ed o o e wo
millennia and p oduced many lawed o incomple e
p oo s, e lec ing a signi ican his o ical misconcep-
ion [24]. I was no un il he 19 h cen u y ha
ma hema icians like Ca l F ied ich Gauss (wo k-
ing on pa allelism 1779–1844), Lobache sky (1829–
30), and Bolyai (1832) demons a ed ha en i ely
consis en non-Euclidean geome ies could be con-
s uc ed by eplacing he i h pos ula e wi h al e -
na i e e sions [24, 25]. In a le e o Bolayi (17-
DEC-1799) Gauss w o e: “I is ue ha I ha e
come upon much which by mos people would be
held o cons i u e a p oo : bu in my eyes i p o es
as good as no hing.”
Simila ly, in he con ex o p esen , c i ical,
sys ema ic e iew, pa allelism, hough seemingly
s aigh o wa d a i s glance, demands ca e ul
sc u iny. Jus as o Euclid’s pa allel pos ula e, an-
aly ical pa allelism in quan i a i e assays mus no
be assumed bu mus be igo ously es ed, subs an-
ia ed, and impo an ly lack o pa allelism mus be
unde s ood.
2.1 Pa allelism and s anda d cu es
The accu a e de e mina ion o analy e concen a-
ions using a s anda d cu e equi es ha he di-
lu ion se ies o he es sample exhibi s pa allelism
wi h he s anda d cu e [26]. This means he wo
cu es mus be simila unc ions, di e ing only by
a scaling ac o along he dose axis (in p esen sys-
ema ic e iew his is always he y-axis), so ha
in e pola ion yields alid and ep oducible esul s.
Wi hou demons a ed pa allelism, calcula ions de-
i ed om a s anda d cu e may lead o inaccu a e
quan i ica ion [26]. In i s mos simple o m in e po-
la ion is linea . Hence he o e i y/es ablish linea -
i y o a labo a o y es in Table 1. Linea in e po-
la ion is a me hod o cu e i ing using linea poly-
nomials o cons uc new da a poin s wi hin he
ange o a disc e e se o known da a poin s [27]. I
is absolu ely c ucial o ecognise ha o bioma ke
concen a ions om samples calcula ions a e only
pe mi ed o da a poin s wi hin he ange o he
poin s o he s anda d cu e. Expanding om lin-
ea in e pola ion, quad a ic, cubic, ou -pa ame e
logis ic (4PL) [28,29], 4PL wi h loga i hmic scaling
o he dose (y-axis) [30], and i e-pa ame e logis ic
(5PL) [29,31–33] s anda d cu es ha e en e ed con-
empo a y labo a o y ou ine. Ex apola ion is no
allowed [7,27,30]. I is manda o y o demons a e
pa allelism o he selec ed ype o a s anda d cu e
and eal-wo ld samples [20].
C i ical Re iews in Clinical Labo a o y Sciences
Pa allelism Page 5
Despi e i s impo ance, he alida ion o pa al-
lelism p esen s a p ac ical challenge o many labo-
a o y wo k lows [7]. The concep , as e iewed he e
and long known in analy ical chemis y, has only
mo e ecen ly gained sus ained a en ion in he con-
ex o egula o y me hod alida ion o bioma ke
assays [10, 21]. Va iabili y in in e nal s anda ds
(ISV) has been iden i ied as a co e ac o con ibu -
ing o measu emen e o , u he unde sco ing he
need o obus alida ion.
Regula o y agencies ha e begun o add ess he is-
sue o pa allelism. The FDA, h ough i s 2022 M10
Bioanaly ical Me hod Valida ion (BMV) guidance,
and he Eu opean Medicines Agency (EMA) bo h
highligh pa allelism as a c i ical alida ion c i e-
ion [20,34]. Acco ding o he EMA, pa allelism is
de ined as ollows: “Pa allelism demons a es ha
he se ially dilu ed incu ed sample esponse cu e
is pa allel o he calib a ion cu e” [20].
Spiked samples in pa allelism The use o
spiked samples in bioassay alida ion is an in e -
es ing s a egy o assess analy ical pe o mance,
pa icula ly in ela ion o ma ix e ec s and as-
say speci ici y [10]. The use o spiked sam-
ples is equen ly employed in he con ex o
mass spec ome y-based me hods, whe e ma ix-
induced a iabili y is a majo conce n o he ac-
cu acy o quan i ica ion [11, 35]. Cu en guide-
lines sugges ha a ia ions o less han 20% a e
gene ally accep able when compa ing spiked sam-
ples wi h calib a o s o quali y con ols [10,36,37].
Howe e , his app oach ypically ocuses on pe o -
mance a ixed concen a ion le els and does no
explici ly e alua e assay beha iou ac oss wide di-
lu ion anges o in he p esence o o he ac o s ha
apply o eal-wo ld samples [37].
The e o e one o he key limi a ions o spiked
samples is he equi emen o high-concen a ion
ma e ial o enable se ial dilu ion and minimize ma-
ix e ec s [37]. Al hough spiking can be in o ma-
i e unde con olled condi ions, i in oduces se -
e al me hodological conce ns. Fi s , he e e ence
ma e ial used o spiking may di e s uc u ally o
unc ionally om he endogenous bioma ke . This
includes he use o unca ed pep ide sequences o
p o eins de i ed om non-human species [38, 39].
Second, he s abili y o spiked ma e ial can be
subop imal, pa icula ly o e long- e m s o age,
whe e p o ein agg ega ion may occu [39]. Agg e-
ga ion, a known sou ce o e o in pa allelism as-
sessmen [18], canno be adequa ely simula ed us-
ing spiked samples.
Mo eo e , spiked samples do no eplica e he
complexi y o physiological ma ices. As no ed in
p io e iews, hei u ili y is limi ed o e alua ing
in e nal s anda d a iabili y in oduced by physio-
logical di e ences be ween con ol and pa ien ma-
ices [37]. E en when he spiked analy e beha es
addi i ely wi h he endogenous bioma ke , his as-
sump ion may no hold unde all condi ions, and
u he conce ns ega ding non-linea in e ac ions
pe sis [40].
Fo hese easons, eal-wo d, pa ien -de i ed
samples emain he gold s anda d o assessing pa -
allelism [8,9]. Thei use be e e lec s he biochem-
ical and biophysical p ope ies ele an o clinical
measu emen [41]. In he con ex o his sys ema ic
e iew, he use o spiked samples is discussed as a
po en ial sou ce o bias.
2.2 De e mina ion o pa allelism
Visual assessmen o e s a complemen a y and o en
mo e in ui i e app oach o e alua ing pa allelism
in bioassays [26]. Unlike o mal s a is ical es ing
(see supplemen a y ma e ials), g aphical me hods
allow analys s o inspec he beha iou o dilu ion
cu es di ec ly, which is pa icula ly aluable in he
p esence o noisy o incomple e da ase s. S a is-
ical expe ise equi ed o p ope ly in e p e hy-
po hesis es ing me hods [42–45] is no commonly
included in he co e aining o labo a o y sci-
en is s, clinicians, o egula o y e iewe s. This
gap may lead o an o e eliance on s a is ical ou -
comes, po en ially o e looking meaning ul de ia-
ions in cu e beha iou ac oss ex ended dilu ion
anges [23]. In eal-wo ld applica ions, comple e
pa allelism is a e, while pa ial pa allelism is o en
obse ed wi hin limi ed dilu ion anges [23]. This
p ac ical obse a ion has led o he ope a ional de i-
ni ion o pa ial pa allelism, acknowledging ha as-
says may beha e accep ably wi hin speci ic, p e-
de ined anges wi hou mee ing idealized c i e ia
ac oss he en i e cu e [46].
The alue o isual me hods o de ec ing lack
o pa allelism has been ecognized in he li e a u e
since he landma k pape by Plikay is [26], and has
been expanded by subsequen s udies [23, 47–49].
Figu e 1 illus a es he mos common pa e ns ob-
se ed.
The in e p e a ion o he pa ial pa allelism plo s
goes back o Euclid’s geome y. The wo lines in
he pa ial pa allelism plo need o be pa allel. To
be mo e p ecise hey need o be pa allel o each
o he , on he y-axis, o a leas pa o he g aph
(Pa e n 3 in Figu e 1). Hence he e m pa ial
pa allelism. A e ical o se is pe mi ed, which
is also en i ely consis en wi h he equi emen s o
s a is ical es ing [43]. In e p e a ion o pa ial pa -
allelism plo s is bes i da a o he s anda d cu es
a e p esen . Fo he pu pose o his sys ema ic e-
iew pa ial pa allelism plo s will be c ea ed o
each s udy included. Based on he g aphical in e -
p e a ion, which is p esen ed o each s udy, a bi-
na y decision will be made: p esence (see Figu e 1
C i ical Re iews in Clinical Labo a o y Sciences
Pa allelism Page 6
Figu e 1: Visual assessmen o pa allelism. The igu e illus a es cha ac e is ic pa e ns used o com-
pa e dilu ion anges (x-axis) o he assay calib a ion s anda d (blue do ed line) wi h heo e ical sample
esponses (black lines). Such isual pa e ns p o ide an accessible means o de e mining whe he pa al-
lelism is main ained o los and o m he basis o ou app oach o e alua ing pa allelism ac oss he assays
included in his c i ical and sys ema ic e iew. Pa e n 1 shows an appa en inc ease in bioma ke con-
cen a ion wi h g ea e sample dilu ion, a phenomenon ypically associa ed wi h he elease o bioma ke
molecules om p o ein agg ega es [18]. Pa e n 2 shows a dec ease in measu ed bioma ke concen a-
ion wi h successi e dilu ion s eps, commonly obse ed when concen a ions app oach he assay’s lowe
limi o de ec ion (LoD; see Table 1). This unde sco es why egula o y au ho i ies equi e alida ion o
analy ical sensi i i y [7]. Pa e n 3 demons a es ha pa allelism may be achie ed wi hin a limi ed
dilu ion ange bu los again a highe dilu ions. This pa e n is e med pa ial pa allelism [23]. As
highligh ed h oughou his e iew, obse ed pa allelism is mos o en pa ial.
pa e n 3) o absence (Figu e 1 pa e ns 1&2, and
o he pa e ns o de ia ion om pa allel alignmen )
o pa ial pa allelism.
3 Clinical con ex
The clinical implica ions o epo ing a i icially el-
e a ed bioma ke concen a ions due o lack o pa -
allelism ( o example Pa e n 1 in Figu e 1) can be
illus a ed using neu o ilamen s. Bo h N L [50,51]
and N H [52,53] ha e been es ablished as aluable
p ognos ic bioma ke s in pa ien s wi h Guillain-
Ba é synd ome (GBS) [54,55].
GBS is cha ac e ized by e ol ing pa alysis, which
in se e e cases can comp omise espi a o y unc-
ion and necessi a e in ensi e ca e uni (ICU) ad-
mission wi h mechanical en ila ion. While mos
pa ien s expe ience ansien pa alysis ollowed by
subs an ial eco e y, a subse de elop long- e m
disabili ies, including he pe manen loss o am-
bula ion. Ea ly iden i ica ion o high- isk pa ien s
would he e o e be o g ea clinical alue, enabling
pe sonalized managemen s a egies anging om
ICU admission decisions o ea men escala ion.
Ta ge ing high-e icacy, bu o en mo e cos ly,
ea men s o pa ien s a g ea es isk o e s he
dual bene i o op imizing clinical ou comes and
imp o ing esou ce alloca ion. Enhanced unc-
ional eco e y in his subg oup would no only im-
p o e pa ien quali y o li e bu also educe down-
s eam heal hca e cos s, including hose associa ed
wi h long- e m disabili y, loss o wo k capaci y, in-
c eased ca e needs, and b oade economic dead-
weigh losses. The impo ance o such s a egies
is e lec ed in Wo ld Heal h O ganiza ion (WHO)
guidance de eloped in collabo a ion wi h he Uni-
e sal Heal h Co e age Pa ne ship (UHC Pa ne -
ship). Thei epo , co e ing 115 coun ies and ep-
esen ing mo e han h ee billion people, explici ly
ecognizes he essen ial ole o clinical labo a o y
se ices in heal hca e deli e y [56]. While he e-
po does no speci ically add ess echnical pa am-
e e s such as pa allelism, his le el o de ail is be-
yond i s scope, i implies ha such issues may be
inco po a ed unde he emi o Na ional Quali y
Con ol Labo a o ies [56].
To illus a e how he lack o demons a ed pa -
allelism can ha e wide- anging consequences, we
highligh one ecen example om he li e a u e
o N L in GBS [57]. In his s udy, N L concen a-
ions we e measu ed using a comme cially a ailable
assay. The manu ac u e ecommends quan i ica-
ion a a dilu ion o 1:4. Howe e , o a subse o
samples, measu emen s we e pe o med a a much
highe dilu ion (1:400) wi hou demons a ing pa -
C i ical Re iews in Clinical Labo a o y Sciences
Pa allelism Page 7
allelism be ween he dilu ed samples and he cali-
b a ion s anda d ac oss his ex ended ange. The
in e ed implica ions a e subs an ial:
•In la ed bioma ke concen a ions: Repo ed
alues anged om app oxima ely 1 pg/mL o
nea ly 10,000 pg/mL, which is se e al o de s
o magni ude highe han expec ed unde ei-
he physiological a ia ion o disease- ela ed
pa hology.
•Po en ial pa ien misclassi ica ion: Ele a ed
N L le els we e associa ed wi h mo e se e e
pheno ypes de ined by elec odiagnos ic c i e-
ia [54, 58]. The da a imply ha a i icially
in la ed concen a ions may ha e been in e -
p e ed as ma ke s o i e e sible axonal degen-
e a ion.
•Bias in p edic i e modeling: P edic i e mod-
els a e in aluable o isk s a i ica ion, bu i
hei s a is ical signi icance a ises om a i i-
cially ele a ed bioma ke concen a ions in a
subg oup o se e ely a ec ed pa ien s, hese
models a e unlikely o eplica e in independen
coho s, clinical ials, o clinical p ac ice.
•Impac on ea men e alua ion: The s udy e-
po ed no ea men e ec on N L le els. One
clinical in e p e a ion migh be o wi hhold
a po en ially e ec i e he apy. Howe e , he
ex eme a iabili y in epo ed concen a ions
likely inc eased noise and obscu ed eal e ec s,
de ia ing om p io ea men ials ha ad-
he ed o alida ed dilu ion anges [2,3].
•Risk o delibe a e misuse: A pa icula ly con-
ce ning possibili y a ises in he con ex o
ea men ials. I placebo samples we e ana-
lyzed a a highe dilu ion ange han ea men
samples unde Pa e n 1 condi ions (Figu e 1),
o a a lowe ange unde Pa e n 2 condi ions,
his would in oduce sys ema ic bias a o ing
a posi i e d ug e ec . Such p ac ices would
cons i u e scien i ic misconduc .
This example unde sco es, on mul iple le els,
he necessi y o ensu ing igo ous and eliable
bioma ke quan i ica ion. Inaccu acies no only
isk pa ien misclassi ica ion bu also unde mine
p edic i e modeling, ea men e alua ion, and ul-
ima ely egula o y con idence. To e alua e how
consis en ly his p inciple has been upheld in he
ield, we now p oceed wi h a sys ema ic e iew o
he li e a u e.
4 Me hods
The p o ocol o his sys ema ic e iew was sub-
mi ed o he PROSPERO egis y and published
on he PROSPERO websi e. The s udy p o o-
col can be sea ched unde he egis a ion numbe
CRD42024568766. The 2020 PRISMA guidelines
o epo ing sys ema ic e iews a e ollowed [59].
The 27-i em PRISMA checklis is uploaded (see
sec ion 7.1).
4.1 Eligibili y C i e ia
Inclusion C i e ia: All s udies in ol ing he anal-
ysis o pa allelism o bioma ke s in body luids
we e conside ed o inclusion. No exclusion c i e-
ia we e applied based on pa icipan demog aph-
ics o speci ic disease condi ions. Body luids in-
cluded ce eb ospinal luid (CSF), u ine, sali a, and
o he ele an bodily luids. Va ious analy ical
echniques we e conside ed, including immunoas-
says, mass spec ome y, ELISA, and o he el-
e an me hods o quan i a i e bioma ke de ec-
ion [11–13,21]. Da a was pe mi ed o be de i ed
om pu ely analy ical s udies, expe imen al s udies
o clinical se ings.
Exclusion C i e ia: S udies no in ol ing
bioma ke sampling om body luids. Resea ch
ocused solely on issue bioma ke s o bioma k-
e s de i ed om non- luid sou ces. A icles lack-
ing de ailed me hodology o insu icien da a on
bioma ke sampling echniques [7]. Case epo s,
e iews, and opinion a icles wi hou o iginal e-
sea ch da a.
4.2 In o ma ion Sou ces
Two da abases we e sea ched, Medline and Google
Schola .
4.3 Sea ch S a egy
A sea ch o he MEDLINE da abase was con-
duc ed co e ing he pe iod a e publica ion o
lack o pa allelism o N H [18]. The da es en-
e ed o he sea ch s a egy we e be ween 9-Dec-
2010 and 12-July-2024. The e we e no language
es ic ions. The En ez P og amming U ili ies
(E-u ili ies), p o ided by he Na ional Cen e o
Bio echnology In o ma ion (NCBI), we e used in a
Py hon sc ip which can be downloaded om he
PROSPERO egis e .
Sea ch S a egy De ails: The sea ch e ms o
he bioma ke s ( i s sea ch e m) we e: “neu o ila-
men ”, “neu o ilamen s”, “ au p o ein”, “T- au”, “P-
au”, “glial ib illa y acidic p o ein”, “amyloid be a”,
“ubiqui in C- e minal hyd olase 1”, “neu og anin”,
and “YKL-40”. The sea ch e ms o he analy i-
cal me hods (second sea ch e m) we e: “me hod”,
“de elopmen ”, “linea i y”, “pa allelism”, and “dou-
bling dilu ion”. The i s sea ch e m and he sec-
ond sea ch e m we e combined indi idually. These
combina ions we e coded in py hon. The comple e
C i ical Re iews in Clinical Labo a o y Sciences
Pa allelism Page 8
Figu e 2: PRISMA low diag am o he li e a u e
sea ch, based on he PRISM empla e code [60].
li e a u e sea ch code can be downloaded om he
s udy p o ocol on he
PROSPERO Regis y unde i em numbe 17. The
li e a u e sea ch was pe o med on 02-SEP-2024.
4.4 S udy Selec ion
Neu odegene a ion is a p e alen ea u e in a ious
human diseases, and bioma ke s play a c ucial ole
in i s indi ec assessmen . The e o e, no speci ic
disease- ela ed es ic ions we e imposed. Limi a-
ions we e ocused on he analy ical de elopmen o
bioma ke es s [7]. S udies we e selec ed based on
he a ailabili y o quan i a i e body luid samples.
Selec ion P ocess: All s udies iden i ied unde -
wen a ull- ex e iew and e iew o supplemen a y
da a whe e a ailable.
Tools Used: A sp eadshee was kep using he
PubMed Iden i ie (PMID) o each s udy. S ud-
ies included and excluded a e e iew we e clea ly
ma ked as such, including he eason o ha deci-
sion.
4.5 Da a Collec ion P ocess and
Da a i ems
Da a Ex ac ion: The da a we e ex ac ed by
hand om he alues p o ided o samples a each
dilu ion s ep in o a sp eadshee . I da a we e only
a ailable in a igu e, he co esponding au ho was
con ac ed by email, con aining he PROSPERO
s udy numbe , asking o sha e he aw da a.
Au ho con ac : Two email eques s o sha -
ing he da a i ems equi ed we e sen o all co -
esponding au ho s. The i s email was sen on
12-SEP-2024. The second email was sen o he
non- esponde s on 12-OCT-2024.
Da a I ems: The h ee speci ic a iables o
which da a we e collec ed included: 1. dilu ion s ep,
2. measu ed bioma ke concen a ion, 3. sample
ype (s anda d, body luid, a i icial/spiked).
4.6 Risk o Bias in Indi idual S ud-
ies
The isk o bias in indi idual s udies was e alu-
a ed by ca e ully e iewing he me hodologies used
in sample selec ion and p epa a ion. Speci ically,
we assessed whe he he samples we e na i e o
spiked wi h p o ein s anda ds and whe he bo h
he samples and s anda ds we e dilu ed using he
same bu e , ensu ing ha i was indeed he co ec
dilu ion bu e [7]. Addi ionally, we eco ded he se-
lec ion o app op ia e disease and con ol samples,
no ed whe he he labo a o y analys was blinded
du ing he expe imen s, and conside ed he ep o-
ducibili y o he expe imen s h ough epea assess-
men s.
4.7 Summa y Measu es
The summa y measu es used in his e iew include
he de ia ion om he line o uni y, which is ep e-
sen ed by he ho izon al line a a y- alue o one in
he pa ial pa allelism plo s [23]. The da a a e ca e-
go ical, indica ing he p esence o absence o pa ial
pa allelism (yes/no). Fo s udies whe e pa allelism
is obse ed, con inuous da a we e p esen ed, speci -
ically he dilu ion ange wi hin which pa ial pa -
allelism was demons a ed. These wo e ec mea-
su es will be analysed o each sample indi idually
and, when meaning ul (n>5), also as a g oup mean.
4.8 Syn hesis o Resul s
The syn hesis o esul s is p ima ily isual, u il-
ising pa ial pa allelism plo s [23] de i ed om
he included s udies. These isual ep esen a ions
a e hen summa ised in able o ma and comple-
men ed by a na a i e syn hesis.
4.9 Risk o Bias Ac oss S udies
The p ima y ocus is on de e mining whe he pa -
ial pa allelism exis s be ween a sample om an
indi idual wi h a speci ic disease and he p o ein
s anda d used in he es . P ac ically his equi es
o p o ide da a o a dilu ion se ies o he sample
in o he assay bu e ( he same dilu ion bu e ha
is used o he s anda d cu e) [7]. The e o e, a
high isk o bias was assigned i no such sample
was a ailable and ins ead, an a i icial sample was
c ea ed by spiking he body luid collec ed wi h he
p o ein. Simila ly, i di e en bu e s we e used o
dilu ion be ween he sample and he s anda d, he
bias was also a ed as high. A mode a e isk o bias
C i ical Re iews in Clinical Labo a o y Sciences
Pa allelism Page 9
was assigned when an app op ia e disease sample
was no used, o i he labo a o y analys was no
blinded, o i da a on ep oducibili y we e missing.
In all o he cases, he isk o bias was a ed as low.
5 Resul s
5.1 S udy Selec ion
The low diag am in Figu e 2 ou lines he p ocess
o s udy selec ion, de ailing he numbe o eco ds
iden i ied, included, and excluded [60]. A comp e-
hensi e li e a u e sea ch ini ially yielded 39 a icles
o u he e alua ion [16,35,61–97]. Addi ionally,
a sea ch on Google Schola , using he same e ms,
iden i ied i e mo e e e ences [5, 98–101]. All 44
a icles [5,16,35,61–101], including supplemen a y
analyses whe e a ailable, we e ho oughly e iewed.
The co esponding au ho was con ac ed wice pe
email o sha ing da a i his was necessa y o c e-
a ing pa ial pa allelism plo s. Fo ou a icles,
no p esen ly alid au ho con ac de ails could be
ound [69, 70, 73, 75]. Fo he emaining pape s, a
ca e ul examina ion o all da a a ailable esul ed in
he exclusion o 25 a icles [16,70,73,75,76,78–97]
due o he absence o da a necessa y o gene a -
ing pa ial pa allelism plo s as pe ou p ede ined
p o ocol. Consequen ly, 19 s udies we e selec ed o
u he e alua ion [5,35,61–69,71,72,74,77,98–101].
5.2 S udy Cha ac e is ics
The cha ac e is ics o he included s udies a e sum-
ma ised in Table 2. The majo i y o s udies ocused
on human samples, p ima ily using plasma o CSF,
wi h only a ew in es iga ing se um. Fou s udies
examined b ain issue homogena es, while wo used
oden issue samples. All s udies included con ol
samples, and mos also inco po a ed samples om
a a ie y o disease condi ions, wi h Alzheime ’s
disease (AD) being he mos equen ly s udied.
The bioma ke s analyzed ac oss he s udies in-
cluded he Amyloid β agmen s, Aβ1−40, Aβ1−42,
Aβoligome s, α-synuclein, DJ-1, o al au p o ein
(Tau), phospho au p o eins (pTau181, pTau217,
pTau231), Apolipop o ein E (ApoE) iso o ms E2,
E3, E4, neu o ilamen ligh chain (N L), neu o ila-
men hea y chain (N H), GFAP, and poly(GP).
5.3 Risk o Bias Wi hin S udies
The isk o bias ac oss he included s udies is sum-
ma ised in Table 3. Only one s udy was assessed
as ha ing a low isk o bias [61]. This s udy p o-
ided clea documen a ion on key ac o s, including
blinding and sample handling.
Eigh s udies we e a ed as ha ing a mode a e
isk o bias [5,35,63,64,66,74,98,100], mainly due
o incomple e in o ma ion ega ding blinding p o-
cedu es, pa icula ly in ela ion o he blinding o
he analys .
Ano he , eigh s udies we e ca ego ised as ha -
ing a high isk o bias [62, 65, 67, 68, 71, 72, 77, 99].
This a ing was due ei he o he use o spiked sam-
ples [62,65,67,68,72,77,99], o o incomple e docu-
men a ion on whe he he samples we e spiked and
how hey we e p ocessed [71].
5.4 Resul s o Indi idual S udies
De ailed esul s o each included s udy a e sum-
ma ised in Table 4.
5.4.1 Amyloid β
The amyloid cascade hypo hesis, p oposed by
Ha dy and Higgins [102], has posi ioned amyloid
β(Aβ) as a cen al pa hological d i e in AD,
ollowing p o eoly ic clea age o he amyloid p e-
cu so p o ein (APP). Quan i ica ion o Aβpep-
ides, pa icula ly Aβ1−42 and Aβ1−40, and hei
a ios has since become in eg al o subsequen e i-
sions o diagnos ic c i e ia o AD [103–105]. How-
e e , agg ega ion-p one p ope ies o Aβ[106,107]
complica e immunoassay quan i ica ion, as epi-
ope masking and al e ed con o ma ions impai
an ibody ecogni ion and dis up dilu ional pa al-
lelism [18].
On e iew o indi idual pa ial pa allelism plo s
his eme ges as a consis en and pe sis en analy -
ical p oblem:
•Aβ1−40: None o he assessed immunoassays
achie ed pa ial pa allelism ac oss h ee inde-
penden s udies [65,69,98]. Fo each s udy de-
ia ions om expec ed pa allelism was shown
in he pa ial pa allelism plo s (see Figu es 12,
14, 16).
•Aβ1−42: Likewise, mos s udies ailed o
demons a e pa ial pa allelism [35, 65, 68, 69,
98]. A single s udy epo ed success ul pa al-
lelism [72]; howe e , his s udy was lagged as
high isk o bias (Table 3), as i in ol ed spik-
ing immunodeple ed pooled plasma wi h syn-
he ic Aβ, a me hod ha may no eplica e
he con o ma ional complexi y o endogenous
pep ides (see Figu es 6, 4, 12, 14, 16, 18).
•O he A βspecies: No e idence o pa ial
pa allelism was ound o unca ed Aβpep-
ides [64], agg ega ed o ms [62], o oligome ic
Aβin EDTA plasma samples [67]. Howe e , in
he la e s udy, he use o ci a e o hepa in
as an icoagulan s enabled pa ial pa allelism,
highligh ing he impo ance o p e-analy ical
a iables o u u e immunoassay de elopmen
(see Figu e 21).
C i ical Re iews in Clinical Labo a o y Sciences
Pa allelism Page 16
Figu e 8: Pa ial Pa allelism Plo o se ial dilu ion o Amyloidβpep ides in plasma. This plo illus a es
he pa ial pa allelism o se ial dilu ions o a ious amyloid-βpep ides in plasma. Da a we e analyzed
ac oss comple e dilu ion anges ( i e da a poin s) o compa ison in he pa ial pa allelism plo s. Such
da a we e a ailable o 4 ou o 18 (22%) plo s p esen ed in Figu e 4 o e e ence [64]. The pep ides
included Aβ6−40, Aβ5−40, Aβ1−38, Aβ1−40. The plo s demons a e a lack o pa ial pa allelism ac oss
he es ed dilu ion anges.
C i ical Re iews in Clinical Labo a o y Sciences
Pa allelism Page 17
Figu e 9: Pa ial pa allelism plo o N H quan-
i ica ion ac oss di e en sample ypes. Sample A
ep esen s bu e spiked wi h N H, while Samples
B–D a e de i ed om na i e CSF [66]. The aw
da a we e p o ided by he co esponding au ho .
The plo shows inc easing N H concen a ions wi h
inc easing dilu ion s eps, while he s anda d cu e
demons a es pa ial pa allelism only wi hin he 1:1
o 1:16 dilu ion ange. O e all, he esul s indica e
a lack o pa ial pa allelism ac oss he ull dilu ion
ange es ed.
ion cu es ac oss six labo a o ies employing one o
mo e o se en assays [76]. Pa allelism was epo ed
as mean pe cen ages pe labo a o y, wi h esul s
anging om 74% o 344%, indica ing signi ican
a iabili y and inconsis ency.
Fu he complica ing he assessmen , some s ud-
ies ailed o p o ide key me hodological de ails,
such as sample p epa a ion p o ocols o calib a ion
s anda ds, which may ha e in oduced addi ional
biases o wha was e iewed in Table 3. Addi-
ionally, a ia ions in expe imen al design, such as
he use o non-s anda d dilu ion ma ices o un e -
i ied spiking p ocedu es, may ha e u he educed
he compa abili y and ep oducibili y o pa allelism
es ing. This unde sco es he need o s ic e ad-
he ence o s anda dised guidelines and mo e ans-
pa en epo ing o me hodological de ails o min-
imise bias and imp o e he eliabili y o pa ial pa -
allelism assessmen s. Ou analysis e ealed sub-
s an ial a iabili y in pa allelism assessmen . The
implica ions o hese indings a e explo ed in he
discussion sec ion.
6 Discussion
This sys ema ic e iew p o ides a comp ehensi e
analysis o pa allelism es ing in neu odegene a-
ion bioma ke assays. The p incipal inding is
ha all cu en bioma ke es s exhibi only pa -
ial [23], a he han ull, pa allelism. Pa ial pa -
allelism plo s, which can be eadily gene a ed om
exis ing da a, o e a s aigh o wa d isual ool o
compa ison. Howe e , he ange o pa ial pa al-
Figu e 10: Pa ial Pa allelism Plo o Neu o ila-
men Ligh Chain (N L) quan i ica ion. The plo il-
lus a es he quan i ica ion o N L in se um samples
spiked wi h 500 ng/mL o he pep ide s anda d [5].
The g aph o he le displays he indi idual di-
lu ion cu es, while he g aph o he igh shows
he a e aged da a om 10 spiked se um samples.
The aw da a we e p o ided by he co esponding
au ho . The g aph demons a es ha pa allelism is
demons a ed, on a g oup le el, o a dilu ion ange
o 1:2–1:8 o ecombinan N L in se um using he
assay bu e .
C i ical Re iews in Clinical Labo a o y Sciences
Pa allelism Page 18
Figu e 11: Pa ial pa allelism plo s o se ial dilu-
ion o CSF measu ing pTau217 using PT3xHT43
and PT3xPT82 assays. Da a we e ex ac ed om
Figu e 2A in e e ence [74]. The plo s do no
demons a e pa ial pa allelism; howe e , his may
be in luenced by da a noise. A e aged da a om
addi ional samples es ed wi hin he dilu ion ange
o 1:8 o 1:64 could p o ide u he cla i y.
Figu e 12: Pa ial Pa allelism plo o Aβ1−42 and
Aβ1−40 quan i ied om plasma samples spiked wi h
he espec i e Amyloidβpep ides. The da a we e
aken om supplemen a y Table 1 in e e ence [98].
The plo indica es a lack o pa ial pa allelism
ac oss he dilu ion ange (na i e sample o 1:4)
es ed.
lelism is ypically na ow, spanning app oxima ely
h ee doubling dilu ion s eps. This inding has
c i ical implica ions o in e p e ing s udies ha
dilu e samples beyond his ange, as such p ac-
ices can lead o inaccu acies. Bioma ke concen-
a ions may be o e es ima ed when pa ial pa al-
lelism plo s de ia e upwa ds (e.g., Figu e 12) o un-
de es ima ed wi h downwa d de ia ions (e.g., Fig-
u e 15). The clinical con ex o such inaccu acies
has been discussed o one example whe e neu o-
ilamen s we e quan i ied a a dilu ion o 1:400 in-
s ead o 1:4. As seen in he p esen e iew, o he
bioma ke s used in he con ex o neu odegene a-
ion a e a ec ed as well. No ably howe e , nea
pe ec pa ial pa allelism [7] can some imes be
achie ed on a g oup le el (e.g., Figu e 10).
The dis inc ion be ween g oup-le el and
indi idual-le el pa ial pa allelism (see Figu e 10)
highligh s wo key poin s. Fi s , bioma ke assays
showing g oup-le el pa ial pa allelism may be
sui able o clinical ials, especially as egula o y
agencies like he FDA and EMA inc easingly
accep bioma ke -based endpoin s o apid d ug
app o als in neu odegene a ion. Second, he ab-
sence o pa allelism in indi idual samples wa an s
u he in es iga ion. Tes s measu ing p o eins
in ol ed in agg ega ion- ela ed pa hologies may
ha bo hidden biases, which could in luence esul s
and in e p e a ions [138]. The e ec o age will be
one o he mos ob ious demog aphic ac o s o be
in es iga ed u he [21,139]. Thi d, pa ial ag ee-
men be ween di e en assays o quan i ica ion o
N L implies p esence o pa ial pa allelism be ween
he wo es s [140]. Howe e , he same me hod
compa ison s udy also showed o e es ima ion o
N L le els by one assay compa ed o ano he a
highe concen a ions [140], indica ing a lack o
pa allelism ou side a na ow dilu ion ange. This
is en i ely consis en wi h he clinical example,
N L in GBS, discussed ea lie .
A 20% ailu e a e in pa ial pa allelism a he in-
di idual sample le el indica es ha one in i e sam-
ples may no yield analy ically compa able esul s.
This deg ee o non-pa allelism, i a ibu able o
biological phenomena such as p o ein agg ega ion
inhe en o disease pa hology [18,141], could in o-
duce signi ican bias in o quan i a i e in e p e a-
ions. I is he e o e essen ial ha de ia ions om
pa allelism a e no dismissed as echnical a e ac s
bu a e ins ead igo ously in e oga ed. While ma-
ix e ec s a e equen ly ci ed, hey do no accoun
o all scena ios and a e p edominan ly a conce n
in mass spec ome y-based pla o ms [11–13]. Ad-
di ional causes include endogenous p o ein–p o ein
in e ac ions, biochemical dissimila i y be ween na-
i e analy es and ecombinan s anda ds, analy e
ins abili y, subop imal assay sensi i i y, and sup-
aphysiological bioma ke concen a ions exceeding
C i ical Re iews in Clinical Labo a o y Sciences
Pa allelism Page 19
Figu e 13: Pa ial Pa allelism Plo o Aβoligome quan i ica ion. This igu e p esen s he pa ial
pa allelism plo o he quan i ica ion o Aβin CSF and PBS [62]. Expe imen s we e conduc ed wi h
inc easing i e a ions: Sample A had ze o i e a ions, Sample B had one i e a ion, p og essing o Sample
E wi h ou i e a ions. The aw da a we e kindly sha ed by he co esponding au ho . The plo s
demons a e a lack o pa ial pa allelism ac oss he dilu ion ange es ed. The ull dilu ion ange is
displayed in he plo s on he le . The zoomed-in plo s on he igh highligh ha Aβoligome s a e
o e es ima ed a dilu ions beyond 1:10.
C i ical Re iews in Clinical Labo a o y Sciences
Pa allelism Page 20
Figu e 14: Pa ial Pa allelism Plo o amyloid
β1−42 and amyloid β1−40 quan i ied om a i i-
cial samples o ecombinan human se um albumin
which was spiked wi h ull leng h ecombinan amy-
loid β1−42 and amyloid β1−40 p o eins a concen a-
ions abou h ee imes abo e he highes s anda d.
The da a we e aken om Table 6 in e e ence [65].
The plo s do no demons a e pa ial pa allelism.
Figu e 15: Pa ial Pa allelism plo o o al Tau
spiked in o PBS. The da a we e aken om able
5 in e e ence [77]. The e is lack o pa ial pa al-
lelism.
Figu e 16: Pa ial Pa allelism plo o spiked CSF
Aβ1−40 (78,300 ng/L) dilu ed in an assay bu e .
The da a we e aken om he op le g aph in
Figu e 2 in e e ence [69]. The plo s do no demon-
s a e pa ial pa allelism.
Figu e 17: Pa ial pa allelism plo s o o al Tau
and pTau (181P) assays. Da a we e de i ed om
“ ep esen a i e esul s om one sample” in Supple-
men a y Figu e 1 o e e ence [71]. The CSF sample
dilu ion se ies in bu e was es ima ed om he x-
axis (1, 0.95, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2,
0.15, 0.10), and measu ed concen a ions we e ap-
p oxima ed om he y-axis. Pa ial pa allelism was
achie ed wi hin a dilu ion ange o 1:1.43 o 1:2.5
o he pTau181 assay bu no o he o al Tau as-
say.
Figu e 18: Pa ial Pa allelism plo o Aβ1−42 om
plasma dilu ed in o a sample bu e . Da a we e
aken om he log-scaled (0–100) y-axis om Fig-
u e 3A in e e ence [72]. The plo shows pa ial
pa allelism wi hin a dilu ion ange o 1:4–1:16.
C i ical Re iews in Clinical Labo a o y Sciences
Pa allelism Page 21
Figu e 19: Pa ial Pa allelism plo s o quan i i-
ca ion o poly-glu amin expansions om na i e
plasma samples. The da a we e aken om Figu e
3H in e e ence [100]. The plo o he le shows he
indi idual dilu ion cu es. The g aph o he igh
shows he a e age o 6 plasma samples. On a g oup
le el, he plo shows pa ial pa allelism wi hin a di-
lu ion ange o 1:4–1:16.
Figu e 20: Pa ial Pa allelism plo o quan i i-
ca ion o phospho yla ed (pTau181, pTau231) au
p o ein om samples spiked wi h he pep ide s an-
da d. The da a we e aken om Supplemen a y
Table 3 in e e ence [99]. Pa ial pa allelism is no
achie ed in his plo .
he assay’s dynamic ange [11–13,18,141,142].
The e o e a c i ical inding in his sys ema ic e-
iew is he po en ial publica ion bias, as s udy ab-
s ac s o en o e s a e he p esence o pa allelism.
Only 14% o he included s udies demons a ed
clea e idence o pa ial pa allelism [5, 61, 71, 72,
100,101]. In e es ingly, se e al s udies epo ed on
pa allelism in be ween di e en samples ins ead o
in be ween samples and he s anda d. This poin
could be emphasized s onge in u u e guidelines
on ha moniza ion o assay alida ion and imple-
men a ion o quali y con ol.
The e iew encoun e ed se e al limi a ions s em-
ming om inconsis encies in he epo ing o
me hodological de ails. Dilu ion anges a ied
widely ac oss s udies, wi h a bi a y s eps (e.g.,
1:4, 1:5, 1:6 e sus 1:1,250, 1:2,500, 1:5,000) [69].
While some dilu ion anges could be in e ed e -
ospec i ely [71], his was o en based on a sin-
gle ep esen a i e sample, p e en ing he gene a-
ion o pa allelism plo s o o he da a. Addi ion-
ally, some epo ed alues exceeded he assay mea-
su emen ange [71], indica ing inapp op ia e ex-
apola ion beyond he s anda d cu e. Fo he
s anda d cu e he dilu ion da a was equen ly
no gi en. Ins ead, pa allelism in be ween sam-
ples, a he han be ween samples and he s anda d
cu e was shown. Table 3 u he highligh s ha
no all samples we e dilu ed in he assay bu e , a
c i ical limi a ion when es ing pa allelism wi h a
s anda d cu e [7].
O e all he numbe o samples and s anda d
cu es was oo small o meaning ul s a is ical mod-
elling [42, 43]. The e is need o la ge num-
be s [143] o samples, s anda ds and da apoin s
(o he dilu ion anges) o obus s a is ical e i-
dence [143,144]. Fo example, one s udy ha was
able o demons a e pa ial pa allelism, did so o
a e y na ow dilu ion ange o 1:1.43–1:2.5 based
on one single sample ha was conside ed o be
ep esen a i e [71]. The only s udy wi h a e y
la ge ange o pa ial pa allelism 1:100–1:10,000 e-
po s his close o he de ec ion limi (16 M) o he
es [67].
Mo eo e , he e was conce ning use o spiked o
a i icial samples in many s udies [5,35,61,62, 65–
69, 72, 74, 77, 99, 100], ins ead o he ecommended
use o na i e samples [8, 9]. Fo one s udy i e-
mained unclea i samples we e spiked o no [71].
Taken oge he he use o spiked samples educes
he gene alisabili y o he indings o c ea ing ep-
esen a i e pa allelism plo s. Simila issues wi h
ha e been epo ed o o he g oups o bioma k-
e s [145]. S onge adhe ence o es ablished guide-
lines is ecommended [8].
Likewise, gene alisabili y is hampe ed by he lack
o sha ing da a on he s anda d cu e o he dilu-
ion s eps p esen ed [5, 35, 61–65, 67–69, 71, 72, 74,
C i ical Re iews in Clinical Labo a o y Sciences
Pa allelism Page 22
Figu e 21: Pa ial Pa allelism Plo s o Aβoligome quan i ica ion in di e en sampling bu e s. The
igu e shows pa ial pa allelism plo s o Aβoligome s in EDTA, Ci a e, and Hepa in bu e s. Samples
A–E a e spiked plasma, and Sample F is spiked PBS [67]. The aw da a we e kindly sha ed by he
co esponding au ho . The plo s e eal a lack o pa allelism o EDTA, while pa ial pa allelism is
achie ed in Ci a e and Hepa in bu e s be ween dilu ions o 1:100 and 1:10,000 on a g oup le el.
C i ical Re iews in Clinical Labo a o y Sciences
Pa allelism Page 23
77, 98–100]. On e ision o he Figu es in p esen
sys ema ic e iew i is no possible o s a e wi h ab-
solu e ce ain y i he s anda d cu e o e lays he
(blue) line o uni y. I would be desi able o ha e,
in u u e s udies, da a om a ep esen a i e num-
be o s and cu es a ailable. These da a poin s do
no need o be exac ly o he same dilu ion s eps as
o samples i he isual ep esen a ion wi h pa ial
pa allelism plo s is chosen. This is ano he ad an-
age o he isual app oach compa ed o s a is ical
echniques.
On c i ical e iew o he se up o sample dilu-
ion i should be men ioned ha o example one
o he inally excluded s udies [16] dilu ed blood
samples om pa ien s wi h mul iple scle osis in o
blood samples o heal h con ols. Fo assessmen
o pa allelism be ween pa ien samples and he as-
say’s s anda d cu e i is manda o y o use he he
same bu e solu ion [7,20,21].
Con lic ing pa allelism esul s we e ound o he
pTau181 assay. One s udy did show pa ial pa -
allelism [71] (Figu e 17B) and one did no [101]
(Figu es 5A-F). Clea ly, on di ec isual compa -
ison o hese pa ial pa allelism plo s i is e iden
ha di e en dilu ion anges we e used. The sho
s e ch o pa ial pa allelism (1:1.43–1:1.25) om
one s udy [71] is o a dilu ion ange lowe han
he one shown o he o he (1:4–1:32 and a:5–
1:40) [101]. Because a hi d s udy did also no
demons a e pa ial pa allelism o pTau181 [99],
a balanced in e p e a ion would sugges absence o
pa allelism o his bioma ke .
Ano he limi a ion o his sys ema ic e iew
is he es ic ion o s udies published be ween
2010 and 2024, a subse o he 32-yea s since
he o mula ion o he amyloid cascade hypo he-
sis (1992–2024) [102]. This ime ame was cho-
sen in en ionally, as conce ns ega ding he e ec
o p o ein agg ega ion on dilu ional pa allelism in
immunoassays we e i s aised in 2010 [18], and
only subsequen ly acknowledged in in luen ial whi e
pape s [12, 21, 146] and egula o y guidance docu-
men s [20,34,147].
Taken oge he , his c i ical and sys ema ic e-
iew emphasises he impo ance o adhe ing igo -
ously o guidelines o pa allelism es ing [12,20,21,
34, 146]. These guidelines a e embedded in a well
de ined amewo k o labo a o y es alida ion
ha is endo sed by egula o y au ho i ies (Table 1).
I obse ed, he eason o lack o pa ial pa allelism
needs o be explained as i has been done in he con-
ex o p o ein agg ega ion [18]. Au ho s should ex-
plici ly epo he dilu ion ange o e which pa ial
pa allelism is achie ed and highligh po en ial bi-
ases, such as o e - o unde es ima ion o bioma ke
concen a ions, when dilu ions exceed his ange.
Fo clinical ials, i is c ucial o clea ly doc-
umen he ange o pa ial pa allelism, he dilu-
ions pe o med, and s a egies o minimise biases.
Regula o y agencies migh conside manda ing he
inclusion o da a om andom indi idual samples
o demons a e pa allelism a bo h indi idual and
g oup le els. Wi hou clea e idence o pa ial
pa allelism, i would be inad isable o quan i y
bioma ke s o neu odegene a ion a a ying dilu-
ion s eps o ac oss di e en ime poin s in clinical
ials. In summa y, ou c i ical and sys ema ic e-
iew highligh s bo h p og ess and pe sis en chal-
lenges in pa allelism es ing. These insigh s in o m
ecommenda ions o u u e assay alida ion and
inclusion in o he es alida ion amewo k o eg-
ula o y au ho i ies (e.g. unde p ecision and bias o
unde linea i y in Table 1).
7 Conclusion
This sys ema ic e iew iden i ied pa ial pa allelism
in only a small p opo ion o bioma ke es s o
neu odegene a ion. A likely biological eason is he
p esence o p o ein agg ega es, a key pa hological
ea u e in many neu odegene a i e diseases. Whe e
pa ial pa allelism is absen , he da a sugges spe-
ci ic dilu ion anges whe e i could po en ially be
achie ed. To ad ance he ield, u u e esea ch
mus align wi h es ablished guidelines [20,147], en-
su ing anspa en epo ing ha allows indepen-
den esea che s and egula o y bodies o e alua e
pa ial pa allelism h ough s anda dised plo s.
Re e ences
[1] Abdelhak A, Foschi M, Abu-Rumeileh S,
e al. Blood GFAP as an eme ging bioma ke
in b ain and spinal co d diso de s. Na u e Re-
iews Neu ology. 2022 Feb;18(3):158–172.
[2] Hause SL, Ba -O A, Cohen JA, e al. O a u-
mumab e sus e i lunomide in mul iple scle-
osis. New England Jou nal o Medicine. 2020
Aug;383(6):546–557.
[3] Mille TM, Cudkowicz ME, Genge A, e al.
T ial o an isense oligonucleo ide o e sen
o SOD1 ALS. New England Jou nal o
Medicine. 2022 Sep;387(12):1099–1110.
[4] Khalil M, Teunissen CE, Lehmann S, e al.
Neu o ilamen s as bioma ke s in neu ological
diso de s — owa ds clinical applica ion. Na-
u e Re iews Neu ology. 2024 Ap ;20(5):269–
287.
[5] Lee S, Pla ina T, Singh CM, e al. De elop-
men o a highly sensi i e neu o ilamen ligh
chain assay on an au oma ed immunoassay
pla o m. F on ie s in Neu ology. 2022 Jul;13.
C i ical Re iews in Clinical Labo a o y Sciences
Pa allelism Page 24
[6] Pe zold A. The 2022 Lady Es elle Wol son
Lec u eship on Neu o ilamen s. Jou nal o
Neu ochemis y. 2022 Aug;163(3):179–219.
A ailable om: h ps://disco e y.
ucl.ac.uk/id/ep in /10146661/1/
LadyEs elleWol sonLec u e.mp4.
[7] Pum J. A p ac ical guide o alida ion and
e i ica ion o analy ical me hods in he clin-
ical labo a o y. Ad ances in Clinical Chem-
is y, Vol 38. 2019;:215–281.
[8] And easson U, Pe e -Liaude A, an Waal-
wijk an Doo n LJC, e al. A p ac ical guide
o immunoassay me hod alida ion. F on ie s
in Neu ology. 2015 Aug;6.
[9] Bu is C, Ashwood E, edi o s. Tie z Tex -
book o Clinical Chemis y. 2nd ed. WB
Saunde s Company; 1994.
[10] Zhang J, Dasgup a A, Ayala R, e al. In e -
nal s anda d a iabili y: oo cause in es i-
ga ion, pa allelism o e alua ing ackabil-
i y and p ac ical conside a ions. Bioanalysis.
2024 May;:1–6.
[11] Richa ds S, Ama a adi L, Pillu la R, e al.
2016 Whi e Pape on ecen issues in bio-
analysis: ocus on bioma ke assay alida ion
(BAV): (Pa 3–LBA, bioma ke s and im-
munogenici y). Bioanalysis. 2016;8(23):2475–
2496.
[12] Piccoli S, Meh a D, Vi ali i A, e al. 2019
Whi e pape on ecen issues in bioanalysis:
FDA immunogenici y guidance, gene he -
apy, c i ical eagen s, bioma ke s and low
cy ome y alida ion (Pa 3 — ecommen-
da ions on 2019 FDA immunogenici y guid-
ance, gene he apy bioanaly ical challenges,
s a egies o c i ical eagen managemen ,
bioma ke assay alida ion, low cy ome y
alida ion & CLSI H62). Bioanalysis. 2019;
11(24):2207–2244.
[13] Fe nández-Me zle C, Acke mann B, Ga o-
olo F, e al. Bioma ke assay alida ion by
mass spec ome y. The AAPS Jou nal. 2022;
24(3):66.
[14] Pe zold A, Al in as A, And eoni L, e al. Neu-
o ilamen ELISA alida ion. Jou nal o Im-
munological Me hods. 2010 Jan;352(1-2):23–
31. A ailable om: h ps://disco e y.
ucl.ac.uk/id/ep in /95615/.
[15] Oeckl P, Ja del C, Salachas F, e al. Mul-
icen e alida ion o CSF neu o ilamen s as
diagnos ic bioma ke s o ALS. Amyo ophic
La e al Scle osis and F on o empo al Degen-
e a ion. 2016 Ap ;17(5-6):404–413.
[16] Wilson D, Chan D, Chang L, e al. De el-
opmen and mul i-cen e alida ion o a ully
au oma ed digi al immunoassay o neu o ila-
men ligh chain: owa d a clinical blood es
o neu onal inju y. Clinical Chemis y and
Labo a o y Medicine. 2024 Jan;62:322–331.
[17] Boo h RA, Be iaul D, Schneide R, e al.
Valida ion and gene a ion o age-speci ic e -
e ence in e als o a new blood neu o ilamen
ligh chain assay. Clinica Chimica Ac a. 2025
Sep;577:120447.
[18] Lu CH, Kalma B, Malaspina A, e al. A
me hod o solubilise p o ein agg ega es o
immunoassay quan i ica ion which o e comes
he neu o ilamen hook e ec . Jou nal o Neu-
oscience Me hods. 2011 Feb;195(2):143–150.
A ailable om: h ps://disco e y.ucl.
ac.uk/id/ep in /678078.
[19] Musso G, Zocca a o M, Gallo N, e al. Hook-
e ec in maglumi immunoassay o se um
an i-gad an ibodies in neu ological diso de s:
When “w ong” ma ix is he igh choice.
Clinica Chimica Ac a. 2024 May;558:119679.
[20] Commi ee o Medicinal P oduc s o Hu-
man Use. ICH guideline M10 on bioanaly ical
me hod alida ion and s udy sample analysis.
Eu opean Medical Agency. 2023 Jan;M10.
[21] He sey S, Kelle S, Ma hews J, e al. 2021
whi e pape on ecen issues in bioanal-
ysis: ISR o bioma ke s, liquid biopsies,
spec al cy ome y, inhala ion/o al & mul i-
speci ic bio he apeu ics, accu acy/LLOQ o
low cy ome y (pa 2 — ecommenda ions
on bioma ke s/CDx assays de elopmen &
alida ion, cy ome y alida ion & inno a-
ion, bio he apeu ics PK LBA egula ed bio-
analysis, c i ical eagen s & posi i e con ols
gene a ion). Bioanalysis. 2022 May;:627–692.
[22] Jucke M, Walke LC. Sel -p opaga ion
o pa hogenic p o ein agg ega es in neu-
odegene a i e diseases. Na u e. 2013 Sep;
501(7465):45–51.
[23] Pe zold A. Pa ial pa allelism plo s.
APPS Applied Sciences. 2024 Jan;
14(2):602. A ailable om: h ps:
//disco e y.ucl.ac.uk/id/ep in /
10185313/1/pa allel- 9-accep ed.pd .
[24] Winge RM. Gauss and non-euclidean geom-
e y. Bulle in o he Ame ican Ma hema ical
Socie y. 1925;31(7):356–358.
[25] Jenko szky L, Lake MJ, Solo ie V. János
Bolyai, Ca l F ied ich Gauss, Nikolai
C i ical Re iews in Clinical Labo a o y Sciences
Pa allelism Page 25
Lobache sky and he New Geome y:
Fo ewo d. Symme y. 2023 Ma ;15(3):707.
[26] Plikay is BD, Holde PF, Pais LB, e al. De-
e mina ion o pa allelism and nonpa allelism
in bioassay dilu ion cu es. Jou nal o Clini-
cal Mic obiology. 1994 Oc ;32(10):2441–2447.
[27] Meije ing E. A ch onology o in e pola ion:
om ancien as onomy o mode n signal and
image p ocessing. P oceedings o he IEEE.
2002 Ma ;90(3):319–342.
[28] Rodba d D, Hu D. S a is ical analysis o
adioimmunoassays and immuno adiome ic
(labelled an ibody) assays. A gene alized
weigh ed, i e a i e, leas -squa es me hod o
logis ic cu e i ing. NSA-31-000958. 1974;
:165–192OSTI ID:4277004.
[29] Rodba d D, Munson P, De Lean A. Imp o ed
cu e- i ing, pa allelism es ing, cha ac e -
iza ion o sensi i i y and speci ici y, alida-
ion, and op imiza ion o adioligand assays.
Radioimmunoassay and ela ed p ocedu es in
medicine; 1978.
[30] Yang H. Eme ging non-clinical bios a is ics
in biopha maceu ical de elopmen and man-
u ac u ing. Chapman and Hall/CRC; 2016.
[31] P en ice RL. A gene aliza ion o he p obi
and logi me hods o dose esponse cu es.
Biome ics. 1976 Dec;32(4):761.
[32] Dudley RA, Edwa ds P, Ekins RP, e al.
Guidelines o immunoassay da a p ocessing.
Clinical Chemis y. 1985 Aug;31(8):1264–
1271.
[33] Go schalk PG, Dunn JR. De e mining
he e o o dose es ima es and minimum
and maximum accep able concen a ions
om assays wi h nonlinea dose- esponse
cu es. Compu e Me hods and P og ams in
Biomedicine. 2005 Dec;80(3):204–215.
[34] Food and D ug Adminis a ion. M10 bioan-
aly ical me hod alida ion and s udy sample
analysis. 2022 ; 2023.
[35] Cullen VC, F edenbu g RA, E ans C, e al.
De elopmen and ad anced alida ion o an
op imized me hod o he quan i a ion o
aβ42 in human ce eb ospinal luid. AAPS
Jou nal. 2012 Sep;14:510–8.
[36] Fu Y, Ba kley D, Li W, e al. E alua ion,
iden i ica ion and impac assessmen o ab-
no mal in e nal s anda d esponse a iabil-
i y in egula ed lcms bioanalysis. Bioanalysis.
2020 Ap ;12(8):545–559.
[37] J Wool E. Lea ning how o in e p e ‘dange -
ous’ in e nal s anda d beha io s. Bioanalysis.
2019 Sep;11(18):1679–1684.
[38] Lawson NS, Ha en GT, Williams GW. Ana-
ly e s abili y in clinical chemis y quali y con-
ol ma e ials. CRC C i ical Re iews in Clini-
cal Labo a o y Sciences. 1982 Jan;17(1):1–50.
[39] Robe s CJ. P o ein agg ega ion and i s im-
pac on p oduc quali y. Cu en Opinion in
Bio echnology. 2014 Dec;30:211–217.
[40] Tu J, Benne P. Pa allelism expe imen s o
e alua e ma ix e ec s, selec i i y and sensi-
i i y in ligand-binding assay me hod de el-
opmen : p os and cons. Bioanalysis. 2017 Jul;
9(14):1107–1122.
[41] Loh TP, Lim CY, Se hi SK, e al. Ad ances
in in e nal quali y con ol. C i ical Re iews
in Clinical Labo a o y Sciences. 2023 May;
60(7):502–517.
[42] Faya P, Rauk AP, G i i hs KL, e al. A
cu e simila i y app oach o pa allelism es -
ing in bioassay. Jou nal o Biopha maceu ical
S a is ics. 2020 Ma ;30(4):721–733.
[43] Go schalk PG, Dunn JR. Measu ing pa -
allelism, linea i y, and ela i e po ency in
bioassay and immunoassay da a. Jou nal
o Biopha maceu ical S a is ics. 2005 May;
15(3):437–463.
[44] Guy S, Shephe d MF, Bowye AE, e al. How
o assess pa allelism in ac o assays: coe i-
cien o a ia ion o esul s wi h di e en dilu-
ions o slope a io? In e na ional Jou nal o
Labo a o y Hema ology. 2023 Dec;45(2):229–
240.
[45] Jonkman JN, Sidik K. Equi alence es ing
o pa allelism in he ou -pa ame e logis ic
model. Jou nal o Biopha maceu ical S a is-
ics. 2009 Aug;19(5):818–837.
[46] Smi h WC, Si ampalam GS. Concep ual and
s a is ical issues in he alida ion o analy ic
dilu ion assays o pha maceu ical applica-
ions. Jou nal o Biopha maceu ical S a is-
ics. 1998 Jan;8(4):509–532.
[47] Gil inan D. Selec ed opics in pa allel line
bioassay: Po ency es ima ion by pa allel line
biossay echniques. De elopmen s in biologi-
cals. 2002;107:47–56.
[48] Klein J, Capen R, Mancinelli R, e al. Valida-
ion o assays o use wi h combina ion ac-
cines. Biologicals. 1999 Ma ;27(1):35–41.
C i ical Re iews in Clinical Labo a o y Sciences
Pa allelism Page 32
Supplemen a y in o ma ion
Acknowledgmen s The sha ing o he aw da a
needed o cons uc he pa ial pa allelism plo s is
acknowledged om he ollowing au ho s: Oli e
Bannach, Die e Willbold and Ma leen J.A. Koel-
Simmelink.
Au ho con ibu ions A.P. designed and pe -
o med all expe imen s, analysed he da a, p e-
pa ed he igu es and w o e he manusc ip . D.C.
e iewed and co- egis e ed he p o ocol on PROS-
PERO and e iewed he manusc ip d a . J.P. e-
iewed he manusc ip d a .
Disclosu e o in e es The au ho s epo he e
a e no compe ing in e es s o decla e. This s udy
was no unded.
E hical app o al decla a ions This is a sys-
ema ic e iew on exis ing da a om s udies which
had e hical pe mission.
Da a a ailabili y & Da a deposi ion All da a
ha e been uploaded o Figsha e [150]. The py hon
code o he sea ch s a egy is openly a ailable o
download om he PRESTO egis y.
C i ical Re iews in Clinical Labo a o y Sciences
Supplemen a y Ma e ials o: “Pa allelism in Neu odegene a i e Bioma ke Tes s. . . ” Page 33
S a is ical me hods o de e mi-
na ion o pa allelism
The assessmen o pa allelism o igina es in he s a-
is ical li e a u e and is ypically amed as a bi-
na y decision: pa allelism is ei he p esen o ab-
sen [43]. This decision is es ed h ough s a is-
ical hypo heses embedded wi hin o mal models.
His o ically, he i s hypo hesis documen ed is Eu-
clid’s i h pos ula e. I is a hypo hesis ha can be
es ed, and as we lea ned om his o y, i can ake
cen u ies o disco e ha p esumed ma hema ical
solu ions e en ually u ned ou o be w ong. The
con empo a y s a is ical app oaches o es ing pa -
allelism all ha e in common ha hey build on he
p obabili y heo y [144]. Wi h ha comes he law
o la ge numbe s [143]. Only wi h la ge numbe s
he e is some gua an ee ha he a e ages om an-
dom e en s p o ide somehow s able long- e m e-
sul s [143,144]. Tha implies ha s a is ical me h-
ods o es ing o pa allelism on small numbe s, a e
open o c i icism.
One es ablished app oach o s a is ical es ing
o pa allelism is he ex a sum-o -squa es anal-
ysis o a iance (ANOVA) me hod, which com-
pa es esidual sum o squa es be ween nes ed mod-
els (RSSEnonpa ) [148, 149]. Fo example his ap-
p oach o ms he basis o using F and χ2s a is-
ics o es o pa allelism [43]. The au ho s com-
pa e isual and s a is ical es ing, showcas ing he
s eng hs and weaknesses o hese me hods. The
key message, o p esence o nonpa allelism, is o
decompose RSSEcons in o he componen o non-
pa allel o igine o RSSEnonpa and wha can be
desc ibed by andom a ia ion RSSE ee. Simpli-
ied, RSSEcons =RSSEnonpa +RSSE ee. He e he
de ini ion o nonpa allelism is he ex a e o ha
comes because o lack o simila i y be ween wo
cu es as pa o RSSEnonpa . The p ac ical cal-
cula ion o RSSEnonpa depends, and his is e y
impo an o ealise, on he assump ion o no mally
dis ibu ed da a and p esence o pa allelism. Wi h
his he law o la ge numbe s applies because he χ2
es is used, which wo ks on a dis ibu ed andome
a iable: d cons =Ns d +Nuk −(P+ 1).
To s a is ically es o his one needs o calcula e
he p obabili ies o he dose (x) o wo models.
1. The ee model (SSE ee) is desc ibed as:
SSE ee(ps d,puk) =
Ns d
X
i=1
ws d
iys d
i− (xs d
i;ps d)2
+
Nuk
X
i=1
wuk
iyuk
i− (xuk
i;puk)2
2. The cons ain model (SSEcons ) is desc ibed
as:
SSEcons ( , p) =
Ns d
X
i=1
ws d
iys d
i− (xs d
i;p)2
+
Nuk
X
i=1
wuk
iyuk
i− ( xuk
i;p)2
Ve y elegan ly he au ho s elabo a e, ci ing com-
p ehensi e e iews, ha highligh he limi a ions o
a ious s a is ical ac o s in his con ex [43]. The
applica ion o s a is ical models o pa allelism as-
sessmen is no wi hou limi a ions. As no ed by
Go schalk e al.,“The exis ence o simila i y be-
ween wo ma hema ical unc ions is no di icul o
de e mine. I is less s aigh o wa d o de e mine
he deg ee o pa allelism be ween wo unc ions ha
a e no exac ly simila .” [43].
Such demons a ion o lack o simila i y has been
p oposed o be sol ed by employing Bayesian o
equen is app oaches [42]. Wi h Bayesian pos-
e io p obabili y one can es pa allel equi alence
h ough:
p(γ, xL, xU) =
P min
ρmax
x∈[xL,xU]| (θ1, x)− (θ2, x +ρ)|< γ
da a
The auho s gi e p a ical examples, based on
simala ed da a, ha deli e a p- alue, using
his p obabilis ic app oach. The simula ed da a
gi e bioma ke concen a ion anging om 0.02–
125,000 a bi a y uni s (Table 1 in e e ence [42]).
In his example he s anda d e o (SE) equals
1
η2Va hAˆ
θi. The e o e, wi h α= 0.05
and δ= 0.85, he p obabili y calucla es as
P T2n−p>λ(ˆ
θ)−δ
SE = 0.062. This is no sig-
ni ican . Consequen ly, simila i y be ween cu es
canno be decla ed. This implies ha he e is no
e idence o pa allelism in bespoke example ( o i-
sual compa ison see Figu e 1 in e e ence [42]. This
is an ex eme example o demons a e lack o pa -
allelism, because he wo cu es c oss o e be ween
log2and log4).
C i ical Re iews in Clinical Labo a o y Sciences
Supplemen a y Ma e ials o: “Pa allelism in Neu odegene a i e Bioma ke Tes s. . . ” Page 34
As in oduced abo e, he 4PL and 5PL s anda d
cu es a e now equen ly used o i ing dose-
esponse cu es and consequen ly o ele ance o
discussed s a is ical e alua ions o pa allelism be-
ween sample and s anda d cu es [42–45]. One
inal wo d o cau ion is wa an ed he e, when em-
ploying highly pa ame e ized non-linea cu e mod-
els, which can lead o o e i ing, he e is a isk
o in oduce bias in o he pa allelism me ics ha
p o ide he da a ed ino abo e desc ibed s a is ical
models.
As Smi h no ed, “The condi ion [o pa allelism]
and i s impo ance a e ela i ely unknown o bio-
analy ical chemis s and many consul ing s a is i-
cians” [46]. In ha wo k, he au ho s illus a ed in-
e p e a i e challenges using simula ed da a, show-
ing how de ia ions in only a po ion o he cu e
can complica e decision-making. Fo his eason,
ou c i ical e iew p ima ily elied on isual me h-
ods o assess he p esence o absence o pa al-
lelism [23,26,47–49]. Visual inspec ion o e s an in-
ui i e and anspa en way o iden i y de ia ions,
making he concep accessible o a b oade scien i ic
audience, including labo a o y p ac i ione s who
may no ha e ad anced s a is ical expe ise. Im-
po an ly, we do no conside isual and s a is ical
me hods o be mu ually exclusi e. Ra he , hey a e
complemen a y: isual assessmen p o ides a p ac-
ical i s -line e alua ion, while s a is ical analyses
can se e as con i ma o y ools, o add ess speci ic
ques ions in selec ed si ua ions.
C i ical Re iews in Clinical Labo a o y Sciences
