PLOS One | h ps://doi.o g/10.1371/jou nal.pone.0335434 No embe 21, 2025 1 / 20
OPEN ACCESS
Ci a ion: Ma ínez-San os M, Yba a M,
Pi es ME, Ce esoni C, Ma ínez-López E,
Sancho-Pelluz J, e al. (2025) Ci cula ing
mic oRNAs as bioma ke s o diabe ic
e inopa hy s age iden i ica ion: A DTA
sys ema ic e iew and me a-analysis. PLoS One
20(11): e0335434. h ps://doi.o g/10.1371/
jou nal.pone.0335434
Edi o : Yalong Dang, Sanmenxia Cen al
Hospi al, Henan Uni e si y o Science and
Technilogy, CHINA
Recei ed: May 20, 2025
Accep ed: Oc obe 12, 2025
Published: No embe 21, 2025
Pee Re iew His o y: PLOS ecognizes he
bene i s o anspa ency in he pee e iew
p ocess; he e o e, we enable he publica ion
o all o he con en o pee e iew and
au ho esponses alongside inal, published
a icles. The edi o ial his o y o his a icle is
a ailable he e: h ps://doi.o g/10.1371/jou nal.
pone.0335434
Copy igh : © 2025 Ma ínez-San os e al
.
This
is an open access a icle dis ibu ed unde he
RESEARCH ARTICLE
Ci cula ing mic oRNAs as bioma ke s o
diabe ic e inopa hy s age iden i ica ion: A DTA
sys ema ic e iew and me a-analysis
Mi iam Ma ínez-San os1,2,3, Ma ía Yba a1,2,3, Ma ia E. Pi es1,2,3, Chia a Ce esoni1,2,3,
Elías Ma ínez-López4, Ja ie Sancho-Pelluz2,3, Ma ia Ol a 2,3*, Jo ge M. Ba cia1,2,3
1 Escuela de Doc o ado Uni e sidad Ca ólica de Valencia San Vicen e Má i , Valencia, Spain, 2 Facul ad
de Medicina y Ciencias de la Salud, Uni e sidad Ca ólica de Valencia San Vicen e Má i , Valencia, Spain,
3 Cen o de In es igación T aslacional San Albe o Magno, Uni e sidad Ca ólica de Valencia San Vicen e
Má i , Valencia, Spain, 4 Depa men o Gene al and Diges i e Su ge y, Hospi al Uni e si a io Doc o
Pese , Valencia, Spain
* ma ia.ol a@uc .es
Abs ac
Pu pose
To e alua e he diagnos ic accu acy o ci cula ing miRNAs in dis inguishing be ween
di e en diabe ic e inopa hy (DR) s ages in ype 2 diabe es melli us (T2DM).
Me hods
We conduc ed a sys ema ic e iew and me a-analysis in acco dance wi h PRISMA-
DTA and Coch ane guidelines. The p o ocol was no egis e es and no ex e nal und-
ing was ecei ed. A comp ehensi e sea ch was pe o med in PubMed, CENTRAL,
Scopus, Web o Science, ScienceDi ec , and ClinicalT ials (up o Janua y 2025) o
iden i y diagnos ic es accu acy s udies on ci cula ing miRNAs o DR. Eligible s ud-
ies included h ee p ede ined compa isons: heal hy con ols e sus DR (CTL s DR),
T2DM wi hou DR e sus DR (T2DM s DR), and non-p oli e a i e e sus p oli e a-
i e DR (NPDR s PDR). DR diagnosis was con i med using undus luo escein angi-
og aphy and/o undus examina ion. Two e iewe s independen ly conduc ed s udy
selec ion, da a ex ac ion, and isk o bias assessmen wi h QUADAS-2; ce ain y o
e idence was assessed using GRADE. Da a we e syn hesized using a bi a ia e
andom-e ec s me a-analysis, wi h subg oup analyses, me a- eg ession, and sensi-
i i y analyses o explo e he e ogenei y. Da a we e syn hesized ia a bi a ia e
andom-e ec s me a-analysis, wi h subg oup analyses, me a- eg ession, and sensi-
i i y es s o explo e he e ogenei y.
PLOS One | h ps://doi.o g/10.1371/jou nal.pone.0335434 No embe 21, 2025 2 / 20
Resul s
Six een s udies (1849 pa icipan s; 21 miRNAs) we e included. Fo CTL s DR (7
s udies), pooled sensi i i y was 77% (70–82) and speci ici y 84% (77–89), AUC 0.86
(0.82–0.89). Fo T2DM s DR (9 s udies), sensi i i y was 81% (75–86) and speci ic-
i y 80% (71–87), AUC 0.88 (0.84–0.91). Fo NPDR s PDR (12 s udies), sensi i i y
was 84% (79–87) and speci ici y 82% (76–88), AUC 0.90 (0.87–0.93). He e ogenei y
a ose chie ly om sample ma ix, no maliza ion s a egies and in e -s udy exp ession
ends. Pa ien selec ion posed he g ea es bias isk.
Conclusions
Ci cula ing miRNAs exhibi p omising diagnos ic accu acy o di e en ia ing among
a ious s ages o DR. Howe e , u u e la ge, p ospec i e s udies in di e se popula-
ions and s anda dized p e-analy ical p o ocols a e equi ed o con i m and ansla e
hese indings.
In oduc ion
Diabe ic e inopa hy (DR) is one o he leading causes o ision loss among
wo king-age adul s wo ldwide [1]. As he global p e alence o ype 2 diabe es melli-
us (T2DM) con inues o ise, so does he incidence o DR, posing a g owing public
heal h conce n [2]. The condi ion p og esses h ough well-de ined s ages, s a ing
wi h non-p oli e a i e diabe ic e inopa hy (NPDR) and po en ially ad ancing o
p oli e a i e diabe ic e inopa hy (PDR), cha ac e ized by neo ascula iza ion and an
inc eased isk o e inal de achmen o hemo hage [3]. The cumula i e occu ence
o p og ession om NPDR o ision- h ea ening complica ions has been es ima ed a
app oxima ely 14–16%. wi h he isk o p og ession o ision loss ising signi ican ly,
eaching nea ly 58% [4,5]. Accu a e de ec ion and s aging o DR a e he e o e essen-
ial o imely clinical decision-making and o p e en ing i e e sible ision loss [4].
Cu en diagnos ic me hods o DR include undus examina ion and luo escein
angiog aphy (FA) among o he s, FA is he gold s anda d o s aging due o i s high
sensi i i y in de ec ing mic o ascula damage [6,7]. Fundus examina ion is mo e
accessible bu less sensi i e, pa icula ly in ea ly disease [8]. FA, while accu a e, is
in asi e and esou ce-in ensi e, limi ing i s ou ine use [9]. These limi a ions highligh
he need o non-in asi e, accessible bioma ke s ha could suppo ea ly de ec ion
and imp o e sc eening and isk s a i ica ion in b oade clinical se ings [10,11].
Ci cula ing mic oRNAs (miRNAs) ha e eme ged as p omising candida es o his
ole. MiRNAs a e small, non-coding RNAs in ol ed in he pos - ansc ip ional eg-
ula ion o gene exp ession and a e de ec able in a ious biological luids, including
se um, plasma, aqueous humo , and ex acellula esicles [12,13]. Thei high s abili y
in ci cula ion, disease-speci ic exp ession pa e ns, and accessibili y h ough non-
in asi e sampling make hem a ac i e ools o bioma ke disco e y [11]. Al hough
nume ous s udies ha e in es iga ed he diagnos ic po en ial o ci cula ing miRNAs in
e ms o he C ea i e Commons A ibu ion
License, which pe mi s un es ic ed use,
dis ibu ion, and ep oduc ion in any medium,
p o ided he o iginal au ho and sou ce a e
c edi ed.
Da a a ailabili y s a emen : All ele an da a
a e wi hin he manusc ip and i s Suppo ing
In o ma ion iles.
Funding: The p esen wo k ecei ed in e nal
unds om Cen o de In es igación T aslacional
SanAlbe o Magno (CITSAM, UCV) and ex e nal
unds om Agencia Es a al de In es igación
Española (PID2020-117875GB-10), Ins i u o
de Salud Ca los III (ISCIII, PI21/00083) and
he Eu opean Union esea ch und, HORIZON
MSCA 2021-DN-01-01_RETORNA 101073316
and Gene ali a Valenciana ACIF 2023-128-001.
The unde s had no ole in he s udy design,
da a collec ion and analysis, decision o pub-
lish, o p epa a ion o he manusc ip .
Compe ing in e es s: The au ho s ha e
decla ed ha no compe ing in e es s exis .
PLOS One | h ps://doi.o g/10.1371/jou nal.pone.0335434 No embe 21, 2025 3 / 20
DR, ma ked he e ogenei y in sample ypes, analy ical me hods (e.g., RT-qPCR, mic oa ays, NGS), and a ge miRNAs
has limi ed compa abili y ac oss s udies [14–18].
No ably, one p io me a-analysis has e iewed he use o ci cula ing miRNAs o DR de ec ion [19], i did no pe o m
any s a i ica ion by disease s age o ype o con ol g oup. This ep esen s a no able gap in he cu en li e a u e, as he
abili y o dis inguish ea ly om ad anced s ages o DR is c i ical o clinical decision-making.
The e o e, he aim o his sys ema ic e iew and me a-analysis is o e alua e he diagnos ic accu acy o ci cula ing
miRNAs in DR. Speci ically, we assess hei pe o mance in dis inguishing be ween heal hy con ols s DR pa ien s, T2DM
s DR, and NPDR s PDR s ages. This s a i ied app oach add esses a c i ical unme need by sys ema ically analyzing
he diagnos ic alue o miRNAs ac oss clinically ele an disease s ages and con ol popula ions.
Me hods
Sea ch s a egy
This sys ema ic e iew and me a-analysis we e conduc ed ollowing he P e e ed Repo ing I ems o PRISMA-
Diagnos ic Tes Accu acy (PRISMA-DTA) [20] o u he in o ma ion consul (S1 and S2 Tables in S1 File). This sys-
ema ic e iew was no egis e ed. We pe o med ex ensi e sea ch in he ollowing da abases: PubMed, CENTRAL,
Scopus, Web o Science, Science Di ec , and Clinical T ials. The las upda e o his e iew was on Janua y 20, 2025.
The sea ch s a egy was designed o iden i y ele an s udies e alua ing he diagnos ic accu acy o miRNAs in DR. A
combina ion o Medical Subjec Headings (MeSH) e ms and ee- ex keywo ds was applied, using Boolean ope a o s
(AND, OR, NOT) o e ine he sea ch. The main opics included: DR, miRNAs, exp ession p o iling, bioma ke s, and
biological sample ypes (se um, plasma, aqueous humo , ex acellula esicles). Fo he comple e sea ch s a egies
please e e o (S1 Tex in S1 File).
Eligibili y c i e ia
This sys ema ic e iew and me a-analysis ollow he Popula ion–Index es –Ta ge condi ion (PIT) s uc u e, as ecom-
mended by he Coch ane. We ocused on (P) Popula ion: human pa icipan s diagnosed wi h DR a a ious s ages (NPDR
o PDR), as well as indi iduals wi hou DR, including heal hy con ols and pa ien s wi h T2DM wi hou DR; (I) Index es :
miRNA exp ession le els measu ed in se um, plasma, aqueous humo , o ex acellula esicles, using alida ed ech-
niques such as quan i a i e eal- ime (RT-qPCR), mic oa ays, o nex -gene a ion sequencing (NGS); and (T) Ta ge
condi ion: DR. We included s udies published om 2014 onwa d, as long as hey e alua ed he diagnos ic accu acy o
miRNAs ac oss di e en s ages o DR con i med using FA and undus examina ion. To be eligible, s udies also needed
o epo key accu acy me ics such as: sensi i i y, speci ici y, o a ea unde he cu e (AUC). On he o he hand, we
excluded s udies ha ocused on o he ypes o diabe es, in i o o in animal models.
S udy selec ion and da a ex ac ion
Two independen e iewe s (M.M-S and E.M-L.) assessed s udies o eligibili y based on p ede ined inclusion and exclu-
sion c i e ia. Full- ex a icles o po en ially ele an s udies we e e iewed, and any disc epancies we e esol ed by con-
sensus; when necessa y, a hi d e iewe (M.O.) p o ided a bi a ion.
F om each included s udy, we ex ac ed da a on s udy cha ac e is ics (au ho , yea , coun y, design), popula ion
de ails, biological sample ype, index es pla o m, no maliza ion s a egy, and diagnos ic accu acy me ics such
as: sensi i i y, speci ici y and AUC. The comple e ex ac ion da ase is p o ided in S3 Table in S1 File. In o ma ion
on miRNA exp ession ends and cu -o alues was eco ded when a ailable. Duplica e eco ds we e iden i ied and
emo ed in EndNo e X9, ollowing he me hodology desc ibed by Kwon e al. [21]. All e e ences we e managed using
EndNo e X9 so wa e.
PLOS One | h ps://doi.o g/10.1371/jou nal.pone.0335434 No embe 21, 2025 4 / 20
Quali y assessmen
The isk o bias and applicabili y conce ns o he included s udies we e assessed using he QUADAS-2 ool [22], applied
independen ly by wo e iewe s ac oss he ou s anda d domains: pa ien selec ion, index es , e e ence s anda d, and
low/ iming. Disc epancies we e esol ed by consensus. In addi ion, he GRADE app oach [23], adap ed o diagnos ic
es accu acy s udies, was used o e alua e he o e all s eng h o e idence and guide ecommenda ions.
S a is ical analysis
We calcula ed 2 × 2 con ingency ables ue posi i es (TP), alse posi i es (FP), alse nega i es (FN), and ue nega i es
(TN) o each included s udy. Pooled sensi i i y, speci ici y, likelihood a ios (PLR, NLR), diagnos ic odds a io (DOR),
and SROC cu e we e es ima ed using a bi a ia e andom-e ec s model, ecommended o diagnos ic es accu acy
me a-analyses. This model join ly accoun s o sensi i i y and speci ici y, including hei co ela ion and be ween-s udy
a iabili y. Since mos s udies did no epo diagnos ic h esholds, a hie a chical HSROC model was no applicable.
Ins ead, we gene a ed empi ical SROC cu es om bi a ia e es ima es o isualize o e all diagnos ic pe o mance.
He e ogenei y was assessed using Coch an’s Q- es and he I2 s a is ic de i ed om he bi a ia e model. An I2 alue
abo e 50% o a p- alue < 0.05 was conside ed indica i e o subs an ial he e ogenei y. To in es iga e i s po en ial sou ces,
p e-speci ied uni a iable me a- eg essions we e pe o med wi hin he bi a ia e amewo k, using a iables such as sam-
ple size, coun y o o igin, biological specimen ype no maliza ion s a egy, and miRNA exp ession pa e n. To assess he
obus ness o ou indings, sensi i i y analyses we e ca ied ou using lea e-one-ou me hods and in luence diagnos ics.
Model assump ions we e e alua ed ia esidual de iance plo s and bi a ia e no mali y es s. Finally, o e alua e po en-
ial publica ion bias, Deeks’ unnel plo asymme y es was pe o med. Fagan nomog ams we e gene a ed o ansla e
likelihood a ios in o pos - es p obabili ies, acili a ing clinical in e p e a ion. Pe cen ages a e epo ed as whole numbe s
ounded o he nea es in ege ; exac es ima es and 95% con idence in e als a e p o ided in he co esponding ables
and igu es. All s a is ical analyses we e pe o med a 95% CI, using STATA 18 (STATA Co po a ion, College S a ion, TX,
USA), inco po a ing he MIDAS package o me a-analysis o DTA. Addi ional de ails a e p o ided in S4, S5 and S6 Tables
in S1 File.
Resul s
S udy cha ac e is ics and quali y assessmen
A o al o 454 a icles we e iden i ied om a ious da abases, including: CENTRAL (n = 2), Scopus (n = 122), PubMed
(n = 77), Clinical T ials (n = 1), Web o Science (n = 125), and Science Di ec (n = 127). Addi ionally, 6 abs ac s we e
e ie ed om g ey li e a u e (ARVO). A e he emo al o 270 duplica es, 180 eco ds we e sc eened based on i le and
abs ac , esul ing in 60 ull- ex a icles being assessed o eligibili y. Among hese, 38 s udies we e excluded o he ol-
lowing easons: in i o s udies (n = 8), e iews (n = 4), animal s udies (n = 6), and s udies wi h mic o ascula complica ions
o he han DR (n = 20). Ul ima ely, 16 s udies me he inclusion c i e ia and we e included in he quan i a i e and quali a i e
syn hesis me a-analysis (Fig 1).
These 16 s udies analyzed 1.849 pa ien s and in es iga ed 21 dis inc miRNAs. Among hese miRNAs, 6 we e de ec ed
in plasma, 14 in se um, and 1 in exosomes. In e ms o s udy design, 14 s udies we e case-con ol [14–18,24–31], while 2
we e c oss-sec ional s udies [32,33]. Geog aphically, he majo i y o s udies we e om China (n = 11)
[14,17,18,24,25,27,28,30,33,34], ollowed by Egyp (n = 3) [16,26,31], I aly (n = 1) [29], and Indonesia (n = 1) [32]. Rega d-
ing he analy ical app oach, 3 s udies assessed miRNA panels [15,27,29], whe eas he emaining 13 s udies ocused on
single miRNA analysis [14,16–18, 24–26,28,30,32–34]. In e ms o compa ison g oups, 9 s udies ha e a single ype o
compa ison, whe eas 7 s udies had mul iple compa isons ac oss di e en disease s ages. The dis ibu ion o compa isons
was as ollows: 6 s udies analyzed CTL s DR [24,25,27,32–34], 7 s udies analyzed T2DM s DR [14,16,24,26,28–30],
PLOS One | h ps://doi.o g/10.1371/jou nal.pone.0335434 No embe 21, 2025 5 / 20
Fig 1. Flow diag am de ailing he selec ion o s udies included in he diagnos ic accu acy me a-analysis.
h ps://doi.o g/10.1371/jou nal.pone.0335434.g001
PLOS One | h ps://doi.o g/10.1371/jou nal.pone.0335434 No embe 21, 2025 6 / 20
Table 1. Quan i a i e and quali a i e cha ac e is ics o included s udies.
S udy
(yea )
Coun-
y
S udy
Design
Com-
pa a-
ion
miRNAs Exp es-
sion
Speci-
men
Me hod No mal-
iza ion
TP TN FP FN Sen
%
Spe
%
To al
(n)
Jiang e
al., 2017
China Case
con ol
CTL
s DR
miR-21 Up Plasma RT-qPCR U6 82 104 11 42 66.1 90.4 239
Qin e
al., 2017
China Case
con ol
CTL
s DR
miR-126 Down Plasma RT-qPCR U6 32 53 6 7 81.25 90.34 98
Wan e
al., 2017
China Case
con ol
CTL
s DR
miR-7 Down Se um RT-qPCR MIR2911 58 54 20 18 76 73 150
Wan e
al., 2017
China Case
con ol
CTL
s DR
miR-7 (exosome) Down Exo-
some
RT-qPCR MIR2911 57 57 17 18 75 77 149
Liu e
al., 2018
China C oss
Sec-
ional
CTL
s DR
miR-211 Up Se um RT-qPCR U6 54 29 4 10 85 87 97
Li e al.,
2019
China Case
con ol
CTL
s DR
miR-4448, miR-
338-3p, miR-190a-5p,
mi 485-5p, miR-9-5p
Up/
Down
Se um RNA-Seq DESeq2 9 10 1 1 90 90.9 21
Su as-
mia i e
al., 2023
Indo-
nesia
C oss
sec-
ional
CTL
s DR
miR-126 Down Se um RT-qPCR miRNA-
328-3p
9 8 2 2 75 50 21
Qin e
al., 2017
China Case
con ol
T2DM
s DR
miR-126 Down Plasma RT-qPCR U6 69 42 2 12 84.8 94.9 125
Shake
e al.,
2019
Egyp Case
con ol
T2DM
s DR
miR-20b Down Se um RT-qPCR SNORD68 31 18 12 19 62 60 80
Shake
e al.,
2019
Egyp Case
con ol
T2DM
s DR
miR-17-3p Down Se um RT-qPCR SNORD68 46 17 13 4 92 56.7 80
Yin e
al., 2020
China Case
con ol
T2DM
s DR
miR-210 Up Se um RT-qPCR U6 92 32 8 18 83.6 80 150
San a-
i o e
al., 2021
I aly Case
con ol
T2DM
s DR
miR-25-3p, miR-
320b, miR-495-3p
Up/
Down
Plasma RT-qPCR miR-19-5p,
miR-
125a-5p
17 9 1 3 85 85 30
Wang e
al., 2021
China Case
con ol
T2DM
s DR
miR-374a Up Se um RT-qPCR U6 110 58 12 27 80.3 82.9 207
Saleh e
al., 2022
Egyp Case
con ol
T2DM
s DR
miR-93 Down Se um RT-qPCR miRNA-16 68 69 11 12 85 86 160
Saleh e
al., 2022
Egyp Case
con ol
T2DM
s DR
miR-152 Up Se um RT-qPCR miRNA-16 68 58 22 12 85 72 160
Zhao e
al., 2023
China Case
con ol
T2DM
s DR
miR-221-3p Up Se um RT-qPCR U6 110 84 18 48 69.7 82.3 260
Qing e
al., 2014
China Case
con ol
NPDR
s
PDR
miR-21, miR-181c,
miR-1179
Up Se um RT-qPCR U6 74 86 4 16 82 95 180
Jiang e
al., 2017
China C oss
Sec-
ional
NPDR
s
PDR
miR-21 Up Plasma RT-qPCR U6 37 58 15 14 72.5 79.5 124
Shake
e al.,
2019
Egyp Case
con ol
NPDR
s
PDR
miR-20b Down Se um RT-qPCR SNORD68 14 23 7 6 70 76.6 50
Shake
e al.,
2019
Egyp Case
con ol
NPDR
s
PDR
miR-17-3p Down Se um RT-qPCR SNORD68 10 24 6 10 50 80 50
(Con inued)
PLOS One | h ps://doi.o g/10.1371/jou nal.pone.0335434 No embe 21, 2025 7 / 20
9 s udies analyzed NPDR s PDR [15–18,26,28,30,31,34]. The p ima y echnique employed o miRNA de ec ion was
RT-qPCR in 15 s udies, while RNA-seq was used in one s udy [27]. Rega ding no maliza ion me hods, 9 s udies u ilized
U6 [14,15,18,24,28,30,31,33,34], while he emaining 7 s udies applied di e en miRNA no maliza ion me hods
[16,17,25–27,29,32] (Table 1).
The QUADAS-2 assessmen showed ha he main sou ces o bias came om how pa ien s we e selec ed and
how he index es was applied. A high isk o bias was mos ly linked o pa ien selec ion, o en due o non-
andom sampling me hods and e ospec i e s udy designs. In he index es domain, he e we e some conce ns
as well, pa icula ly in s udies ha didn’ clea ly de ine diagnos ic h esholds ahead o ime. On he o he hand, he
e e ence s anda d and he low and iming o he s udies gene ally showed a lowe isk o bias, wi h mos s udies
ollowing p ope diagnos ic p ocedu es and easonable imelines. When i came o applicabili y, mos s udies posed
low conce n ac oss all domains. Howe e , a ew showed mino issues ela ed o he index es , mainly because o
di e ences in how i was applied ac oss s udies. A de ailed summa y o he isk o bias and applicabili y conce ns is
shown in (Fig 2).
S udy
(yea )
Coun-
y
S udy
Design
Com-
pa a-
ion
miRNAs Exp es-
sion
Speci-
men
Me hod No mal-
iza ion
TP TN FP FN Sen
%
Spe
%
To al
(n)
Hui e
al, 2019
China Case
con ol
NPDR
s
PDR
miR-126 Down Plasma RT-qPCR cel-miR-
39-3p
28 31 12 8 78.4 73.2 79
Ma e
al., 2019
China Case
con ol
NPDR
s
PDR
miR-93 and miR-21 Up Plasma RT-qPCR U6 39 30 4 3 92 89 76
Ma e
al., 2019
China Case
con ol
NPDR
s
PDR
miR-93 Up Plasma RT-qPCR U6 37 28 6 5 89 81 76
Ma e
al., 2019
China Case
con ol
NPDR
s
PDR
miR-21 Up Plasma RT-qPCR U6 38 24 10 4 90 71 76
Yin e
al., 2020
China Case
Con ol
NPDR
s
PDR
miR-210 Up Se um RT-qPCR U6 41 45 15 9 84.2 78.9 110
Wang e
al., 2021
China Case
Con ol
NPDR
s
PDR
miR-374a Up Se um RT-qPCR U6 61 51 13 10 84.2 78.8 135
Saleh e
al., 2022
Egyp Case
con ol
NPDR
s
PDR
miR-93 Down Se um RT-qPCR miRNA-16 34 25 15 6 85 63 80
Saleh e
al., 2022
Egyp Case
con ol
NPDR
s
PDR
miR-152 Up Se um RT-qPCR miRNA-16 34 32 8 6 85 80 80
Salem
e al.,
2022
Egyp Case
con ol
NPDR
s
PDR
miR-181c Up Se um RT-qPCR U6 54 60 0 6 90 100 120
Salem
e al.,
2022
Egyp Case
con ol
NPDR
s
PDR
miR-1179 Up Se um RT-qPCR U6 54 48 12 6 90 80 120
TP: T ue Posi i es, TN: T ue Nega i es; FP: False Posi i es; FN: False Nega i es; Sen: Sensi i i y.
h ps://doi.o g/10.1371/jou nal.pone.0335434. 001
Table 1. (Con inued)
PLOS One | h ps://doi.o g/10.1371/jou nal.pone.0335434 No embe 21, 2025 8 / 20
Diagnos ic accu acy o miRNAs in CTL s DR, T2DM s DR, and NPDR s PDR compa isons
A o al o 7 s udies con ibu ed da a o he compa ison o CTL s DR. The pooled es ima es om he andom‐e ec s
model showed a summa y sensi i i y o 77% (70–82), wi h an I2 o 47%, and a summa y speci ici y o 84% (77–89),
wi h an I2 o 62%, indica ing mode a e he e ogenei y (Fig 3A-B). The SROC cu e yielded an AUC o 0.86 (0.84–0.92)
(Fig 3C), sugges ing mode a e‐ o‐high o e all accu acy. To unde s and he clinical ele ance o hese indings, a Fagan
nomog am was cons uc ed using he ac ual p e- es p obabili y o DR in he popula ion s udied 22% [2]. The plo e ealed
a posi i e miRNA esul aises he p obabili y o ha ing he disease o 58%, while a nega i e esul educes i o jus 7%
(Fig 4A). In pa allel, a sca e ma ix was used o isualize he ela ionship be ween he likelihood a ios ac oss all miRNA
es s included in his compa ison. The pooled posi i e likelihood a io (PLR) was 4.77 (3.19–7.13), indica ing ha pa ien s
wi h DR a e nea ly i e imes mo e likely o es posi i e han hose wi hou he disease. On he o he hand, he pooled
nega i e likelihood a io (NLR) was 0.29 (0.23–0.37), sugges ing a no able educ ion in he p obabili y o disease ollowing
a nega i e esul (Fig 4A-B). Las ly, Deeks’ es indica ed no signi ican p esence o publica ion bias (p = 0.27) (Fig 5A).
Da a om 9 s udies in T2DM s DR compa ison indica ed a summa y sensi i i y o 81% (75–86), wi h a speci ici y o
80% (71–87), wi h an I2 o 73%, sugges ing mode a e o high he e ogenei y in bo h measu es (Fig 3D-E). The SROC
cu e epo ed an AUC o 0.88 (0.80–0.95), indica ing s ong o e all diagnos ic accu acy (Fig 3F). Fagan nomog am
was cons uc ed using he ac ual p e- es p obabili y o DR in his popula ion 28% [35]. The plo showed ha a posi i e
miRNA esul would inc ease he p obabili y o de ec ing DR o 61%, while a nega i e esul would lowe he likelihood
o jus 8%. This e lec s a meaning ul shi in pos - es p obabili ies, ein o cing he ole o miRNAs in helping clinicians
di e en ia e be ween uncomplica ed diabe es and he onse o DR (Fig 4C). The sca e ma ix illus a ed he dis ibu-
ion o diagnos ic pe o mance showed he PLR was 4.1 (2.8–6.1), sugges ing ha pa ien s wi h DR a e app oxima ely
ou imes mo e likely o es posi i e compa ed o hose wi h T2DM alone. Meanwhile, he NLR was 0.23 (0.17–0.32),
indica ing a subs an ial dec ease in he p obabili y o disease ollowing a nega i e es (Fig 4D). The Deeks’ unnel plo
analysis showed no signi ican e idence o publica ion bias (p = 0.81), indica ing a low isk ha he esul s a e in luenced
by publica ion bias (Fig 5B).
Twel e s udies we e included in his analysis compa ing NPDR s PDR. The me a-analysis ound a pooled sensi i i y o 84%
(79–87) wi h an I2 o 56% and a speci ici y o 82% (76–88) wi h an I2 o 77%, showing mode a e- o-high he e ogenei y, (Fig 3G-
H). The SROC cu e demons a ed an AUC o 0.90 (0.82–0. 91), poin ing o high o e all diagnos ic accu acy (Fig 3I). The Fagan
nomog am was cons uc ed using a p e- es p obabili y o 17% [5], based on he ac ual p e alence obse ed in his popula ion.
Fig 2. Risk o bias and applicabili y assessmen using he QUADAS-2 ool.
h ps://doi.o g/10.1371/jou nal.pone.0335434.g002
PLOS One | h ps://doi.o g/10.1371/jou nal.pone.0335434 No embe 21, 2025 9 / 20
Fig 3. Fo es plo s o sensi i i y and speci ici y and SROC cu es showing diagnos ic accu acy o miRNAs: CTL s DR (A-C), T2DM s DR (D-
F), and NPDR s PDR (G-I) compa isons.
h ps://doi.o g/10.1371/jou nal.pone.0335434.g003
PLOS One | h ps://doi.o g/10.1371/jou nal.pone.0335434 No embe 21, 2025 16 / 20
Biological insigh s and candida e miRNAs
Among he s udies included, se e al miRNAs appea ed ecu en ly, highligh ing hei po en ial ele ance ac oss di e en expe -
imen al con ex s. The mos equen ly epo ed was miR-126 [17,24,32], iden i ied in h ee sepa a e s udies. This miRNA has
Table 6. Ce ain y o e idence o he diagnos ic accu acy o miRNAs acco ding o he GRADE app oach in T2DM s DR.
P e es p obabili y (global p e alence o DR in T2DM pa ien s): 28.41%$
Pooled sensi i i y: 80% IC95% (0.77–0.83)
Pooled speci ici y: 80% IC95% (0.76–0.83)
T2DM s DR
Ou come Numbe
o s udies
S udy design Risk
o bias
Indi ec
e idence
Incon-
sis ency
Imp e-
cision
Publica-
ion bias
E ec
x1000*
Quali y
T ue posi i es 7 (1012
pa ien s)
Case-con ol (n = 7) High Se ious2Se ious3No
se ious
No
de ec ed
227
(22.7%)
⊕⊕◯◯
Low
T ue nega i es 7 (1012
pa ien s)
Case-con ol (n = 7) High Se ious2Se ious3No
se ious
No
de ec ed
570
(57%)
⊕⊕⊕◯
Mode a e
False Posi i es 7 (1012
pa ien s)
Case-con ol (n = 7) High Se ious2Se ious3No
se ious
No
de ec ed
146
(14.6%)
⊕⊕◯◯
Ve y low
False nega i es 7 (1012
pa ien s)
Case-con ol (n = 7) High Se ious2Se ious3No
se ious
No
de ec ed
57
(5.7%)
⊕⊕⊕◯
Mode a e
*Numbe o pa ien s pe 1000 es ed o a p e alence o 28.41%
2: The e idence was a ed as se ious due o a iabili y in sample ypes (plasma s. se um) and di e ences in miRNA exp ession pa e ns (o e exp es-
sion s. unde exp ession), which may a ec he applicabili y o he esul s.
3: He e ogenei y was a ed as se ious due o high inconsis ency, wi h an I2 o 72.5% o sensi i i y and 73.1% o speci ici y, indica ing subs an ial a i-
abili y among s udies.
$: Hashemi, H., Rez an, F., Pakzad, R., Ansa ipou , A., Heyda ian, S., Yek a, A., … Khabazkhoob, M. (2021). Global and Regional P e alence o Diabe -
ic Re inopa hy; A Comp ehensi e Sys ema ic Re iew and Me a-analysis. Semina s in Oph halmology, 37(3), 291–306.
h ps://doi.o g/10.1371/jou nal.pone.0335434. 006
Table 7. Ce ain y o e idence o he diagnos ic accu acy o miRNAs acco ding o he GRADE app oach in NPDR s PDR.
P e es p obabili y (global p e alence o PDR): 17%$
Pooled sensi i i y: 84% IC95% (0.81–0.86)
Pooled speci ici y: 82% IC95% (0.79–0.85)
NPDR s PDR
Ou come Numbe o
s udies
S udy design Risk o
bias
Indi ec
e idence
Inconsis-
ency
Imp ecision Publica ion
bias
E ec
x1000*
Quali y
T ue posi i es 9 (954 pa ien s) Case-con ol
(n = 9)
High Se ious2Se ious3No se ious No de ec ed 142
(14.2%)
⊕⊕◯◯
Low
T ue nega i es 9(954 pa ien s) Case-con ol
(n = 9)
High Se ious2Se ious3No se ious No de ec ed 678
(67.8%)
⊕⊕⊕◯
Mode a e
False Posi i es 9 (954 pa ien s) Case-con ol
(n = 9)
High Se ious2Se ious3No se ious No de ec ed 152
(15.2%)
⊕◯◯◯
Ve y low
False nega i es 9 (954 pa ien s) Case-con ol
(n = 9)
High Se ious2Se ious3No se ious No de ec ed 28
(2.8%)
⊕⊕⊕◯
Mode a e
*Numbe o pa ien s pe 1000 es ed o a p e alence o 17%
2: The e idence was a ed as se ious due o a iabili y in sample ypes (plasma s. se um) and di e ences in miRNA exp ession pa e ns.
3: He e ogenei y was a ed as se ious due o high inconsis ency, wi h an I2 o 55.6% o sensi i i y and 77.1% o speci ici y, indica ing subs an ial a i-
abili y among s udies.
$: Yang QH, Zhang Y, Zhang XM, Li XR. P e alence o diabe ic e inopa hy, p oli e a i e diabe ic e inopa hy and non-p oli e a i e diabe ic e inopa hy in
Asian T2DM pa ien s: a sys ema ic e iew and Me a-analysis. In J Oph halmol. 2019 Feb 18;12(2):302–311
h ps://doi.o g/10.1371/jou nal.pone.0335434. 007
PLOS One | h ps://doi.o g/10.1371/jou nal.pone.0335434 No embe 21, 2025 17 / 20
been linked o DR h ough i s ole in ascula in eg i y and angiogenesis egula ion, wi h dec eased le els obse ed in plasma
and i eous samples o a ec ed pa ien s [37]. miR-21, was also epo ed in 3 s udies [15,18,34] and has been implica ed in
e inal angiogenesis and in lamma ion in he diabe ic con ex [38]. Addi ionally, miR-181c [15,25], miR-1179 [15,31] and miR-
93 [16,18] we e each epo ed in wo s udies, miR-93 has been associa ed wi h inc eased DR isk in T2DM [39].
Clinical ansla ion and he need o s anda diza ion
Ou me a-analysis e ealed conside able he e ogenei y. One majo sou ce o a iabili y s ems om he ype o biologi-
cal ma ix used. Al hough bo h plasma and se um we e employed, plasma may o e a mo e eliable p o ile o ci cula ing
miRNAs [40]. Unlike se um, plasma a oids he con ounding elease o pla ele -de i ed miRNAs du ing coagula ion, which
can dis o exp ession p o iles and lead o inconsis en esul s [41]. Ano he key issue is he lack o a uni e sally accep ed
in e nal con ol o no maliza ion. While U6 small nuclea RNA was he mos commonly used e e ence gene ac oss
included s udies, i is p edominan ly nuclea and may deg ade in cell- ee condi ions such as plasma o se um, hus
in oducing bias [42]. Al e na i e e e ence miRNAs like miR-16-5p ha e demons a ed g ea e s abili y and may ep esen
mo e app op ia e no maliza ion candida es in ex acellula RNA esea ch [43].
O he me hodological ac o s, including he ime elapsed be ween sample collec ion and p ocessing [44], he ype o
miRNA ex ac ion ki , and de ec ion pla o m used, con ibu e u he o be ween-s udy a iabili y [45,46]. This lack o uni-
o mi y hinde s he compa abili y o esul s and educes hei gene alizabili y ac oss di e en clinical con ex s.
To o e come hese ba ie s, he ield u gen ly needs s anda dized p e-analy ical p o ocols, consensus-based epo ing
guidelines, and he de elopmen o alida ed mul i-miRNA diagnos ic panels [47]. Fu he mo e, aining and ce i ica ion
o echnical pe sonnel in ol ed in miRNA handling and da a in e p e a ion would help minimize human e o and inc ease
ep oducibili y [48,49]. Wi hou hese imp o emen s, he in eg a ion o miRNAs in o clinical diagnos ic wo k lows will
emain heo e ical, ega dless o hei p omising s a is ical pe o mance [50].
Limi a ions
This me a-analysis has se e al limi a ions ha mus be conside ed when in e p e ing he indings. Fi s , he p o ocol was no eg-
is e ed in P ospe o. Second, a p edominan numbe o he included s udies we e conduc ed in Chinese popula ions, which may
in oduce demog aphic bias and limi he gene alizabili y o he indings o o he e hnic g oups, pa icula ly Wes e n coho s. While
his does no comp omise in e nal alidi y, i unde sco es he need o b oade geog aphic ep esen a ion in u u e s udies.
Thi d, he s udy designs we e p ima ily case-con ol and c oss-sec ional, which a e inhe en ly mo e p one o bias han
p ospec i e coho s udies [51]. These designs may o e es ima e diagnos ic accu acy due o spec um bias o inapp o-
p ia e pa ien selec ion [52]. Fu he mo e, andomized clinical ials a e en i ely lacking, e lec ing bo h ope a ional and
me hodological challenges in conduc ing such s udies in his ield.
Fou h, p e-analy ical and analy ical he e ogenei y was subs an ial ac oss s udies. Va ia ions in he biological sample
ype, no maliza ion s a egies, and miRNA isola ion and de ec ion pla o ms con ibu ed signi ican ly o inconsis ency
in esul s. Fi h, mos s udies did no epo explici diagnos ic cu -o alues o indi idual miRNAs. Al hough bi a ia e
andom-e ec s modeling allows o obus es ima ion o sensi i i y, speci ici y, and AUC, he absence o h eshold alues
limi s he clinical applicabili y o he indings [53]. Diagnos ic h esholds a e essen ial o guiding eal-wo ld decision-
making and should be es ablished and alida ed in u u e esea ch. The lack o cu -o s also e lec s a b oade issue: lim-
i ed s a is ical aining and me hodological s anda diza ion among many in es iga o s in he ield [54].
Conclusion
This me a-analysis demons a es ha ci cula ing miRNAs exhibi p omising diagnos ic accu acy o dis inguishing among
a ious s ages o DR, suppo ing hei ole as p ac ical, non-in asi e bioma ke s. By s a i ying he analysis in o h ee clin-
ically ele an compa ison g oups (CTL s DR; T2DM s DR; and NPDR s PDR), we educed in e -s udy he e ogenei y
PLOS One | h ps://doi.o g/10.1371/jou nal.pone.0335434 No embe 21, 2025 18 / 20
and gene a ed mo e p ecise and clinically meaning ul es ima es o diagnos ic pe o mance. To u he enhance hei ans-
la ional po en ial, miRNA exp ession p o iling should be in eg a ed wi h es ablished clinical assessmen s and alida ed in
well-designed p ospec i e coho s. We call upon he scien i ic and clinical communi y o es ablish in e na ional consensus
on sample p ocessing, no maliza ion p o ocols, and diagnos ic h eshold de ini ion. Only h ough igo ous s anda diza ion
and p ospec i e alida ion can ci cula ing miRNAs be success ully inco po a ed in o ou ine diagnos ic wo k lows.
Suppo ing in o ma ion
S1 File. S1 Tex . Comple e sea ch s a egy. S1 Table. P isma DTA abs ac checklis . S2 Table. P isma DTA checklis .
S3 Table. Ex ac ion da a. S4 Table. Da ase used o STATA me a-analysis and me a- eg ession (CTL s DR). S5 Table.
Da ase used o STATA me a-analysis and me a- eg ession (T2DM s DR). S6 Table. Da ase used o STATA me a-
analysis and me a- eg ession (NPDR s PDR).
(ZIP)
Au ho con ibu ions
Concep ualiza ion: Mi iam Ma ínez-San os, Elías Ma ínez-López, Jo ge M. Ba cia.
Da a cu a ion: Ma ía Yba a, Ma ia E. Pi es, Chia a Ce esoni.
Fo mal analysis: Mi iam Ma ínez-San os, Elías Ma ínez-López.
Funding acquisi ion: Jo ge M. Ba cia, Ma ia Ol a, Ja ie Sancho-Pelluz.
In es iga ion: Mi iam Ma ínez-San os.
Me hodology: Mi iam Ma ínez-San os, Elías Ma ínez-López, Ma ia Ol a.
P ojec adminis a ion: Jo ge M. Ba cia, Ja ie Sancho-Pelluz.
Supe ision: Jo ge M. Ba cia, Ma ia Ol a, Ja ie Sancho-Pelluz.
Visualiza ion: Mi iam Ma ínez-San os, Ma ía Yba a.
W i ing – o iginal d a : Mi iam Ma ínez-San os, Ma ía Yba a, Ma ia Ol a.
Re e ences
1. Thomas RL, Halim S, Gu udas S, Si ap asad S, Owens DR. IDF Diabe es A las: A e iew o s udies u ilising e inal pho og aphy on he global
p e alence o diabe es ela ed e inopa hy be ween 2015 and 2018. Diabe es Res Clin P ac . 2019;157:107840. h ps://doi.o g/10.1016/j.dia-
b es.2019.107840 PMID: 31733978
2. Teo ZL, Tham YC, Yu M, Chee ML, Rim TH, Cheung N. Global P e alence o Diabe ic Re inopa hy and P ojec ion o Bu den h ough 2045: Sys em-
a ic Re iew and Me a-analysis. Oph halmology. 2021;128(11):1580–91.
3. Li H, Liu X, Zhong H, Fang J, Li X, Shi R, e al. Resea ch p og ess on he pa hogenesis o diabe ic e inopa hy. BMC Oph halmol. 2023;23(1):372.
h ps://doi.o g/10.1186/s12886-023-03118-6 PMID: 37697295
4. Cunha-Vaz J, Mendes L, Res e-Fe ei a D. Unde s anding nonp oli e a i e diabe ic e inopa hy p og ession using nonin asi e imaging. Eye (Lond).
2025;39(14):2627–34. h ps://doi.o g/10.1038/s41433-025-03901-3 PMID: 40646245
5. Yang Q-H, Zhang Y, Zhang X-M, Li X-R. P e alence o diabe ic e inopa hy, p oli e a i e diabe ic e inopa hy and non-p oli e a i e diabe ic e inop-
a hy in Asian T2DM pa ien s: a sys ema ic e iew and Me a-analysis. In J Oph halmol. 2019;12(2):302–11. h ps://doi.o g/10.18240/ijo.2019.02.19
PMID: 30809489
6. Ho ie S, Ohno-Ma sui K. P og ess o Imaging in Diabe ic Re inopa hy-F om he Pas o he P esen . Diagnos ics (Basel). 2022;12(7):1684. h ps://
doi.o g/10.3390/diagnos ics12071684 PMID: 35885588
7. Li X-Y, Wang S, Dong L, Zhang H. Compa ison o undus luo escein angiog aphy and undus pho og aphy g ading c i e ia o ea ly diabe ic e inop-
a hy. In J Oph halmol. 2022;15(2):261–7. h ps://doi.o g/10.18240/ijo.2022.02.11 PMID: 35186686
8. Panwa N, Huang P, Lee J, Keane PA, Chuan TS, Richha iya A, e al. Fundus Pho og aphy in he 21s Cen u y--A Re iew o Recen Technological
Ad ances and Thei Implica ions o Wo ldwide Heal hca e. Telemed J E Heal h. 2016;22(3):198–208. h ps://doi.o g/10.1089/ mj.2015.0068 PMID:
26308281
PLOS One | h ps://doi.o g/10.1371/jou nal.pone.0335434 No embe 21, 2025 19 / 20
9. Okonkwo ON, Hassan AO, Bogunjoko T, Akinye A, Akanbi T, Agweye C. Low a es o op ical cohe ence omog aphy u iliza ion in he diagnosis and
managemen o e ino ascula diseases in a lowe middle-income economy. Nige J Clin P ac . 2023;26(7):1011–6. h ps://doi.o g/10.4103/njcp.
njcp_911_22 PMID: 37635588
10. Chond ozoumakis G, Cha zimichail E, Hab a O, Vouno ypidis E, Papanas N, Ga ziou as Z, e al. Re inal Bioma ke s in Diabe ic Re inopa hy: F om
Ea ly De ec ion o Pe sonalized T ea men . J Clin Med. 2025;14(4):1343. h ps://doi.o g/10.3390/jcm14041343 PMID: 40004872
11. F udd K, Si ap asad S, Raman R, K ishnakuma S, Re a hy YR, Tu owski P. Diagnos ic ci cula ing bioma ke s o de ec ision- h ea ening diabe ic
e inopa hy: Po en ial sc eening ool o he u u e?. Ac a Oph halmol. 2022;100(3):e648–68. h ps://doi.o g/10.1111/aos.14954 PMID: 34269526
12. Smi -McB ide Z, Mo se LS. Mic oRNA and diabe ic e inopa hy-bioma ke s and no el he apeu ics. Ann T ansl Med. 2021;9(15):1280. h ps://doi.
o g/10.21037/a m-20-5189 PMID: 34532417
13. Simó-Se a O, Simó R, He nández C. Ci cula ing Bioma ke s o Diabe ic Re inopa hy: An O e iew Based on Physiopa hology. J Diabe es Res.
2016;2016:5263798. h ps://doi.o g/10.1155/2016/5263798 PMID: 27376090
14. Zhao L, Pan Q. Highly-Exp essed MiR-221-3p Dis inc ly Inc eases he Incidence o Diabe ic Re inopa hy in Pa ien s Wi h Type 2 Diabe es Melli us.
T ansl Vis Sci Technol. 2023;12(10):17. h ps://doi.o g/10.1167/ s .12.10.17 PMID: 37856104
15. Qing S, Yuan S, Yun C, Hui H, Mao P, Wen F, e al. Se um miRNA bioma ke s se e as a inge p in o p oli e a i e diabe ic e inopa hy. Cell
Physiol Biochem. 2014;34(5):1733–40. h ps://doi.o g/10.1159/000366374 PMID: 25427542
16. Saleh AA, El-He nawy SM, Kasemy ZA, Alhagaa AA, Nooh MZ, A a a ES. Mi-RNA-93 and Mi-RNA-152 in he Diagnosis o Type 2 Diabe es and
Diabe ic Re inopa hy. B J Biomed Sci. 2022;79:10192. h ps://doi.o g/10.3389/bjbs.2021.10192 PMID: 35996507
17. Hui Y, Yin Y, Tian DH. Combina ion o se um HMGB-1 and se um miR-126 achie e high p edic i e powe o de ec p oli e a i e diabe ic e inopa hy:
a Chinese popula ion-based s udy. In e na ional Jou nal o Clinical and Expe imen al Medicine. 2019;12(5):5909–14.
18. Ma Y, Zhou LX, Liu Y, Wu W. P edic ion o e inopa hy in ype 2 diabe es melli us by combined de ec ion o mic oRNA93 and mic oRNA21. In e na-
ional Eye Science. 2019;19(9):1550–3.
19. Ma L, Wen Y, Li Z, Wu N, Wang Q. Ci cula ing Mic oRNAs as Po en ial Diagnos ic Bioma ke s o Diabe ic Re inopa hy: A Me a-Analysis. F on
Endoc inol (Lausanne). 2022;13:929924. h ps://doi.o g/10.3389/ endo.2022.929924 PMID: 35898469
20. Page MJ, McKenzie JE, Bossuy PM, Bou on I, Ho mann TC, Mul ow CD, e al. The PRISMA 2020 s a emen : an upda ed guideline o epo ing
sys ema ic e iews. BMJ. 2021;372:n71. h ps://doi.o g/10.1136/bmj.n71 PMID: 33782057
21. Kwon Y, Lemieux M, McTa ish J, Wa hen N. Iden i ying and emo ing duplica e eco ds om sys ema ic e iew sea ches. J Med Lib Assoc.
2015;103(4):184–8. h ps://doi.o g/10.3163/1536-5050.103.4.004 PMID: 26512216
22. Whi ing PF, Ru jes AWS, Wes wood ME, Malle S, Deeks JJ, Rei sma JB, e al. QUADAS-2: a e ised ool o he quali y assessmen o diagnos ic
accu acy s udies. Ann In e n Med. 2011;155(8):529–36. h ps://doi.o g/10.7326/0003-4819-155-8-201110180-00009 PMID: 22007046
23. B ozek JL, Akl EA, Jaeschke R, Lang DM, Bossuy P, Glasziou P, e al. G ading quali y o e idence and s eng h o ecommenda ions in
clinical p ac ice guidelines: Pa 2 o 3. The GRADE app oach o g ading quali y o e idence abou diagnos ic es s and s a egies. Alle gy.
2009;64(8):1109–16. h ps://doi.o g/10.1111/j.1398-9995.2009.02083.x PMID: 19489757
24. Qin L-L, An M-X, Liu Y-L, Xu H-C, Lu Z-Q. Mic oRNA-126: a p omising no el bioma ke in pe iphe al blood o diabe ic e inopa hy. In J Oph hal-
mol. 2017;10(4):530–4. h ps://doi.o g/10.18240/ijo.2017.04.05 PMID: 28503423
25. Wan S, Wang J, Wang J, Wu J, Song J, Zhang C-Y, e al. Inc eased se um miR-7 is a p omising bioma ke o ype 2 diabe es melli us and i s
mic o ascula complica ions. Diabe es Res Clin P ac . 2017;130:171–9. h ps://doi.o g/10.1016/j.diab es.2017.06.005 PMID: 28646700
26. Shake OG, Abdelaleem OO, Mahmoud RH, Abdelgha a NK, Ahmed TI, Said OM, e al. Diagnos ic and p ognos ic ole o se um miR-20b, miR-
17-3p, HOTAIR, and MALAT1 in diabe ic e inopa hy. IUBMB Li e. 2019;71(3):310–20.
27. Li ZM, Dong Y, He C, Pan XC, Liu DY, Yang JL, e al. RNA-Seq Re ealed No el Non-p oli e a i e Re inopa hy Speci ic Ci cula ing MiRNAs in
T2DM Pa ien s. F on ie s in Gene ics. 2019;10.
28. Yin C, Lin X, Sun Y, Ji X. Dys egula ion o miR-210 is in ol ed in he de elopmen o diabe ic e inopa hy and se es a egula o y ole in e inal
ascula endo helial cell p oli e a ion. Eu J Med Res. 2020;25(1):20. h ps://doi.o g/10.1186/s40001-020-00416-3 PMID: 32498701
29. San o i o D, To o L, De Na dis V, Ma can onio P, D’Aloisio R, Mas opasqua A, e al. Plasma mic oRNA signa u e associa ed wi h e inopa hy in
pa ien s wi h ype 2 diabe es. Sci Rep. 2021;11(1):4136. h ps://doi.o g/10.1038/s41598-021-83047-w PMID: 33602976
30. Wang Z, Zhang X, Wang Y, Xiao D. Dys egula ion o miR-374a is in ol ed in he p og ession o diabe ic e inopa hy and egula es he p oli e a ion
and mig a ion o e inal mic o ascula endo helial cells. Clin Exp Op om. 2022;105(3):287–92. h ps://doi.o g/10.1080/08164622.2021.1913043
PMID: 33941051
31. Salem TI, Eldin NB, Alhusseini NF, Abdullah OA, Ahmed NE. Exp ession p o ile o mic oRNAs may be p omising in diagnosis o p oli e a i e dia-
be ic e inopa hy: an Egyp ian s udy. In J Diabe es De C ies. 2022;43(1):36–44. h ps://doi.o g/10.1007/s13410-022-01044-9
32. Su asmia i NMA, Su ya hi NMA, Andayani A. Se um mic oRNA-126 exp ession as a bioma ke o diabe ic e inopa hy. Uni Med. 2023;42(2):121–
7. h ps://doi.o g/10.18051/uni med.2023. 42.121-127
33. Liu H-N, Li X, Wu N, Tong M-M, Chen S, Zhu S-S, e al. Se um mic oRNA-221 as a bioma ke o diabe ic e inopa hy in pa ien s associa ed wi h
ype 2 diabe es. In J Oph halmol. 2018;11(12):1889–94. h ps://doi.o g/10.18240/ijo.2018.12.02 PMID: 30588418
PLOS One | h ps://doi.o g/10.1371/jou nal.pone.0335434 No embe 21, 2025 20 / 20
34. Jiang Q, Lyu XM, Yuan Y, Wang L. Plasma miR-21 exp ession: an indica o o he se e i y o Type 2 diabe es wi h diabe ic e inopa hy. Biosci Rep.
2017;37(2).
35. Hashemi H, Rez an F, Pakzad R, Ansa ipou A, Heyda ian S, Yek a A, e al. Global and Regional P e alence o Diabe ic Re inopa hy; A Comp e-
hensi e Sys ema ic Re iew and Me a-analysis. Semin Oph halmol. 2022;37(3):291–306. h ps://doi.o g/10.1080/08820538.2021.1962920 PMID:
34402390
36. Zhou H, Peng C, Huang D-S, Liu L, Guan P. mic oRNA Exp ession P o iling Based on Mic oa ay App oach in Human Diabe ic Re inopa hy: A
Sys ema ic Re iew and Me a-Analysis. DNA Cell Biol. 2020;39(3):441–50. h ps://doi.o g/10.1089/dna.2019.4942 PMID: 32101049
37. P amanik S, Saha C, Chowdhu y S, Bose C, Bha acha yya NP, Mondal LK. Dec eased Le els o miR-126 and miR-132 in Plasma and Vi eous
Humo o Non-P oli e a i e Diabe ic Re inopa hy Among Subjec s wi h Type-2 Diabe es Melli us. Diabe es Me ab Synd Obes. 2022;15:345–58.
h ps://doi.o g/10.2147/DMSO.S346097 PMID: 35153496
38. Roy D, Modi A, Khokha M, Sankanagouda S, Yada D, Sha ma S, e al. Mic oRNA 21 Eme ging Role in Diabe ic Complica ions: A C i ical
Upda e. Cu Diabe es Re . 2021;17(2):122–35. h ps://doi.o g/10.2174/1573399816666200503035035 PMID: 32359340
39. Zou H-L, Wang Y, Gang Q, Zhang Y, Sun Y. Plasma le el o miR-93 is associa ed wi h highe isk o de elop ype 2 diabe ic e inopa hy. G ae es
A ch Clin Exp Oph halmol. 2017;255(6):1159–66. h ps://doi.o g/10.1007/s00417-017-3638-5 PMID: 28382439
40. Blondal T, Jensby Nielsen S, Bake A, And easen D, Mou i zen P, W ang Teilum M, e al. Assessing sample and miRNA p o ile quali y in se um and
plasma o o he bio luids. Me hods. 2013;59(1):S1-6. h ps://doi.o g/10.1016/j.yme h.2012.09.015 PMID: 23036329
41. Fa aldi M, Sansoni V, Pe ego S, Goma asca M, Ko as J, Ziemann E, e al. S udy o he p eanaly ical a iables a ec ing he measu emen
o clinically ele an ee-ci cula ing mic oRNAs: ocus on sample ma ix, pla ele deple ion, and s o age condi ions. Biochem Med (Zag eb).
2020;30(1):010703. h ps://doi.o g/10.11613/BM.2020.010703 PMID: 31839723
42. Xiang M, Zeng Y, Yang R, Xu H, Chen Z, Zhong J, e al. U6 is no a sui able endogenous con ol o he quan i ica ion o ci cula ing mic oRNAs.
Biochem Biophys Res Commun. 2014;454(1):210–4. h ps://doi.o g/10.1016/j.bb c.2014.10.064 PMID: 25450382
43. Klo en V, Neumann MHD, Di Pasquale F, Sp enge -Haussels M, Sha e JM, Schlumpbe ge M, e al. Mul icen e E alua ion o Ci cula ing Plasma
Mic oRNA Ex ac ion Technologies o he De elopmen o Clinically Feasible Re e se T ansc ip ion Quan i a i e PCR and Nex -Gene a ion
Sequencing Analy ical Wo k Flows. Clin Chem. 2019;65(9):1132–40. h ps://doi.o g/10.1373/clinchem.2019.303271 PMID: 31235535
44. So niko a EA, Kisele a AV, Meshko AN. E ec o plasma and se um s o age condi ions on ci cula ing mic oRNA le els. Ca dio asc The P e .
2024;23(11):4180. h ps://doi.o g/10.15829/1728-8800-2024-4180
45. Lee I, Bax e D, Lee MY, Sche le K, Wang K. The Impo ance o S anda diza ion on Analyzing Ci cula ing RNA. Mol Diagn The . 2017;21(3):259–
68. h ps://doi.o g/10.1007/s40291-016-0251-y PMID: 28039578
46. Yoon H, Belmon e KC, Kas en T, Ba eman R, Kim J. In a- and In e -indi idual Va iabili y o mic oRNA Le els in Human Ce eb ospinal Fluid: C i i-
cal Implica ions o Bioma ke Disco e y. Sci Rep. 2017;7(1):12720. h ps://doi.o g/10.1038/s41598-017-13031-w PMID: 28983117
47. Kozoma a A, Bi gaoanu M, G i i hs-Jones S. miRBase: om mic oRNA sequences o unc ion. Nucleic Acids Res. 2019;47(D1):D155–62. h ps://
doi.o g/10.1093/na /gky1141 PMID: 30423142
48. Suzuki K, Yamaguchi T, Kohda M, Tanaka M, Takemu a H, Waki a M, e al. Es ablishmen o p eanaly ical condi ions o mic oRNA p o ile analysis
o clinical plasma samples. PLoS One. 2022;17(12):e0278927. h ps://doi.o g/10.1371/jou nal.pone.0278927 PMID: 36516194
49. Ki schne MB, Kao SC, Edelman JJ, A ms ong NJ, Vallely MP, an Zandwijk N, e al. Haemolysis du ing sample p epa a ion al e s mic oRNA
con en o plasma. PLoS One. 2011;6(9):e24145. h ps://doi.o g/10.1371/jou nal.pone.0024145 PMID: 21909417
50. Zendjabil M. P eanaly ical, analy ical and pos analy ical conside a ions in ci cula ing mic oRNAs measu emen . Biochem Med (Zag eb).
2024;34(2):020501. h ps://doi.o g/10.11613/BM.2024.020501 PMID: 38882585
51. Ru jes AWS, Rei sma JB, Vandenb oucke JP, Glas AS, Bossuy PMM. Case-con ol and wo-ga e designs in diagnos ic accu acy s udies. Clin
Chem. 2005;51(8):1335–41. h ps://doi.o g/10.1373/clinchem.2005.048595 PMID: 15961549
52. Yang B, Olsen M, Vali Y, Langendam MW, Takwoingi Y, Hyde CJ, e al. S udy designs o compa a i e diagnos ic es accu acy: A me hodological
e iew and classi ica ion scheme. J Clin Epidemiol. 2021;138:128–38. h ps://doi.o g/10.1016/j.jclinepi.2021.04.013 PMID: 33915262
53. V bin CM. Recommenda ions o epo ing measu es o diagnos ic accu acy. Cy opa hology. 2023;34(3):185–90. h ps://doi.o g/10.1111/cy .13208
PMID: 36648307
54. Hassanzad M, Hajian-Tilaki K. Me hods o de e mining op imal cu -poin o diagnos ic bioma ke s wi h applica ion o clinical da a in ROC analysis:
an upda e e iew. BMC Med Res Me hodol. 2024;24(1):84. h ps://doi.o g/10.1186/s12874-024-02198-2 PMID: 38589814
