RESEARCH ARTICLE Open Access
CRABP1, C1QL1 and LCN2 a e bioma ke s o
di e en ia ed hy oid ca cinoma, and
p edic ex a hy oidal ex ension
Rica do Celes ino
1,2,3,4
, To inn Nome
3,5
, Ana Pes ana
1,2,6
, And eas M. Ho
3,5
, A. Ped o Gonçal es
1,6,7
,
Luísa Pe ei a
1,2
, B uno Ca adas
1,2
, Ca a ina Eloy
1,2
, T ine Bjø o
8,9
, Manuel Sob inho-Simões
1,2,10,11
,
Rol I. Sko heim
3,5*
and Paula Soa es
1,2,10*
Abs ac
Backg ound: The p ognos ic a iabili y o hy oid ca cinomas has led o he sea ch o accu a e bioma ke s a he
molecula le el. Follicula hy oid ca cinoma (FTC) is a ypical example o di e en ia ed hy oid ca cinomas (DTC) in
which challenges a e aced in he di e en ial diagnosis.
Me hods: We used high- h oughpu pai ed-end RNA sequencing echnology o s udy ou cases o FTC wi h di e en
deg ee o capsula in asion: wo minimally in asi e (mFTC) and wo widely in asi e FTC (wFTC). We sea ched by genes
di e en ially exp essed be ween mFTC and wFTC, in an a emp o ind bioma ke s o hy oid cance diagnosis and/o
p og ession. Selec ed bioma ke s we e alida ed by eal- ime quan i a i e PCR in 137 ozen hy oid samples and in an
independen da ase (TCGA), e alua ing he diagnos ic and he p ognos ic pe o mance o he candida e bioma ke s.
Resul s: We iden i ied 17 genes signi ican ly di e en ially exp essed be ween mFTC and wFTC. C1QL1,LCN2,CRABP1
and CILP we e di e en ially exp essed in DTC in compa ison wi h no mal hy oid issues. LCN2 and CRABP1 we e also
di e en ially exp essed in DTC when compa ed wi h ollicula hy oid adenoma. Addi ionally, o e exp ession o LCN2
and C1QL1 we e ound o be independen p edic o s o ex a hy oidal ex ension in DTC.
Conclusions: We conclude ha he unde exp ession o CRABP1 and he o e exp ession o LCN2 may be use ul
diagnos ic bioma ke s in hy oid umou s wi h ques ionable maligni y, and he o e exp ession o LCN2 and C1QL1 may
be use ul o p ognos ic pu poses.
Keywo ds: CRABP1,CILP,C1QL1,LCN2, Cance , Thy oid, Bioma ke , TCGA, RNA sequencing, Gene exp ession
Backg ound
Thy oid cance is he mos equen ype o endoc ine
cance wi h an incidence o 12 cases pe 100,000 indi-
iduals [1, 2]. Mo e han 95% o hy oid cance cases
o igina e om ollicula epi helial cells [3]. Papilla y hy-
oid ca cinoma (PTC) and ollicula hy oid ca cinoma
(FTC) a e he mos common his o ypes. Despi e he
o e all good p ognosis o hese wo main his o ypes o
di e en ia ed hy oid ca cinoma (DTC) [1], some cases
p og ess, de elop dis an me as ases and acqui e an un-
p edic able esponse o ea men .
The inc easing incidence o hy oid cance has led o he
sea ch o good bioma ke s ha can help in he diagnosis
o malignancy and/o p edic he clinical beha iou o he
umou s. Un il now, clinical and his opa hological p og-
nos ic ac o s emain he only obus elemen s o be used
o diagnosis and p ognosis o pa ien s wi h hy oid u-
mou s [4], al hough new ma ke s a e e ealing some diag-
nos ic o p ognos ic alue pe se. As an example, BRAF
mu a ions ha e been shown o be use ul o p edic ing e-
cu ence and/o disease pe sis ence [5], bu mos ly when
associa ed wi h o he clinicopa hological ea u es. Recen ly,
TERT p omo e mu a ions e ealed an independen
* Co espondence: [email p o ec ed];[email p o ec ed]
3
Depa men o Molecula Oncology, Ins i u e o Cance Resea ch,
No wegian Radium Hospi al, Oslo Uni e si y Hospi al, P.O.Box 4953 Nydalen,
0424 Oslo, No way
1
i3S - Ins i u o de In es igação e Ino ação em Saúde, Uni e sidade do Po o,
Rua Al edo Allen, 208, 4200-135 Po o, Po ugal
Full lis o au ho in o ma ion is a ailable a he end o he a icle
© The Au ho (s). 2018 Open Access This a icle is dis ibu ed unde he e ms o he C ea i e Commons A ibu ion 4.0
In e na ional License (h p://c ea i ecommons.o g/licenses/by/4.0/), which pe mi s un es ic ed use, dis ibu ion, and
ep oduc ion in any medium, p o ided you gi e app op ia e c edi o he o iginal au ho (s) and he sou ce, p o ide a link o
he C ea i e Commons license, and indica e i changes we e made. The C ea i e Commons Public Domain Dedica ion wai e
(h p://c ea i ecommons.o g/publicdomain/ze o/1.0/) applies o he da a made a ailable in his a icle, unless o he wise s a ed.
Celes ino e al. BMC Cance (2018) 18:68
DOI 10.1186/s12885-017-3948-3
p ognos ic alue ega ding dis an me as asis and su i al
o pa ien s wi h hy oid cance [5].
The classi ica ion o benign and malignan hy oid u-
mou s a he his ological le el s ill has limi a ions. Many
ollicula pa e ned umou s a e ypical examples o his
di icul y. The classical his ological c i e ion o dis in-
guish FTC om ollicula hy oid adenoma (FTA) is he
p esence o any image o capsula o ascula in asion
[6]. This ci cums ance limi s he diagnos ic accu acy o
ine needle aspi a ion biopsy (FNAB) in p e-su gical
g ounds. FTC is subclassi ied in o minimally in asi e
FTC (mFTC) and widely in asi e FTC (wFTC) [3], wi h
he la e ha ing a wo se p ognosis han mFTC [3, 7, 8].
The molecula mechanisms ha o ches a e he in a-
si eness o FTC cance cells a e poo ly unde s ood.
In o de o iden i y molecula al e a ions associa ed o
hy oid cance in asion, we used FTC as a model o an
encapsula ed umou , s udying umou s wi h di e en
deg ees o in asion: wo mFTC and wo wFTC, using
high- h oughpu pai ed-end RNA sequencing (RNA-seq)
echnology. The bioma ke s p oposed he e we e ali-
da ed in a se ies o DTC, in hy oid cance cell lines and
in he gene exp ession da a om DTC a ailable in The
Cance Genome A las (TCGA) [9].
Me hods
Thy oid cance samples and cell lines
Two wFTC (cases 1 and 2) and wo mFTC (cases 3 and
4) we e collec ed om umou issue biobank (Po o).
These ou umou s om pa ien s wi h 55–82 yea s o
age we e subjec ed o high- h oughpu pai ed-end RNA-
seq analyses. B ie ly, he pa ien /case 1 had a umou
measu ing 2 cm in diame e and p esen ing a p edomin-
an ollicula g ow h pa e n and oncocy ic ea u es; pa-
ien /case 2 had a umou measu ing 5 cm in diame e
and p esen ing a p edominan solid/ abecula g ow h
pa e n; pa ien /case 3 had a umou measu ing 3.7 cm
in diame e and p esen ing a ollicula g ow h pa e n;
and pa ien /case 4 had a umou measu ing 4 cm in
diame e and p esen ing a ollicula g ow h pa e n.
Fo compa a i e and alida ion pu poses, 137 ozen
hy oid samples we e collec ed om umou issue bio-
bank (Po o): 98 DTC [15 FTC (including he ou used
in RNAseq), 23 ollicula a ian o PTC (FVPTC) and
60 PTC], 20 FTA, and 19 no mal samples o adjacen /
con ala e al hy oid issues om pa ien s wi h DTC
om his se ies. The his ology o all DTC was e ised by
wo pa hologis s (CE and MSS) and he inal classi ica-
ion was made acco ding o he WHO c i e ia [3]. The
clinicopa hological ea u es and gene ic al e a ions o he
DTC a e p esen ed in Table 1, and a b ie ly desc ip ion
o he se ies is a ailable in he Addi ional ile 1. Fu he -
mo e, en hy oid cance cell lines we e also used in his
s udy including one cell line de i ed om one FTC wi h
oncocy ic pa e n (XTC1), h ee de i ed om PTC
(BCPAP, K1 and TPC1), and six de i ed om undi e -
en ia ed hy oid ca cinoma (C643, HTH74, KAT4, T238,
T241, 8505C).
High- h oughpu pai ed-end RNA-seq
Lib a y cons uc ion was pe o med using he T uSeq
RNA Sample P ep Ki 2 acco ding o p o ocol (Illumina
Inc., San Diego, CA, USA), including poly-A mRNA isola-
ion, agmen a ion, and gel-based size selec ion. Shea ing
o abou 250 bp agmen s was achie ed using he Co a is
S2 ocused-ul asonica o (Co a is Inc., Wobu n, MA,
USA). Sequencing was pe o med acco ding o he pai ed-
end RNA-seq p o ocols om Illumina o Solexa sequen-
cing on a Genome Analyze IIx wi h pai ed-end module
(Illumina Inc). Se en y-six bps we e sequenced, om each
side o a agmen o abou 250 bp long.
A pa icula a en ion was paid o genes di e en ially
exp essed be ween mFTC and wFTC. Reads ma ked by
he Illumina pipeline (Bus a d.py, OLB 1.6.0 and 1.8.0)
as passed il e ing we e used in he analysis.
Gene exp ession le els in FTCs we e compu ed by
using Cu links 1.1.0 [10] wi h he Illumina iGenomes
Ensembl GRCh37 da a se (2011–06-20) as e e ence, on
eads aligned wi h TopHa 1.3.3 [11]. Gene exp ession
le els we e compa ed be ween he wo wFTC (case 1
and 2) and wo mFTC (case 3 and 4). Resul s we e ep-
esen ed as old change o gene exp ession in wFTC
compa ing wi h mFTC. Values o old change in gene
exp ession we e a loga i hmic scale (log2). Genes and
ansc ip s we e conside ed di e en ially exp essed be-
ween mFTC and wFTC whene e Q alue <0.05.
Isola ion o nucleic acids, e e se- ansc ip ion PCR and
cDNA sange sequencing
F ozen issue samples a ailable om hy oid umou s (n
= 118) and adjacen /colla e al no mal hy oid issues (n
= 19) we e c ushed and homogenized. To al RNA om
issues and hy oid cance de i ed cell lines was ex-
ac ed using TRIzol® Reagen (Ambion®, Li e Technolo-
gies™, CA, USA), acco ding o he manu ac u e ’s
p o ocol. Genomic DNA was ex ac ed using he Gen-
omic DNA Pu i ica ion Ki (Ci omed, Lisbon, Po ugal)
acco ding o he manu ac u e ’s p o ocol.
cDNA was syn hesized om 1 μg o RNA a 42 °C o
60 min, using oligo (dT) p ime s and M-MuLV e e se
ansc ip ase (Fe men as, The mo Scien i ic, S . Leon-Ro ,
Ge many). The e e se- ansc ip ion PCR (RT-PCR) p od-
uc s we e analysed by cDNA Sange sequencing using he
Big Dye e mina o e sion 3.1 cycle (Applied Biosys ems,
Fos e Ci y, CA, USA). The samples we e analysed in an
au oma ed sequencing machine (ABI P ism 3100 Gene ic
Analyse , Applied Biosys ems, Fos e Ci y, CA, USA).
Celes ino e al. BMC Cance (2018) 18:68 Page 2 o 16
Sc eening o PAX8/PPARG,RET/PTC1 and RET/PTC3
ea angemen s, BRAF and NRAS mu a ions, and TERT
p omo e mu a ions in hy oid cance se ies
The 98 DTC we e u he cha ac e ized o selec ed mo-
lecula al e a ions known o be common in hy oid
umou s. PCR ollowed by Sange sequencing was used
on genomic DNA o de ec ing mu a ions in he mos
equen ho spo egions o BRAF (exon 15), NRAS
(exon 2), and p omo e egion o TERT, acco ding o
p e iously desc ibed p ocedu es [12, 13]. In ou p e ious
Table 1 Clinicopa hological and molecula da a o pa ien s wi h di e en ia ed hy oid cance (DTC) included in he s udy
To al FTC FVPTC PTC
To al (%) 98 (100) 15 (15.3) 23 (23.5) 60 (61.2)
Age (n) 98 15 23 60
Mean (yea s) 43 ± 1.6 53 ± 3.9 43 ± 3.1 40 ± 2.1
Tumou size (n) 93 15 22 56
Mean (cm) 2.8 ± 0.18 3.9 ± 0.46 2.4 ± 0.30 2.7 ± 0.24
Gende (n) 98 15 23 60
Female (%) 84 (85.7) 12 (80) 19 (82.6) 53 (88.3)
Male (%) 14 (14.3) 3 (20) 4 (17.4) 7 (11.7)
Capsule (n) 88 15 21 52
Posi i e (%) 45 (51.1) 15 (100) 9 (42.9) 21 (40.4)
Capsula in asion (n) 42 15 8 19
Posi i e (%) 27 (64.3) 15 (100) 2 (25.0) 10 (52.6)
Vascula in asion (n) 89 15 21 53
Posi i e (%) 48 (53.9) 11 (73) 5 (23.8) 32 (60.4)
Lymph node me as asis (n) 93 15 21 57
Posi i e (%) 24 (25.8) –4 (19.0) 20 (35.1)
Ex a hy oidal ex ension (n) 88 15 21 52
Posi i e (%) 27 (30.7) 1 (6.67) 2 (9.52) 24 (46.2)
Dis an me as asis (n) 93 15 21 57
Posi i e (%) 5 (5.38) 1 (6.67) –4 (7.02)
Lymphocy ic hy oidi is (n) 91 15 22 54
Posi i e (%) 40 (44.0) 4 (26.7) 8 (36.4) 28 (51.9)
Oncocy ic (n) 89 15 21 53
Posi i e (%) 17 (19.1) 6 (40.0) –11 (20.8)
PAX8-PPARG ea angemen s (n) 98 15 23 60
Posi i e (%) 2 (2.04) 1 (6.67) 1 (4.35) –
RET/PTC ea angemen s (n) 98 15 23 60
Posi i e (%) 17 (17.3) 1 (6.67) 3 (13.0) 13 (21.7)
RET/PTC1 ea angemen (n) 98 15 23 60
Posi i e (%) 12 (12.2) 1 (6.67) 1 (4.35) 10 (16.7)
RET/PTC3 ea angemen (n) 98 15 23 60
Posi i e (%) 2 (2.04) –1 (4.35) 1 (1.67)
BRAF mu a ion (n) 98 15 23 60
Posi i e (%) 21 (21.4) –2 (8.70) 19 (31.7)
NRAS mu a ion (n) 98 15 23 60
Posi i e (%) 16 (16.3) 5 (33.3) 5 (21.7) 6 (10.0)
TERT p omo e mu a ion (n) 98 15 23 60
Posi i e (%) 3 (3.06) 2 (13.3) 1 (4.35) –
n, numbe o cases wi h a ailable da a
FTC, ollicula hy oid ca cinoma; FVPTC, ollicula a ian o papilla y hy oid ca cinoma; PTC, papilla y hy oid ca cinoma
Celes ino e al. BMC Cance (2018) 18:68 Page 3 o 16
wo ks we e i ied ha HRAS and KRAS mu a ions we e
e y a e (o absen ) e en s. Fo ha eason, we only
sc eened mu a ions o NRAS gene.
The p esence o PAX8-PPARG and RET/PTC ea ange-
men s we e de e mined by RT-PCR ollowed by Sange
sequencing o PCR p oduc s. Some posi i e cases we e
con i med by luo escence in si u hyb idisa ion (FISH),
ollowing he p ocedu es desc ibed p e iously [12].
Real- ime quan i a i e e e se ansc ip ion PCR
Real- ime quan i a i e PCR (qPCR) o C1QL1,LCN2,
IL22RA1,MAMDC2,CILP,ASXL3,CRABP1, and
SCUBE3 genes was pe o med in he wo wFTC (cases 1
and 2) and wo mFTC (cases 3 and 4) used in pai ed-
end RNA-seq, and hei co esponding no mal hy oid
issue. qPCR o C1QL1,LCN2,CILP,andCRABP1 we e
also pe o med in he emaining se ies o DTC, in FTA
and in no mal hy oid issues, as well as in hy oid can-
ce cell lines.
cDNA was ampli ied o he C1QL1,LCN2,IL22RA1,
MAMDC2,CILP,ASXL3,CRABP1, and SCUBE3 by
qPCR using SYBR® G een PCR Mas e Mix (Applied
Biosys ems, Fos e Ci y, CA, USA). A qPCR assay a ge -
ing Hypoxan hine phospho ibosyl ans e ase 1 (HPRT)
was used as he housekeeping con ol. P ime sequences
a e a ailable on eques . Exp ession le els we e ob ained
as he a e age cycle h eshold (CT) o a leas wo ep-
licas. A CT = 35 was assigned o a ge genes ha we e
no exp essed in samples wi h posi i e exp ession o he
housekeeping con ol gene (HPRT).
The ela i e quan i ica ion o a ge genes was de e -
mined using he compa a i e CT me hod (2
-ΔΔCT
)which
was p e iously alida ed by Li ak’s Linea Reg ession
Me hod (Sequence De ec o Use Bulle in 2; Applied Bio-
sys ems) [14]. This p o ided he old changes (2
-ΔΔCT
)in
gene exp ession no malized o an in e nal con ol gene
(HPRT), and ela i e o one pool o no mal hy oid issues
(calib a o ). Exp ession alues (2
-ΔΔCT
)we eno malized
a loga i hmic scale (log2). Fo C1QL1 and LCN2 genes,
old change [log2 (2
-ΔΔCT
)] > 1 was conside ed as gain o
exp ession, and old change [log2 (2
-ΔΔCT
)] ≤1 was es ab-
lished as no mal gene exp ession. Fo CRABP1 and CILP
genes, old change [log2 (2
-ΔΔCT
)] < −1 was conside ed as
loss o exp ession, and old change [log2 (2
-ΔΔCT
)] ≥−1
was es ablished as no mal gene exp ession. The diagnos ic
pe o mance (sensi i i y and speci ici y) o hose cu -o
alues a e p esen ed in Table 2.
Da a om he cance genomic a las (TCGA)
RNASeq ( e sion 2) in o ma ion we e ex ac ed om
TCGA [15] o 393 DTC and 59 hy oid no mal samples,
o C1QL1,LCN2,CRABP1 and CILP genes. Gene ex-
p ession alues o each sample a e no malized ead
coun s, es ima ed by using RSEM so wa e [16].
S a is ical analysis
S a is ical analysis was conduc ed wi h SPSS e sion 22.0
(SPSS Inc., Chicago, IL, USA). The esul s a e exp essed
as pe cen age o mean ± SE. S a is ical analysis was pe -
o med bo h on he whole DTC se ies and on he di e -
en subg oups: FTC, FVPTC and PTC. Recei e
Ope a ing Cha ac e is ics (ROC) cu es o indi idual
bioma ke s we e gene a ed using log2 (2
-ΔΔCT
) gene ex-
p ession alues and hy oid issue ype (DTC agains
no mal o FTA) as inpu . Fo e alua ion o he com-
bined bioma ke panel, he sum o log2 (2
-ΔΔCT
) exp es-
sion alues om genes wi h gain (C1QL1 and LCN2)
and loss (CRABP1 and CILP) in DTC we e used. Spea -
man’s ho es (non-pa ame ic) was used o e alua ing
he co ela ion o he exp essions be ween di e en
genes. Fishe ’s exac es , - es (unpai ed, wo- ailed),
Mann–Whi ney U es , and ANOVA we e used when
app op ia e. The p edic i e alue o C1QL1,LCN2,
CRABP1 and CILP exp ession as a p ognos ic ac o in
hy oid cance and hei co ela ion wi h clinicopa ho-
logical ac o s –age, umou size, and ex a hy oidal ex-
ension we e assessed using uni a ia e and mul i a ia e
logis ic eg ession models. Tes esul s wi h P- alues
<0.05 we e conside ed s a is ically signi ican .
Resul s
Di e en ial gene exp ession in mFTC and wFTC
Di e en ial exp ession analysis o he pai ed-end RNA-
seq da a iden i ied 17 genes wi h signi ican ly di e -
en ial exp ession be ween mFTC and wFTC (Fig. 1,
Addi ional ile 2: Table S1 and Addi ional ile 3: Table S2).
Inc eased gene exp ession o KLK1,NEFL,KIAA1239,
SLC6A17,NXPH4,LCN2,C1QL1 and TRIB3, and de-
c eased gene exp ession o ST6GAL1,SCUBE3,IL22RA1,
MAMDC2,CILP,CYSLTR2,ASXL3,CRABP1 and
LINC00887 we e obse ed in wFTC in compa ison wi h
mFTC. C1QL1,LCN2,IL22RA1,MAMDC2,CILP,ASXL3,
CRABP1,andSCUBE3 genes we e selec ed o alida ion
h ough a cus omized il e ing based on gene exp ession
in hy oid umou s and no mal hy oid issues da ase s
[17–19], assessed om Oncomine™and in he li e a u e
a ailable. Valida ion o exp ession le els was done in he
ou FTCs om pai ed-end RNA-seq and in hei ma ched
no mal issue by qPCR. Gain o C1QL1 and LCN2 exp es-
sion, and loss o IL22RA1,MAMDC2,CILP,ASXL3,
CRABP1,andSCUBE3 exp ession we e con i med in
wFTC compa ed wi h mFTC and be ween he ou FTC
and hei espec i e co esponding no mal hy oid issue
(Addi ional ile 4: igu e S1). Based on he accu acy o
gene exp ession le els ob ained om pai ed-end RNA-seq
and qPCR in he FTC and in he gene exp ession le els in
hei no mal hy oid issues by qPCR, C1QL1,LCN2,
CRABP1 and CILP genes we e selec ed o be es ed as bio-
ma ke s in a well-cha ac e ized se ies o DTC (Table 1).
Celes ino e al. BMC Cance (2018) 18:68 Page 4 o 16
Addi ionally, we iden i ied and expe imen ally e i ied
a se o 21 usion ansc ip s exp essed by DTC
(Addi ional ile 5: igu e S2; Addi ional ile 6: Table S3;
Addi ional ile 7: Table S4). Howe e , he usion an-
sc ip s a e no likely o be bioma ke s because hey we e
also de ec ed in no mal hy oid issues. Fu he de ails
can be ound in he Addi ional ile 8.
C1QL1,LCN2,CRABP1, and CILP as bioma ke s in DTC
A se ies o 98 DTC (15 FTC, 23 FVPTC and 60 PTC)
and a pool o 19 no mal hy oid issues and 20 FTA
we e used o e alua e he pe o mance o C1QL1,LCN2,
CRABP1, and CILP as bioma ke s o DTC. Measu e-
men s o he exp ession le els o hose genes we e done
by qPCR. C1QL1,LCN2,CRABP1 and CILP we e di e -
en ially exp essed in DTC in compa ison wi h no mal
hy oid issues (Addi ional ile 9: igu e S3), wi h s a is-
ical signi icance (P= 0.002, P= 0.005, P< 0.001 and P=
0.018, espec i ely). No ably, such signi ican di e ences
we e also obse ed o LCN2 and CRABP1 exp ession
when compa ing DTC agains FTA (P< 0.001 and P=
0.002, espec i ely), and FTA agains no mal hy oid is-
sues (P= 0.013 and P= 0.022, espec i ely).
In o de o es he a o emen ioned bioma ke s as pu-
a i e diagnos ic ools o hy oid cance , ROC cu es
we e pe o med (Fig. 2, Addi ional ile 10: Table S5). In
he ROC analysis, compa ing DTC e sus no mal hy-
oid (Fig. 2a), he a eas unde o he ROC cu e (AUC)
o C1QL1,LCN2,CRABP1 and CILP we e 0.799 (P=
2.1E-3), 0.783 (P= 3.6E-3), 0.902 (P= 3.4E-5), and 0.687
(P= 5.5E-2), espec i ely. When hese po en ial bio-
ma ke s we e combined in a panel o gene exp ession,
he AUC alue was 0.927 (P= 1.1E-5) in DTC e sus
no mal hy oid (Fig. 2c), p o iding a sligh imp o emen
o e he AUC esul s ob ained in each indi idual gene.
Compa ing DTC wi h FTA, he AUC alues o C1QL1,
LCN2,CRABP1 and CILP we e 0.566 (P= 4.8E-1), 0.898
(P= 1.9E-5), 0.746 (P= 8.3E-3), and 0.675 (P=6.1E-2), e-
spec i ely (Fig. 2b). When hese po en ial bioma ke s we e
combined in a panel o gene exp ession, he AUC alue
was 0.839 (P= 2.7E-4) in DTC e sus FTA (Fig. 2d), no
imp o ing he esul in compa ison wi h hose o he
AUC o LCN2 gene. CRABP1 was he bes bioma ke o
Fig. 1 Di e en ially exp essed genes ound be ween minimally and widely in asi e ollicula hy oid ca cinoma (FTC). Gene exp ession was
measu ed by high- h oughpu pai ed-end RNA sequencing and using s a is ically signi ican alues (Q< 0.05). Genes wi h lowe exp ession in
widely in asi e FTC as compa ed wi h minimally in asi e FTC a e lis ed a he op o he igu e (g een). Genes wi h highe exp ession in widely
in asi e FTC as compa ed wi h minimally in asi e FTC a e lis ed a he bo om o he igu e ( ed). De ails o genes, exp ession le els, s a is ical
alues, and molecula unc ion o he genes a e ound in Addi ional ile 2: Table S1 o di e en ially exp essed genes and in Addi ional ile 3: Table
S2 o di e en ially exp essed ansc ip s
Table 2 Diagnos ic pe o mance o po en ial bioma ke s o di e en ia ed hy oid cance (DTC)
Bioma ke Cu -o Sensi i i y (%) Speci ici y (%)
DTC DTC s no mal hy oid DTC s FTA
C1QL1 >1 62.9 (52.0–72.9) 82.4 (56.6–96.0) 58.8 (33.0–81.5)
LCN2 >1 64.0 (52.9–74.0) 73.3 (44.9–92.1) 83.3 (58.6–96.2)
CRABP1 <−1 84.3 (75.0–91.1) 88.9 (65.3–98.3) 35.7 (12.9–64.8)
CILP <−1 61.4 (50.4–71.6) 66.7 (35.0–89.9) 53.0 (27.9–77.0)
Sensi i i ies and speci ici ies a e indica ed compa ing DTC wi h no mal hy oid issue o ollicula hy oid adenoma (FTA)
Celes ino e al. BMC Cance (2018) 18:68 Page 5 o 16

he de ec ion o DTC when es ed agains no mal hy oid.
On he o he hand, LCN2 was he bes bioma ke o he
de ec ion o DTC when es ed agains FTA.
We also es ed gene exp ession le els om an add-
i ional se ies o 393 DTC and 59 no mal hy oid issues
a ailable in TCGA [15]. C1QL1,LCN2 and CRABP1
we e di e en ially exp essed in DTC in compa ison wi h
no mal hy oid issues (Addi ional ile 11: igu e S4),
wi h s a is ical signi icance (P< 0.001).
Clinical and molecula associa ions wi h gain o C1QL1
exp ession
C1QL1 exp ession le els we e success ul measu ed by
qPCR in 89 DTC (15 FTC, 20 FVPTC and 54 PTC), 10
hy oid cance cell lines, 17 FTA, and 18 ma ched hy-
oid no mal issues (Fig. 3a). Gain o C1QL1 exp ession
in ela ion o hy oid no mal issues was ound in FTC,
FVPTC and PTC (Fig. 3b), bu he di e ences only
achie ed he h eshold o s a is ical signi icance in
FVPTC (P= 0.015) and PTC (P= 0.001). Addi ionally,
eigh ou o en (80%) hy oid cance cell lines showed
gain o C1QL1 exp ession (Fig. 3c). Gain o C1QL1 ex-
p ession was signi ican ly associa ed wi h ex a hy oidal
ex ension (P= 0.003) and lymphocy ic hy oidi is (P=
0.003) in he DTC samples (Table 3). All he cases wi h
BRAF mu a ions p esen gain o C1QL1 exp ession (P
<0.001).InFTC,gaino C1QL1 exp ession was
p esen in 71% o wFTC and absen in mFTC (P=
0.007; Addi ional ile 12: Table S6), and lymphocy ic
hy oidi is was only p esen in FTC wi h gain o
C1QL1 exp ession (P=0.026). In FVPTC, absence o
NRAS mu a ions was signi ican ly associa ed o he
gain o C1QL1 exp ession (P= 0.014; Addi ional ile 13:
Table S7). PTC wi h C1QL1 gain we e signi ican ly la -
ge (2.96 ± 0.33 cm) han PTC wi hou gain o exp es-
sion (1.78 ± 0.25 cm; P= 0.021; Addi ional ile 14:
Table S8). Gain o C1QL1 exp ession in PTC was also sig-
ni ican ly associa ed wi h he p esence o ex a hy oidal
ex ension (P= 0.006), and BRAF mu a ions (P=0.001).
Clinical and molecula associa ions wi h gain o LCN2
exp ession
LCN2 exp ession le els we e success ul measu ed by
qPCR in 86 DTC (14 FTC, 19 FVPTC and 53 PTC), 10
hy oid cance cell lines, 18 FTA, and 16 ma ched hy-
oid no mal issues (Fig. 4a). Gain o LCN2 exp ession
Fig. 2 Recei e Ope a ing Cha ac e is ics (ROC) cu es o gene exp ession o po en ial bioma ke s in hy oid cance s. ROC cu es o gene
exp ession o indi idual and combined po en ial bioma ke s in di e en ia ed hy oid cance s (DTC) e sus no mal hy oid issues and DTC e sus
FTA.Thea eaunde heROCcu e(AUC) ep esen s heaccu acyo heindi idual and combined po en ial bioma ke s o dis inguishing di e en ia ed
hy oid cance s (DTC) om no mal hy oid issue samples o ollicula hy oid adenoma (FTA) (a-d). DTC e sus no mal hy oid o indi idual (a)and
combined (c) po en ial bioma ke s wi h gain (C1QL1and LCN2)andloss(CRABP1 and CILP) o gene exp ession in cance . DTC e sus FTA o indi idual (b)
and combined (d) po en ial bioma ke s wi h gain (C1QL1 and LCN2)andloss(CRABP1 and CILP) o gene exp ession in cance . Asymp o ic signi icance,
s anda d e o and 95% con idence in e al measu emen s o all alues can be ound in Addi ional ile 10: Table S5
Celes ino e al. BMC Cance (2018) 18:68 Page 6 o 16
was ound in FTC, FVPTC and PTC (Fig. 4b), bu he
di e ences only achie ed he h eshold o s a is ical sig-
ni icance in PTC (P< 0.001). Th ee ou o en (30%) o
he hy oid cance cell lines showed gain o LCN2
(Fig. 4c). A a iance wi h DTC, signi ican loss o
LCN2 exp ession was ound in FTA (P= 0.013; Fig. 4b).
Gain o LCN2 exp ession in DTC samples was signi i-
can ly associa ed wi h he p esence o ex a hy oidal ex-
ension (P=0.020),oncocy ic ea u es(P=0.018),and he
p esence o BRAF mu a ions (P= 0.031; Table 3). Oncocy-
ic pa e n o he FTC was signi ican ly associa ed wi h
he gain o LCN2 exp ession (P= 0.016; Addi ional ile 12:
Table S6). PTC wi h gain o LCN2 exp ession we e signi i-
can ly la ge (2.72 ± 0.30 cm) han PTC wi hou gain o
exp ession (1.76 ± 0.34 cm; P= 0.023; Addi ional ile 14:
Table S8). Al hough no s a is ically signi ican (P=
0.130), ex a hy oidal ex ension was mo e equen in
PTC wi h gain (54%) han wi hou gain (31%) o
LCN2 exp ession. No signi ican associa ions ha
could be ela ed wi h gain o LCN2 exp ession we e
obse ed in FVPTC.
Clinical and molecula associa ions wi h loss o CRABP1
exp ession
CRABP1 exp ession le els we e success ul measu ed by
qPCR in 89 DTC (15 FTC, 19 FVPTC and 55 PTC), 10
hy oid cance cell lines, 14 FTA, and 19 ma ched hy-
oid no mal issues (Fig. 5a). Signi ican loss o CRABP1
exp ession was de ec ed in FTC, FVPTC and PTC (P<
0.001), and in FTA (P= 0.023; Fig. 5b). No ably, all he
10 hy oid cance cell lines es ed showed loss o
CRABP1 exp ession (Fig. 5c). Loss o CRABP1 exp es-
sion was associa ed wi h encapsula ed DTC (P= 0.037).
BRAF mu a ions (P= 0.025; Table 3) we e only p esen
in DTC wi h loss o CRABP1 exp ession. No signi ican
associa ions we e ound in FTC, FVPTC and PTC wi h
loss o CRABP1 exp ession.
Clinical and molecula associa ions wi h loss o CILP
exp ession
CILP exp ession le els we e success ully measu ed by
qPCR in 88 DTC (14 FTC, 20 FVPTC and 54 PTC), 10
hy oid cance cell lines, 17 FTA, and wel e ma ched
Fig. 3 Di e en ial gene exp ession o C1QL1 in hy oid umou s and no mal issues. Gene exp ession was measu ed in ollicula hy oid adenoma
(FTA), ollicula hy oid cance (FTC), ollicula a ian o papilla y hy oid ca cinoma (FVPTC), papilla y hy oid ca cinoma (PTC), no mal hy oid
issues and hy oid cance cell lines by eal- ime quan i a i e PCR. His og am showing he gene exp ession o each sample o de ed by exp ession
le els (a). Box plo ep esen a ion (median and Tukey whiske s) showing gene exp ession o each subg oup o hy oid umou s –FTA, FTC, FVPTC
and PTC (b). Gene exp ession o he hy oid cance cell lines (c). Gene exp ession was calib a ed by he pool o no mal hy oid issues. S a is ically
signi ican alues: *P= 0.015, **P= 0.001
Celes ino e al. BMC Cance (2018) 18:68 Page 7 o 16
Table 3 Clinicopa hological and gene ic da a o di e en ia ed hy oid cance s classi ied by classes based on gene exp ession
C1QL1 old change LCN2 old change CRABP1 old change CILP old change
No mal ≤1 Gain > 1 P alue No mal ≤1 Gain > 1 P alue Loss < −1 No mal ≥−1P alue Loss < −1 No mal ≥−1P alue
DTC (n=98)
Age (n) 33 56 NS (0.318) 31 55 NS (0.614) 75 14 NS (0.826) 54 34 NS (0.474)
Mean (yea s) 40.9 ± 2.4 44.2 ± 2.4 42.0 ± 2.2 44.0 ± 2.4 43.4 ± 1.9 41.64 ± 3.7 42.9 ± 2.3 44.0 ± 2.7
Tumou size (n) 31 54 NS (0.063) 29 53 NS (0.680) 71 14 NS (0.220) 51 33 NS (0.582)
Mean (cm) 2.20 ± 0.22 3.10 ± 0.27 2.59 ± 0.32 2.67 ± 0.24 2.86 ± 0.22 2.29 ± 0.43 2.87 ± 0.28 2.49 ± 0.23
Gende (n) 33 56 NS (0.148) 31 55 NS (0.232) 75 14 NS (0.350) 54 34 NS (0.175)
Female (%) 26 (78.8) 50 (89.3) 28 (90.3) 45 (81.8) 63 (84.0) 13 (92.9) 44 (81.5) 31 (91.2)
Male (%) 7 (21.2) 6 (10.7) 3 (9.7) 10 (18.2) 12 (16.0) 1 (7.1) 10 (18.5) 3 (8.8)
Capsule (n) 30 50 NS (0.263) 26 51 NS (0.099) 66 14 0.037 49 30 NS (0.128)
Posi i e (%) 18 (60.0) 25 (50.0) 17 (65.4) 24 (47.1) 39 (59.1) 4 (28.6) 29 (59.2) 13 (43.3)
Capsula in asion (n) 17 23 NS (0.496) 15 23 NS (0.332) 37 3 NS (0.704) 28 11 NS (0.191)
Posi i e (%) 12 (70.6) 15 (65.2) 11 (73.3) 14 (60.9) 25 (67.6) 2 (66.7) 17 (60.7) 9 (81.8)
Vascula in asion (n) 30 51 NS (0.516) 28 51 NS (0.245) 67 14 NS (0.060) 50 30 NS (0.261)
Posi i e (%) 17 (56.7) 30 (58.8) 14 (50.0) 31 (60.8) 42 (62.7) 5 (35.7) 30 (60.0) 15 (50.0)
Lymph node me as asis (n) 31 54 NS (0.398) 30 53 NS (0.073) 71 14 NS (0.397) 51 33 NS (0.074)
Posi i e (%) 7 (22.6) 15 (27.8) 5 (16.7) 18 (34.0) 21 (29.6) 3 (21.4) 10 (19.6) 12 (36.4)
Ex a hy oidal ex ension (n) 30 51 0.003 29 50 0.020 67 13 NS (0.328) 48 32 NS (0.332)
Posi i e (%) 4 (13.3) 23 (45.1) 5 (17.2) 21 (42.0) 23 (34.3) 3 (23.1) 17 (35.4) 9 (28.1)
Dis an me as asis (n) 31 54 NS (0.096) 30 53 NS (0.401) 71 14 NS (0.397) 51 33 NS (0.657)
Posi i e (%) –5 (9.3) 1 (3.33) 4 (7.55) 5 (7.04)) –2 (3.92) 2 (6.06)
Lymphocy ic hy oidi is (n) 32 51 0.003 30 51 NS (0.416) 69 14 NS (0.563) 49 33 NS (0.354)
Posi i e (%) 7 (21.9) 28 (54.9) 12 (40.0) 23 (45.1) 31 (44.9) 6 (42.9) 19 (38.8) 15 (45.5)
Oncocy ic (n) 31 50 NS (0.290) 28 51 0.018 67 14 NS (0.329) 48 32 NS (0.473)
Posi i e (%) 5 (16.1) 12 (24.0) 2 (7.14) 15 (29.4) 13 (19.4) 4 (28.6) 9 (18.8) 7 (21.9)
PAX8-PPARG ea angemen s (n) 33 56 NS (0.371) 31 55 NS (0.360) 75 14 NS (0.843) 54 34 NS (0.386)
Posi i e (%) 1 (3.03) –1 (3.23) –1 (1.33) ––1 (2.94)
RET/PTC ea angemen s (n) 33 56 NS (0.493) 31 55 NS (0.174) 75 14 NS (0.617) 54 34 NS (0.160)
Posi i e (%) 5 (15.1) 10 (17.9) 3 (9.68) 11 (20.0) 12 (16.0) 2 (14.3) 7 (13.0) 8 (23.5)
RET/PTC1 ea angemen (n) 33 56 NS (0.618) 31 55 NS (0.163) 75 14 NS (0.425) 54 34 NS (0.070)
Posi i e (%) 4 (12.1) 7 (12.5) 2 (6.45) 9 (16.4) 10 (13.3) 1 (7.14) 4 (7.41) 7 (20.6)
Celes ino e al. BMC Cance (2018) 18:68 Page 8 o 16
Table 3 Clinicopa hological and gene ic da a o di e en ia ed hy oid cance s classi ied by classes based on gene exp ession (Con inued)
C1QL1 old change LCN2 old change CRABP1 old change CILP old change
No mal ≤1 Gain > 1 P alue No mal ≤1 Gain > 1 P alue Loss < −1 No mal ≥−1P alue Loss < −1 No mal ≥−1P alue
RET/PTC3 ea angemen (n) 33 56 NS (0.629) 31 55
1
75 14
1
54 34 NS (0.614)
Posi i e (%) –1 (1.79) –– –– 1 (1.85) –
BRAF mu a ion (n) 33 56 <0.001 31 55 0.031 75 14 0.025 54 34 NS (0.331)
Posi i e (%) –18 (32.1) 3 (9.68) 16 (29.1) 19 (25.3) –13 (24.1) 6 (17.6)
NRAS mu a ion (n) 33 56 NS (0.213) 31 55 NS (0.300) 75 14 NS (0.090) 54 34 0.001
Posi i e (%) 7 (21.2) 7 (12.5) 6 (19.4) 7 (12.7) 13 (17.3) –14 (25.9) –
TERT p omo e mu a ion (n) 33 56 NS (0.244) 31 55 NS (0.706) 75 14 NS (0.595) 54 34 NS (0.669)
Posi i e (%) –3 (5.36) 1 (3.23) 2 (3.64) 3 (4.00) –2 (3.70) 1 (2.94)
n, numbe o cases wi h a ailable da a; 1, no s a is ics we e compu ed due o cons an numbe s o one ea u e
Bold alues indica e he esul was s a is ically signi ican
Celes ino e al. BMC Cance (2018) 18:68 Page 9 o 16
neoplasm cy ology by Thy oSeq 2 nex -gene a ion sequencing assay.
Cance . 2014;120:3627–34.
41. McI e B, Cas o MR, Mo is JC, Be ne V, Small idge R, Hen y M, Kosok L,
Reddi H. An independen s udy o a gene exp ession classi ie (A i ma) in
he e alua ion o cy ologically inde e mina e hy oid nodules. J Clin
Endoc inol Me ab. 2014;99:4069–77.
42. Ka gi AY, Bus aman e MP, Gulec S. Genomic p o iling o hy oid nodules:
cu en ole o Thy oSeq nex -gene a ion sequencing on clinical decision-
making. Mol Imaging Radionucl The . 2017;26:24–35.
• We accep p e-submission inqui ies
• Ou selec o ool helps you o ind he mos ele an jou nal
• We p o ide ound he clock cus ome suppo
• Con enien online submission
• Tho ough pee e iew
• Inclusion in PubMed and all majo indexing se ices
• Maximum isibili y o you esea ch
Submi you manusc ip a
www.biomedcen al.com/submi
Submi you nex manusc ip o BioMed Cen al
and we will help you a e e y s ep:
Celes ino e al. BMC Cance (2018) 18:68 Page 16 o 16