Liu et al. BMC Gastroenterol (2020) 20:415 https://doi.org/10.1186/s12876-020-01560-0

RESEARCH ARTICLE Open Access Prognostic value of family in early-stage pancreatic ductal adenocarcinoma after pancreaticoduodenectomy and possible mechanisms Jun‑Qi Liu1, Xi‑Wen Liao1, Xiang‑Kun Wang1, Cheng‑Kun Yang1, Xin Zhou1, Zheng‑Qian Liu1, Quan‑Fa Han1, Tian‑Hao Fu1, Guang‑Zhi Zhu1, Chuang‑Ye Han1, Hao Su1, Jian‑Lu Huang1,2, Guo‑Tian Ruan3, Ling Yan3, Xin‑Ping Ye1 and Tao Peng1*

Abstract Background: This study explored the prognostic signifcance of Glypican (GPC) family genes in patients with pan‑ creatic ductal adenocarcinoma (PDAC) after pancreaticoduodenectomy using data from The Cancer Genome Atlas (TCGA) and Expression Omnibus (GEO). Methods: A total of 112 PDAC patients from TCGA and 48 patients from GEO were included in the analysis. The relationship between overall survival and the expression of GPC family genes as well as basic clinical characteristics was analyzed using the Kaplan-Meier method with the log-rank test. Joint efects survival analysis was performed to further examine the relationship between GPC genes and prognosis. A prognosis nomogram was established based on clinical characteristics and prognosis-related genes. Prognosis-related genes were investigated by genome-wide co-expression analysis and gene set enrichment analysis (GSEA) was carried out to identify potential mechanisms of these genes afecting prognosis. Results: In TCGA database, high expression of GPC2, GPC3, and GPC5 was signifcantly associated with favorable survival (log-rank P 0.031, 0.021, and 0.028, respectively; adjusted P value 0.005, 0.022, and 0.020, respectively), and joint efects analysis= of these genes was efective for prognosis prediction. The= prognosis nomogram was applied to predict the survival probability using the total scores calculated. Genome-wide co-expression and GSEA analysis sug‑ gested that the GPC2 may afect prognosis through sequence-specifc DNA binding, transport, cell diferentia‑ tion and oncogenic signatures (KRAS, RAF, STK33, and VEGFA). GPC3 may be related to cell adhesion, angiogenesis, infammatory response, signaling pathways like Ras, Rap1, PI3K-Akt, chemokine, GPCR, and signatures like cyclin D1, p53, PTEN. GPC5 may be involved in transcription factor complex, TFRC1, oncogenic signatures (HOXA9 and BMI1), gene methylation, phospholipid metabolic process, glycerophospholipid metabolism, cell cycle, and EGFR pathway.

*Correspondence: [email protected] 1 Department of Hepatobiliary Surgery, The First Afliated Hospital of Guangxi Medical University, Shuang Yong Rd. 6#, Nanning 530021, Guangxi Zhuang Autonomous Region, People’s Republic of China Full list of author information is available at the end of the article

© The Author(s) 2020. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativeco​ mmons​ .org/licen​ ses/by/4.0/​ . The Creative Commons Public Domain Dedication waiver (http://creativeco​ ​ mmons.org/publi​ cdoma​ in/zero/1.0/​ ) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Liu et al. BMC Gastroenterol (2020) 20:415 Page 2 of 23

Conclusion: GPC2, GPC3, and GPC5 expression may serve as prognostic indicators in PDAC, and combination of these genes showed a higher efciency for prognosis prediction. Keywords: GPC family genes, Pancreatic ductal adenocarcinoma, Prognostic indicator, Mechanism

Background In this study, we explored the relationship between GPC Pancreatic cancer (PC) is related to an unfavorable prog- family genes expression and prognosis of PDAC patients. nosis, and its mortality rate is close to its incidence rate [1]. Te incidence of PC is predicted to rise 40% in the Methods next 10 years in North America and Europe [2], and Patient data according to the latest statistics, PC ranks fourth among Te RNA-sequencing dataset used in this study and cancers directly causing death for men and women in the corresponding clinical data were acquired from Te the United States [3], moreover, by 2030, its rank may Cancer Genome Atlas (TCGA) (https​://porta​l.gdc.cance​ increase to second [4]. In China, the prognostic status r.gov/; accessed September 25, 2019), and DESeq was of PC patients is also severe, and 5-year survival rate of applied to normalize the initial material [24]. To increase patients with PC after age standardization is approxi- reliability of data analysis, previously established inclu- mately 11.7% [5]. Due to the unique biological behaviors sion and exclusion criteria were used [25]. Te inclu- of PC, metastasis is present when patients are diagnosed sion criteria were as follows: (i) survival information was and only 9.7% patients can be diagnosed at an early stage complete; (ii) histology result was confrmed as PDAC; [6]. Furthermore, the 5-year survival rate is 9% for PC at (iii) pathologic stage was I or II; (iv) pancreaticoduo- all stages and 3% at advanced stages [3]. So far, surgical denectomy was carried out on patients. PDAC patients resection remains the best therapy for PC at the early with pathologic stage III or IV and those who under- stage [7]. Terefore, identifying reliable early molecular went other surgical procedures were excluded from the markers to improve prognosis of PC is important. study. According to the above criteria, 112 patients were Glypican (GPC) family genes include six members included in the analysis. Te clinical characteristics (GPC1, GPC2, GPC3, GPC4, GPC5, GPC6), and all of the included in the analysis were age, sex, alcohol history, GPC family are expressed in human [8]. are pathologic stage, histologic grade, radical resection, radi- attached to the cell membrane and function in biological ation therapy, targeted molecular therapy, overall sur- processes such as cell and tissue growth, embryo devel- vival (OS) time, and survival status. Dataset GSE62452 opment, and cell movement [9, 10]. Tey are reported to was downloaded from Gene Expression Omnibus (GEO) be related to multiple diseases including various cancers. database to validate the prognostic value of survival- GPC1 is upregulated in pancreatic cancer [11], esopha- related genes (https​://www.ncbi.nlm.nih.gov/geo/query​/ geal cancer [12], and prostate cancer [13]. Li et al. report acc.cgi?acc=GSE62​452; accessed October 5, 2020). Fol- that GPC1 contributes to the proliferation and motil- lowing the same criteria described above, we included 48 ity of esophageal cancer cells through the PTEN/Akt/β- cases in this study. catenin pathway [14]. Increased level of GPC3 in serum could serve as a marker for hepatoblastoma [15] as well Analysis using public database as (HCC) [16, 17]. GPC3 dele- Te expression status of GPC family genes in diferent tion mutation can help in diagnosis of Simpson-Golabi- normal tissues was analyzed by the Genotype-Tissue Behmel syndrome type 1 (SGBS1), which is a serious Expression (GTEx, https​://www.gtexp​ortal​.org/, accessed genetic disease [18, 19]. Overexpression of GPC5 may October 9, 2019) website [26, 27]. Te Gene Expression accelerate tumor progression of lymphoma [20]. In addi- Profling Interactive Analysis (GEPIA, http://gepia​.cance​ tion, GPC5 may play a role in strengthening the inter- r-pku.cn/, accessed October 9, 2019), an online tool con- action between Patched 1 and Hedgehog signaling in taining 9736 tumors and 8587 normal samples from the rhabdomyosarcoma [21]. GPC5 may serve as a key gene TCGA and the GTEx projects, was used to show expres- afecting the cell cycle of podocytes in kidneys, fnally sion level of each gene in both tumor and normal tissues causing nephrotic syndrome [22]. of PDAC [28]. Te Database for Annotation, Visualiza- Pancreatic ductal adenocarcinoma (PDAC) accounts tion, and Integrated Discovery (DAVID) v6.8 (https://​ for more than 80% of pancreatic neoplasms [1, 23]. How- david​.ncifc​rf.gov/, accessed November 6, 2019) [29, 30] ever, there are few studies on the prognostic value of GPC was chosen to carry out gene enrichment analysis con- family genes in early-stage PDAC after pancreaticoduo- taining (GO) function analysis and Kyoto denectomy despite the poor prognosis of this tumor type. Encyclopedia of Genes and Genomes (KEGG) pathway Liu et al. BMC Gastroenterol (2020) 20:415 Page 3 of 23

a Gene expression for GPC1 (ENSG00000063660.8) b Gene expression for GPC2 (ENSG00000213420.7)

2.5

1.5

2.0 )

1.5 1.0 log10(TPM+1) log10(TPM+1

1.0

0.5

0.5

0.0

NerveSkin -B S B CellsArteryP -V Es BrainA - AdreCo Tn CervixCer Muscv Ov ArteHeartPituitar Br B UterusF He Bla BrainE - AdNu ColonB E BraiBrainn Br Mi B B A LungK Th C B K S SpleBraiS L Pa Cell TesSkinSkinU - O C Ce BraV Ne F SpleenBraiBraiCellBraiSmallBr aProP ColBreaBl M BraAdiposBraiBraC BraiB B LunArt BrainBrai - CellNu Art EsopEsoAd Es T Ki WhStomaAd Art HearKidnH MusclePa Li ra kin ra rostatagin o rt est a rai all sophag rainsop a n re ra d id y o rainid to maiv t var e agi al itu a in o r r hy d e v in in ph ery i ary ry in n - opiaart d ipose in - or ast i-n Maipo n ro lon n mac n er ncre s t eru rvix rvix in - rve lo in - statHip o ddero i i lon - ain ain- Hypot - Caery - e ip o n o renae a n e - - - Su - Cerebe - Cul a s ix le - de - ha - - - s ey id - ey en - ll - is - y n pi n n s n itaryn s r Salin n n n - Corte g n s ry h pha o pharo e le ry - t - e rt - L crear No Ce e a - Coal - En - - S - - L y F A - A r - Ca A HippocampusHyS Putae - - S - Corte S In a EB Not SuSunsn E a - a - Cere- Ce- EB- Intestine - - t - e - Su- - Pu- Cul- a se id y - c y T n A gu Gl Ecto Ti e rontalnt Corten tria us - Sig gus -mygda alivar - M Tra u h p te s - E - E S T n Am e T - Ma A - FrontalSigmoi d Aorta C gusg - gu Bl h l Gl T A - Cortex - s ibiat S re Exposedt or(Lowt s rte a k b ft e T cle S u p V e bsta in s V- c ndpi ibia T re V ran vanterioSu b u cleutamens o(ba us - Vis s M oo ib trial ef Skelet un Expbe ur a - ron nd doc elet ial Ve rior ub l - Gaus mo da o m m isc dull nsverse a tine tr Ex tocerv nal cord ube - yg p m st da tu rona e d a ia t l e Mux ce e Ap us b t M th y ma e l co a o l be bel tr da ocampu s ry bc a x h re Mu- M ce- Gas du nd l Ap Ven llu llar d ar erv rvixa n cin cu id e ( usc la alam G n era a nt x - nsform posed ce l ansfo ve m Glanr cind nt C alamuste d ll m fibrobla co y l tricle pen acc tan b lan ry (b ia ni rd T x rvix la lum la rs ar ut ia or (ba a f ry s uc ra a p triclal He sa ix gu troe asalu asl (Om e p ix r Hemispher Te e y aneo u ccumbenib cularisosal (Otroe en os x d u eous lari us d Ti g (cervrm os (ce rm rminas T gul nig tex sal g ro d e ed (S mi (Blate cort ag mb so ss a ra inal Ileu ed e (L ed ly issu a ra sa b m s a sphes e en ph ga s ue l gangliae d (So rvica te (BA l la e oph ge e ts A9) nt ic ly w l Ileu e cortexs a gansts n u r leg) s ag nglia u al mp er leg m 9) n s (b tu ageal Ju p re (b e m) c-1) upr l c glia gl m ra ex as al Junction ) m ho -1) e ph m ia ) p (B ) c ap ) o ) as ) ubic) al yt ub cyt (B al ga A24 g es ic e A2 n an ) s 4) ct ) glia nglia ion ) )

c Gene expression for GPC3 (ENSG00000147257.13) d Gene expression for GPC4 (ENSG00000076716.8)

2.5 2.0

2.0 1.5

1.5

1.0 log10(TPM+1) log10(TPM+1)

1.0

0.5 0.5

0.0 0.0

LungA N B K Ad P Th A F Ov B S KidneyC Mi C H E A Va St S A A Ut P C E S C T P CellH S B MuscleE B Li B B B B B B BraiB B B B W C EsophaE A A ColoArteN LungPituitUa Ov C Bl Fa Ki Ad CervixCe Ad S Th StoKidneyMino P- r T Ad Br Mu Va Pa Sk H B S Es Sk B Br Br Br BrainB - BCortexBr Br C He BrainB C Br LiveW di ervere id ituitay rt al la ki e n er e so rtery gina- omki drenrt e ro o so mallo es ancre earpl rai so ra v ra ra ra ra rainrain - - Nuclrainra - i ra ra h el s rterrtery - e ter o a dn mall In ro es e earrainpleen ra a ain ai ra rai ai ai e rainel - Cereain - Cehole Blo p as n ipos r ery -lo Coroar dd n rv or vi art ph n ery ru st lon ph lon tis s - e n p in er in in in - in n n in in - o ls op ry rv ar lon -d llo ip rvixi p- Ecto yro ma s tis renala Gls gi nc in ophagusin in in n in n n n lls art ls r ose - t ey - oid pi y er - ix - x a - s a t en - hag ------le - h y n e - us y der p ey os ose id Satatel st -cl Mae na r - Sunt Expo - N - Ca - Fr - - Hy - Hi - - - - - e - ry an S - S - - A a ch Notal S - te - a In - a C - Lef H S C P C A S A H C Fr Cortex B E agus - - G - - ry ian e - Subcut- ch - ea - - S Anterior- Am cing - Su N C - Lef- Putame EB T M M un EC E aliv E tr gusT G A S gus testTi r s ultured fib yp- Skeletus pi audutamer m ubs nt i erebe loodB gus A T S C T T - M Endo - te C S s Atrialp o u ul S ib ammarye Vi T ortec ndocial ibi landorta igm ans t na ygda er ppoc ont V ortaibiaig o ibial ra Tu V st o iv a k inal co t S u ontal y p bstapp c t t V- r o u ial d scera ube toc ary - a un - Mune V o - ebel tantei -transform - moronar n b ed is ine rtexar n mmelet - u d g o o leusu redac V fi b t ebe d bcutaneo ull n xpos ervi A G l oid vers en thal Mul ate (en us ior al Mu l sve e u ce c c y Gl d Ap M n at dala tha cam e e ra a ary x Gl e pp as E - al cord lu l ci ampulla Cor as id lla ervixeral - a a uc Expe ( C ntia n n llu n lla ed a rv en xp s T tricl a c ba (basm a a ac ng r sculari y rs a rvi T and r l s pe r b o lamu tric ( m sf r l (Oment x nd ix tro cu e e r mu os n cumb H te tro e n x (Oe y ed n d (c os a rtex ni pus c bro l b o H Tis (Lo da e ose larrmi oblastse s a (c sal ig ula s emisphex e e rminal Il Tiss ( d a os sa ulate corte gr umb e as rm em us su ge so e al r ( e so o m L agee e l g ( s blasts a ed l w p d (S is nal Il rv ganglia ga a te B d s p us e u ower leg)rv d ( B a l g is e er leg) hage ica ens cort A9 ly h n e ica Suprapua A en an ymp u uprapub e l c ngl ) mphoc ag tu eu l ngl 9 s h m) u (ba re e m) c- ia ) ( glia er al Ju m -1 ) ia) ex (B a m 1) ) x bas h e ) sal l Junctio ( ) oc ytes b B al g yt nction ic ganA24) ic A2 es ) ) 4) ang gli n li efa) a) Gene expression for GPC5 (ENSG00000179399.14) Gene expression for GPC6 (ENSG00000183098.10)

1.4

1.2 1.5

1.0

0.8 1.0 log10(TPM+1) log10(TPM+1) 0.6

0.4 0.5

0.2

0.0 0.0 Ar C N Ar A B Ov C F E Ad Ad CervEix Pr Co Lu Li C SmPituV StomBr K U T H K Sp Te Sk M E B B Sk Br Br Ad B Br B Br Br Br Br B Br H Pa M C W BraiB B B BraiBraiBr BraiB Te BraiPi C B B Ki B Ar L K N A T H ArteryCo lCo Ov P L V HeartS - ELe Sk E BreaM Sk StoC Ad A B Fa UterSpl C A E C P W Mus ter el er teryrte lad a olona l sophiposipos sophaos l n ve erv agina easidn ter hyroearidn l st i in so rainr ain in ain a r rai ainr ain- a ai a ainrai ai ear n u ells h r r r r el r ra ra u i e r hy e r iv ag m s s i ervix dr la el diposs e an ls v r de ry lop on -g r all it ey -u eye -en i n o p - N in e n - in - n in n n cr sc ole ai ai ain ai ai st tui ls ai i dn i ter n dn r ter art - o lo a os e a ophaguin ophagun in m ipo e d ll e l ophagur hole Bl y - e - y - Co ia ta ix a a t s id t - s - r Sah - - - H - nal Gland ------Ht - Lee le - - n n n n n - n n n is n ta n n e n y g ey ve y -r Coroi n n ry t r ina ll - s or - Su n d opianu en s vi cr cl - C - r - Si agu e e - gute - EIntestiry ch - At S ag C C Nu S F Ce P S Ante C Am as E B - Ca- Nucl------Hypot- r - - - y - - - d - - Si- ate In Not Sunt S ac s a er s - x ea e Ti ult TibialAorta n - - E Tr M C M un l u audatee o (bt asalyp ga ubstaro ut p orte ippo S B lood P A F C AmyS H y E Spi C - MeduC T - T At A T t - Mammaryal h - Ece l G C e - E s - bial ured r g Tu s S V ct s an ndo am or rial e Exivarys reb Su c ntalrebela in ygda ft keletaV utamennte ro o ub ip BV ereb er ib C ibi ri ortag ran f es iv n - Subcut T ul - n ood S ona m be - ub isceoc - s n tex dulla - ot le mena (brio V -tran ud n rtex gdals p - n e ia orte al o a m t ti s - s - ar Expos to land ub tu Vi s - docerv k oid G c e M v c e m A pos M e n Expohal ussed (S n l co r x ca en a eus rior ta ta h oc tranal bell l n l Appe oi sv V ne Ga y c redsce ele ry as ut r rvi uscu er er - a pp G uc llu a tia C lar Hem cing la mpust s te l n al ampu c ell l x a d e entr M e e M t fib t a al x s v T ry en ed land(Lowos m amu cc or rd ricle l f ( ac c Cor a tia amu s or umla ar ry rs - s E usc G rvi fi ral uco al rob ro ne ( e ix er Ti n tex (B a ( orm bas (b in ni for d Hemisph e ic Te tr xp lan ed x a b i esopoh O la m s dag a s umbensig sal cerviulca cu as gulatte med( nda le r oeso ularisT d neous robl(Omens x las usm ri inal s ra a e a m a x gr s s cer mi sed ( issu (L a en s ue e is g te d ly l b l ga (BA9) a v g n opha o a ts t I e n A9) pherean c ganglens e ic e al e wer sts age um leu r g u gl l or m n c lymphal I S tu le lia prap (b ia c- tex ph gl or e leu g uprapub l m) al Ju ) m g) ) asal ) 1) o ia (ba ia) tex c-1 re m eal Junc eg ubic) (BA2 cy ) o ) ) t sal (BA24) cyt nct gang es ga e i io 4) s t c) n lia ng ion ) lia) Fig. 1 Gene expression levels of Glypican family genes in normal human organ tissues. a-f Gene expression levels of Glypican1–6, respectively

analysis. Te possible functioning pathways of the genes regression analysis was used to adjust for prognosis-sig- were also investigated by Biological Network Gene nifcant factors. Hazard ratio (HR) and 95% confdence Ontology (BiNGO) in Cytoscape (version 3.7.1) [31]. interval (CI) were considered to estimate the relative risk. Stratifed analysis was carried out based on cer- tain clinical characteristics of the patients for survival- Survival analysis related genes to explore their signifcance in prognosis. Two groups of patients were set up based on 50% cutof To understand the relationship between GPC genes and expression value of each gene both in TCGA database prognosis at a deeper level, joint efects survival analy- and GEO database. Te relationship between OS and sis was taken into consideration. Te survival-signifcant gene expression level as well as basic clinical character- clinical characteristics, clinical factors usually related to istics was analyzed using Kaplan-Meier method with the prognosis of patients with malignant tumors clinically log-rank test. Log-rank P < 0.05 was considered statisti- and prognosis-related genes were included to establish a cally signifcant. Multivariate Cox proportional hazards Liu et al. BMC Gastroenterol (2020) 20:415 Page 4 of 23

ab c

de f

Fig. 2 Gene level distribution of Glypican family genes in pancreatic ductal adenocarcinoma between tumor and normal tissues. a-f Gene level distribution of Glypican1–6 in pancreatic ductal adenocarcinoma between tumor and normal tissues, respectively. Notes: *P < 0.05. prognosis nomogram. Better survival prediction could be Genome‑wide co‑expression analysis made according to the total points. Genome-wide co-expression analysis of prognosis- related genes was performed to investigate their potential biological mechanisms based on TCGA database. A gene Liu et al. BMC Gastroenterol (2020) 20:415 Page 5 of 23

Coreceptor activity involved in , planar cell polarity pathway -log10(P.value) Retinoid metabolic process 15 Proteinaceous extracellular matrix Plasma membrane 10 Wnt signaling pathway, planar cell polarity pathway Extracellular space 5 Lysosomal lumen Integral component of plasma membrane binding Gene number Golgi lumen 2 Glycosaminoglycan metabolic process 3

Gene Ontology Glycosaminoglycan catabolic process 4 Glycosaminoglycan biosynthetic process 5 Anchored component of membrane 6 Anatomical structure morphogenesis 0 250 500 750 Fold enrichment Fig. 3 Function enrichment analysis of Gene Ontology for Glypican family genes completed by the Database for Annotation, Visualization, and Integrated Discovery

with Pearson correlation coefcient > 0.5 and P < 0.05 was Te unpaired t test was used to compare gene expres- considered as a co-expression gene. A co-expression net- sion levels between normal and tumor tissues. Te work was built for each gene related to prognosis and its expression relationship of each GPC gene and its co- co-expressed genes using Cytoscape software (version expressed genes was quantifed by Pearson’s correlation 3.7.1) [32]. GO function analysis and KEGG pathway coefcient. Te correlation plot was constructed using analysis of these genes were also completed using DAVID Cytoscape software (version 3.7.1). All statistical analyses [29, 30]. were performed using SPSS v.25.0 software (IBM, Chi- cago, IL, USA). A P value < 0.05 was considered statisti- cally signifcant. Gene set enrichment analysis To understand the underlying mechanisms of GPC Results genes afecting prognosis, we used Gene Set Enrichment Analysis using public database Analysis (GSEA, http://softw​are.broad​insti​tute.org/gsea/ Te expression status of GPC family genes in tissues index​.jsp, November 6, 2019) [33, 34]. Databases c2 (c2. derived from various normal human organs was analyzed all.v7.0.symbols.gmt) and c6 (c6.all.v7.0.symbols.gmt) in using GTEx (Fig. 1). Te expression level of GPC family the Molecular Signatures Database (MSigDB) [35] were genes was lower in human pancreas than in other organs. used to search for possible pathways based on TCGA Te results of GEPIA analysis showed that expression of database. Enrichment results were considered statisti- GPC1, GPC3, GPC4, and GPC6 was signifcantly higher cally signifcant if the nominal P-value was < 0.05 and the in PDAC tumor tissues than in normal tissues (P < 0.05) false discovery rate (FDR) was < 0.25. (Fig. 2). GO functional enrichment analysis indicated that GPC family genes were mainly involved in composition Statistical analysis of cell membrane, organelles and anchored components Survival analysis was performed using Kaplan-Meier of the membrane, heparan sulfate proteoglycan bind- method with log-rank test. Univariate and multivariate ing, and glycosaminoglycan metabolic process (Fig. 3, survival analyses were performed with Cox proportional Additional fle 1: Table 1). Te results of BiNGO analysis hazards regression model to calculate crude and adjusted (Fig. 4) confrmed those of GO analysis. HRs and 95% CIs. Survival curves were plotted using GraphPad Prism v.7.0 (GraphPad Software Inc., La Jolla, CA). Liu et al. BMC Gastroenterol (2020) 20:415 Page 6 of 23

Fig. 4 Functioning pathways of Glypican family genes carried out by Biological Network Gene Ontology in Cytoscape software

Survival analysis fle 2: Table 2 shows that histologic grade, extent of sur- Te Kaplan-Meier method and log-rank test were used gery, treatment with radiation and targeted molecular to investigate the association between basic clinical therapy were signifcant in OS. GPC family genes were characteristics and OS in TCGA database. Additional Liu et al. BMC Gastroenterol (2020) 20:415 Page 7 of 23

a 100 b 100 Low GPC1 Low GPC2 l 80 l 80 High GPC1 High GPC2 viva viva

ur 60 ur 60 ts ts 40 40

Percen 20 Percen 20 Log-rank P=0.955 Log-rank P=0.031 0 0

0200 400600 8001000120014001600 0200 400600 8001000120014001600 Overall Survival(days) Overall survival (days)

c 100 d 100 Low GPC3 Low GPC4 l 80 l 80 High GPC3 High GPC4 viva viva ur 60 ur 60 ts ts 40 40 Percen 20 Percen 20 Log-rank P=0.021 Log-rank P=0.794 0 0

0200 400600 8001000120014001600 0200 400600 8001000120014001600 Overall survival (days) Overall survival (days)

e 100 f 100 Low GPC5 Low GPC6 80 80

ival High GPC5 ival High GPC6 rv 60 rv 60 su su

40 40

Percent 20 Percent 20 Log-rank P=0.028 Log-rank P=0.083 0 0

0200 400600 8001000120014001600 0200 400600 8001000120014001600 Overall survival (days) Overall survival (days) Fig. 5 Kaplan-Meier survival curves of Glypican family genes for pancreatic ductal adenocarcinoma in The Cancer Genome Atlas database. a-f Kaplan-Meier survival curves of Glypican1–6, respectively divided into two groups based on expression level, and HR = 0.449, 95% CI = 0.258–0.782; adjusted P = 0.022, survival analysis was performed between the two groups. adjusted HR = 0.531, 95% CI = 0.309–0.914; and adjusted Te results (Fig. 5a–f) demonstrated that expression P = 0.020, adjusted HR = 0.525, 95% CI = 0.306–0.902, of GPC2, GPC3, and GPC5 was signifcantly associated respectively). Results of stratifed analysis for GPC2, with survival. Te median survival time (MST) was sig- GPC3, and GPC5 are shown in Table 2. High expression nifcantly longer in patients with high expression of of GPC2 was signifcantly associated with better OS in GPC2, GPC3, and GPC5 than the low expression group patients who were male, were > 60 years old, had histo- (log-rank P = 0.031, 0.021, and 0.028, respectively; MST, logic grade G1 or G2, had R1 or Rx resection or whether 634 days vs. 481 days, 614 days vs. 473 days, and 593 days received radiation therapy. GPC3 expression was related vs. 485 days, respectively, Fig. 5b, c, e and Fig. 6). After to patients who were female, were > 60 years old, had his- adjusting for survival-signifcant clinical parameters in a tologic grade G1 or G2, or did not receive radiation or multivariate Cox proportional hazards regression model, targeted molecular therapy. Moreover, GPC5 could infu- GPC2, GPC3, and GPC5 were still signifcantly asso- ence prognosis of patients who were ≤ 60 years old, had ciated with OS (Table 1) (adjusted P = 0.005, adjusted Liu et al. BMC Gastroenterol (2020) 20:415 Page 8 of 23

abc

Fig. 6 Prognostic models of Glypican2, Glypican3 and Glypican5 for pancreatic ductal adenocarcinoma in The Cancer Genome Atlas database. From top to bottom are expression values of these genes, survival status of patients and expression heatmaps of these genes at low and high expression levels. a-c Prognostic models of Glypican2, Glypican3, and Glypican5, respectively

Table 1 Prognostic value of Glypican family genes in The Cancer Genome Atlas database Gene No. of events (%) MST (days) Crude HR Crude Adjusted HR Adjusted 95% CI P value 95% CI­ a P value­ a

GPC1 Low 36/56(64.3%) 518 1 1 High 33/56(58.9%) 511 0.986(0.610–1.594) 0.955 1.120 (0.661–1.896) 0.674 GPC2 Low 41/56(73.2%) 481 1 1 High 28/56(50.0%) 634 0.589(0.362–0.959) 0.031 0.449(0.258–0.782) 0.005 GPC3 Low 38/56(67.9%) 473 1 1 High 31/56(55.4%) 614 0.568(0.349–0.925) 0.021 0.531(0.309–0.914) 0.022 GPC4 Low 35/56(62.5%) 517 1 1 High 34/56(60.7%) 592 1.066(0.659–1.723) 0.794 1.252(0.744–2.105) 0.397 GPC5 Low 35/56(62.5%) 485 1 1 High 34/56(60.7%) 593 0.577(0.351–0.948) 0.028 0.525(0.306–0.902) 0.020 GPC6 Low 37/56(66.1%) 485 1 1 High 32/56(57.1%) 603 0.647(0.393–1.063) 0.083 0.891(0.501–1.585) 0.693

Abbreviations: MST median survival time, HR hazard ratio, CI confdence interval a Adjusted for histologic grade, targeted molecular therapy, radiation therapy and radical resection

histologic grade G3 or G4, had R1 or Rx resection, or did (MST = 278 days, adjusted P value < 0.001). Te group not receive radiation or targeted molecular therapy. of GPC2 and GPC5 was associated with the high- est risk of death in group I (MST = 278 days, adjusted Joint efects analysis P value < 0.001) and the group combining GPC3 Based on the prognostic signifcance of each GPC fam- and GPC5 showed the poorest prognosis in group i ily gene, we combined every two genes among GPC2, (MST = 278 days, adjusted P value < 0.001). GPC3, and GPC5 to investigate their signifcance in We also analyzed survival associated with the three PDAC prognosis. Te combination of GPC2 and GPC3 genes simultaneously. Group A showed the worst in sur- was associated with worse survival outcome in group 1 vival status (MST = 219 days, adjusted P value = 0.018), Liu et al. BMC Gastroenterol (2020) 20:415 Page 9 of 23 a Adjusted P Adjusted 95%CI 0.043 0.357 0.023 0.049 0.154 0.012 0.610 0.118 0.073 0.075 0.191 0.026 Adjusted HR Adjusted 95% CI 0.197 (0.041–0.951) 0.712 0.346–1.466) 0.479 (0.254–0.904) 0.397 (0.158–0.995) 0.637 (0.342–1.185) 0.305 0.120–0.770) 0.725 (0.210–2.498) 0.571 (0.282–1.154) 0.464 (0.200–1.076) 0.477 (0.211–1.078) 0.635 (0.322–1.254) 0.303 (0.105–0.870) High 20 35 33 13 36 19 18 40 27 29 41 15 Information of radiation therapy was unavailable in unavailable was therapy of radiation Information c Low GPC5 18 31 37 16 38 25 12 33 26 30 39 17 a Adjusted P Adjusted 95%CI 0.938 0.137 0.031 0.041 0.035 0.051 0.461 0.265 0.030 0.772 0.020 0.386 Adjusted HR Adjusted 95% CI 1.061 (0.238–4.729) 0.589 0.293–1.184) 0.504 (0.271–0.938) 0.346 (0.125–0.956) 0.517 (0.280–0.954) 0.368 0.135–1.003) 0.612 (0.166–2.260) 0.668 (0.328–1.359) 0.381 (0.160–0.909) 0.880 (0.372–2.086) 0.442 (0.222–0.880) 0.644 (0.238–1.743) Information of radical resection was unavailable in 2 patients. in 2 patients. unavailable resection was of radical Information b High 18 37 33 10 38 18 19 43 25 31 39 17 GPC3 Low 20 29 37 19 36 26 11 30 28 28 41 15 a Adjusted P Adjusted 95%CI 0.202 0.103 0.020 0.079 0.007 0.024 0.046 0.059 0.124 0.002 0.010 0.180 Adjusted HR Adjusted 95% CI 0.433 (0.119–1.569) 0.532 0.249–1.136) 0.462 (0.241–0.885) 0.453 (0.187–1.095) 0.415 (0.218–0.790) 0.378 0.162–0.880) 0.258 (0.068–0.977) 0.477 (0.222–1.027) 0.481 (0.189–1.221) 0.302 (0.140–0.652) 0.374 (0.178–0.787) 0.495 (0.177–1.383) d High 22 30 28 14 34 24 19 35 23 33 41 15 therapy c b Low GPC2 16 36 42 15 40 20 11 38 30 26 39 17 Information of targeted molecular therapy was unavailable in10 patients unavailable was molecular therapy of targeted Information d Stratifed analysis of Glypican genes in The Cancer Genome Atlas database Genome Atlas Cancer of Glypican genes in The analysis Stratifed ­ therapy ­ resection + G2 + G4 + Rx ≤ 60 > 60 Adjusted for histologic grade, targeted molecular therapy, radiation therapy and radical resection. and radical therapy radiation molecular therapy, targeted grade, histologic for Adjusted

12 patients. 12 patients. 2 Table years) Age R0 No No interval CI confdence ratio, HR hazard Abbreviations: a R1 Yes Yes Sex Female Radiation molecular ­ Targeted Male Histologic grade G1 G3 Radical Liu et al. BMC Gastroenterol (2020) 20:415 Page 10 of 23

Table 3 Joint efects analysis of combination of Glypican genes in The Cancer Genome Atlas database Group No. of events MST (days) Crude HR Crude Adjusted HR Adjusted 95% CI P value 95% CI­ a P value­ a

GPC2 GPC3 + 1 21/26(80.8%) 278 1 0.001 1 < 0.001 2 37/60(61.7%) 568 0.441(0.252–0.772) 0.004 0.350(0.187–0.653) 0.001 3 11/26(42.3%) 702 0.285(0.135–0.598) 0.001 0.173(0.072–0.418) < 0.001 GPC2 GPC5 + I 18/23(78.3%) 278 1 0.001 1 < 0.001 II 40/66(60.6%) 517 0.424(0.237–0.758) 0.004 0.283(0.145–0.554) < 0.001 III 11/23(47.8%) 702 0.253(0.116–0.552) 0.001 0.141(0.057–0.353) < 0.001 GPC3 GPC5 + i 26/38(68.4%) 393 1 0.028 1 0.018 ii 21/36(58.3%) 498 0.685(0.383–1.226) 0.203 0.575(0292–1.129) 0.108 iii 22/38(57.9%) 691 0.443(0.243–0.805) 0.008 0.394(0.206–0.756) 0.005 GPC2 GPC3 GPC5 + + A 13/16(81.2%) 219 1 0.028 1 < 0.001 B 26/39(66.7%) 517 0.494(0.250–0.974) 0.042 0.446(0.212–0.938) 0.033 C 23/42(54.8%) 592 0.269(0.130–0.560) < 0.001 0.176(0.076–0.406) < 0.001 D 7/15(46.7%) 702 0.204(0.079–0.526) 0.001 0.135(0.045–0.403) < 0.001

Group 1:low GPC2 low GPC3; Group 2:low GPC2 high GPC3 or high GPC2 low GPC3; Group 3: high GPC2 high GPC3 + + + + Group I:low GPC2 low GPC5; Group II:low GPC2 high GPC5 or high GPC2 low GPC5; Group III: high GPC2 high GPC5 + + + + Group i:low GPC3 low GPC5; Group ii:low GPC3 high GPC5 or high GPC3 low GPC5; Group iii: high GPC3 high GPC5 + + + + Group A:low GPC2 low GPC3 low GPC5; Group B: high GPC2 low GPC3 low GPC5 or low GPC2 high GPC3 low GPC5 or low GPC2 low GPC3 high GPC5; + + + + + + + + Group C:high GPC2 high GPC3 low GPC5 or high GPC2 low GPC3 high GPC5 or low GPC2 high GPC3 high GPC5; Group D:high GPC2 high GPC3 high + + + + + + + + GPC5 Abbreviations: MST median survival time, HR hazard ratio, CI confdence interval a Adjusted for histologic grade, targeted molecular therapy, radiation therapy and radical resection whereas the best survival was observed in group D analysis between the two groups was carried out. Table 4 (MST = 702 days, adjusted P value < 0.001). Tese data and Fig. 9a–f show that higher expression of GPC3 are shown in Table 3 and Fig. 7a–d showed the survival was signifcantly related to better survival (log-rank curves. P = 0.038) and higher expression of GPC2 and GPC5 was also related to better survival, though not signifcantly (log-rank P 0.337 and 0.090, repectively). Multivariate Prognosis nomogram = Cox proportional hazards regression analysis adjusted Based on the status of each clinical parameter and for prognosis-related clinical characteristics showed that expression levels of GPC2, GPC3, and GPC5, a score for none of these genes was signifcantly correlated to overall each variable was calculated. Te total score could be cal- survival (all adjusted P > 0.05). culated to predict 1-, 2-, and 3- year survival probabili- ties. Te nomogram (Fig. 8) indicated that GPC2, GPC3, Genome‑wide co‑expression analysis of GPC2, GPC3 and GPC5 afected the prognosis of PDAC to diferent and GPC5 in PDAC degrees. Genome-wide co-expression analysis was performed for each of these genes to investigate their related functional Validation dataset to demonstrate the prognostic value pathways through TCGA database. For GPC2 and its co- of survival‑related genes expressed genes, a correlation network was established as To further understand the prognostic value of GPC2, shown in Fig. 10a (Additional fle 4: Table 4). GO analy- GPC3, and GPC5, we acquired the GSE62452 data- sis indicated that GPC2 and its co-expressed genes func- set from GEO database. As shown in Additional fle 3: tioned mainly in sequence-specifc DNA binding, protein Table 3, histologic grade was signifcantly associated with transport, cell diferentiation, and anterior/posterior pat- OS. GPC family genes were also divided into two groups tern specifcation (Fig. 10b, Additional fle 5: Table 5). by the median expression level of each gene and survival Liu et al. BMC Gastroenterol (2020) 20:415 Page 11 of 23

a 100 b 100 Group 1 Group I l 80 Group 2 80 Group II ival viva

Group 3 rv Group III

ur 60 60 su ts 40 40 Percent Percen 20 20 Log-rank P<0.001 Log-rank P<0.001 0 0

0200 400600 8001000120014001600 0 200 400 600 800 1000120014001600 Overall Survival(days) Overall survival (days)

c 100 d 100 Group A Group i l l Group B 80 Group ii 80 Group C viva Group iii viva

ur 60 ur 60 Group D ts ts 40 40

Percen 20 Percen 20 Log-rank P=0.024 Log-rank P<0.001 0 0

0200 400600 8001000120014001600 0 200 400 600 800 1000120014001600 Overall survival (days) Overall Survival(days) Fig. 7 Survival curves of joint efects analysis of Glypican2, Glypican3 and Glypican5 in The Cancer Genome Atlas database. a Survival curve of Glypican2 and Glypican3; b Survival curve of Glypican2 and Glypican5; c Survival curve of Glypican3 and Glypican5; d Survival curve of Glypican2, Glypican3 and Glypican5

Te correlation network for GPC3 and its co-expressed indicated that low GPC2 expression was closely related to genes (Fig. 11a, Additional fle 6: Table 6) identifed oncogenic signatures such as KRAS, RAF1, STK33, and 511 positively co-expressed genes and 25 negatively co- VEGFA (Fig. 13a–f; Additional fle 12: Table 12). GSEA expressed genes. GO analysis of these genes indicated results of c2 enrichment showed that high GPC3 expres- that they were enriched in cell adhesion, angiogenesis, sion was associated with neuroactive ligand receptor and infammatory response (Fig. 11b, Additional fle 7: interaction and GPCR ligand binding (Fig. 14a–c; Addi- Table 7). And KEGG analysis indicated that these genes tional fle 13: Table 13), and c6 enrichment suggested were related to several biological processes, mainly in that high GPC3 expression was correlated to cyclin D1, Ras, Rap1, PI3K-Akt, and chemokine signaling pathways p53, and PTEN (Fig. 13d–f; Additional fle 14: Table 14). (Fig. 11c, Additional fle 8: Table 8). For GPC5, c2 reference indicated that low expression of Te correlation network for GPC5 and its co- GPC5 was related to the EGFR pathway, gene methyla- expressed genes was shown in Fig. 12a and Additional tion status, TFRC1, and the cell cycle (Fig. 15a–d; Addi- fle 9: Table 9. Te results of GO analysis showed that tional fle 15: Table 15) and c6 reference indicated that these genes were associated with transcription factor low GPC5 expression was related to HOXA9 and BMI1 complex and phospholipid metabolic process (Fig. 12b, (Fig. 15e–f; Additional fle 16: Table 16). Additional fle 10: Table 10). KEGG analysis showed that these genes were involved in pancreatic secretion Discussion and glycerophospholipid metabolism (Fig. 12c, Addi- In this research, we studied the relationship between tional fle 11: Table 11). GPC family gene expression and prognosis of early-stage PDAC patients after pancreaticoduodenectomy both in Gene set enrichment analysis TCGA database and GEO database. We concluded that GSEA was carried out to explore possible mechanisms of high expression of GPC2, GPC3, and GPC5 was signif- GPC family genes afecting prognosis of PDAC patients cantly related to favorable prognosis in TCGA database, through TCGA database. Te results of c6 reference suggesting the value of these genes as biomarkers for predicting the prognosis of PDAC patients. Moreover, Liu et al. BMC Gastroenterol (2020) 20:415 Page 12 of 23

012345678910 Points

>60 Age (years) ≤60 Male Sex Female Yes Alcohol history No G3/G4 Histologic grade G1/G2 Stage I Pathologic stage Stage II R1+Rx Radical resection R0 Yes Radiation therapy No No Targeted molecular therapy Yes Low GPC2 High Low GPC3 High Low GPC5 High

Total Points 0510 15 20 25 30 35 40

1−year survival 0.9 0.8 0.70.6 0.4 0.20.1

2−year survival 0.8 0.70.6 0.4 0.20.1

3−year survival 0.70.6 0.4 0.20.1 Fig. 8 Prognosis nomogram for predicting 1-, 2- and 3- overall survival combination of the three genes showed a better predic- the development and prognosis of various cancers. Li tive value for PDAC prognosis. et al. demonstrated that GPC1 is enriched in exosomes GPC family genes may contribute to the malignant produced by colorectal cancer cells HT-29 and HCT- behaviors of tumors and they are closely related to 116, and increased expression level of miR-96-5p and Liu et al. BMC Gastroenterol (2020) 20:415 Page 13 of 23

Table 4 Prognostic value of Glypican family genes in Gene Expression Omnibus database Gene Samples Crude HR Crude Adjusted HR Adjusted (n 48) 95% CI P value 95% CI­ a P value­ a = GPC1 Low 24 1 1 High 24 0.888(0.455–1.735) 0.728 0.675(0.330–1.383) 0.283 GPC2 Low 24 1 1 High 24 0.717(0.362–1.420) 0.337 0.852(0.418–1.738) 0.660 GPC3 Low 24 1 1 High 24 0.468(0.225–0.973) 0.038 0.556(0.259–1.197) 0.134 GPC4 Low 24 1 High 24 1.722(0.869–3.413) 0.115 1.616(0.812–3.216) 0.172 GPC5 Low 24 1 1 High 24 0.556(0.279–1.106) 0.090 0.600(0.298–1.206) 0.151 GPC6 Low 24 1 1 High 24 1.727(0.882–3.381) 0.107 1.554(0.783–3.083) 0.207

Abbreviations: HR hazard ratio, CI confdence interval a Adjusted for histologic grade

miR-149 can restrain both GPC1 expression and cell transport, cell diferentiation and oncogenic signatures proliferation of the tumor, suggesting that GPC1 can be (KRAS, RAF, STK33, and VEGFA). In pancreatic cancer, used as a marker for diagnosis and therapy of colorectal mutation of TP53 at codon 249 can alter the structure of cancer [36]. It is reported that GPC2 could promote the p53, thus afecting its binding to a specifc region of DNA proliferation of neuroblastoma cells as a result of MYCN and enhancing the risk of cancer [46, 47]. A study showed binding to a motif of the promoter of GPC2 and gain of that GDF11 regulates the biological behaviors of pancre- 7q [37]. GPC2 can also be used an efective atic cancer cells to infuence their diferentiation and high prognostic indicator for prostate cancer and neuroblas- expression of GDF11 is associated with favorable OS in toma [37–39]. GPC3 blocks the cell cycle in renal cancer pancreatic cancer [48]. RAF1 accelerates migration and cells 786-O and ACHN at G1 phase [40]. Overexpression invasion of pancreatic cancer and disorders of the RAF1 of GPC3 reduces progression and metastasis of breast pathway are related to worse prognosis in pancreatic cancer cells LM3 through targeting canonical Wnt path- cancer patients [49, 50]. Moreover, microRNA-216a may way [41]. Te GPC5 rs2352028 variant and lower expres- downregulate RAF1 in pancreatic cancer and increase sion of this gene may contribute to increased risk of lung cell apoptosis [51]. VEGFA expression can increase as a cancer [42, 43]. Sun et al. have shown that GPC5 regu- result of the long non-coding RNA (lncRNA) 00511 in lates epithelial–mesenchymal transition to reduce inva- PDAC, which fnally promotes tumor progression. Te sion of prostate cancer cells [44]. Its expression can serve expression level of lnc00511 can be used as an indicator as a prognostic indicator in a cohort of prostate cancer of prognosis in PDAC [52]. patients in China [45]. In this study, we demonstrated the GPC3 is related to cell adhesion, angiogenesis, infam- relationship between OS and expression levels of GPC2, matory response, signaling pathways like Ras, Rap1, GPC3, and GPC5. Combined with results of GEPIA, it PI3K-Akt, chemokine, GPCR, and signatures like cyclin demonstrates their roles as tumor suppressor genes in D1, p53, PTEN. For pancreatic cancer patients, the degree PDAC. of infammatory response can be measured by serum To explore potential mechanisms of GPC genes afect- lactate dehydrogenase level and it is associated with the ing prognosis, we conducted GSEA and genome-wide co- outcome of patients [53]. Angiogenesis is dysregulated in expression analyses. Te results showed that GPC2 was PDAC, and it contributes to proliferation and deteriora- associated with sequence-specifc DNA binding, protein tion of the tumor, making survival of patients worse [54, Liu et al. BMC Gastroenterol (2020) 20:415 Page 14 of 23

a 100 b 100 Low GPC1 Low GPC2 l 80 l 80 High GPC1 High GPC2 viva viva

ur 60 ur 60 ts ts 40 40

Percen 20 Percen 20 Log-rank P=0.728 Log-rank P=0.337 0 0

01020304050607080 01020304050607080 Overall survival (months) Overall survival (months)

c 100 d 100 Low GPC3 Low GPC4 l 80 l 80 va i High GPC3 High GPC4 viva rv u 60 ur 60 ts ts 40 40 Percen 20 Percen 20 Log-rank P=0.038 Log-rank P=0.115 0 0

01020304050607080 01020304050607080 Overall survival (months) Overall survival (months)

e 100 f 100 Low GPC5 Low GPC6 l l 80 80 High GPC5 High GPC6 viva viva ur ur 60 60 s ts t 40 40 rcen e

P 20 Percen 20 Log-rank P=0.090 Log-rank P=0.107 0 0

01020304050607080 01020304050607080 Overall survival (months) Overall survival (months) Fig. 9 Kaplan-Meier survival curves of Glypican family genes for pancreatic ductal adenocarcinoma in Gene Expression Omnibus database. a-f Kaplan-Meier survival curves of Glypican1–6, respectively

55]. Certain mutations of KRAS are associated with the pathway. In pancreatic cancer, the transcription factor response to drugs in PDAC cells [56]. In PDAC associ- hif- 2α can speed up metabolism and promote tumor ated with the KRAS mutation, decitabine therapy inhibits proliferation and high level of hif- 2α correlates with tumor growth [57]. ARF6 is reported to be in close rela- worse OS [61, 62]. Te methylation status of GRAP2, tionship with the Ras pathway and its overexpression is ICAM3, A2ML1, MUC1, and MUC4 can infuence the related to unfavorable prognosis of PDAC patients [58]. expression of these genes, which is associated with sur- PTEN plays a role in pancreatic cancer growth. Te func- vival of pancreatic cancer [63, 64]. Phosphatidylserine tion of PTEN is regulated by HNF1A and fnally afects is related to apoptosis of pancreatic cancer cells with the survival of pancreatic cancer patients [59, 60]. the involvement of microparticles [65]. Stimuli such as GPC5 is involved in the transcription factor complex oxidative stress can make phosphatidylserine appear TFRC1, oncogenic signatures HOXA9 and BMI1, gene outside on the pancreatic cancer cell membrane, fnally methylation, phospholipid metabolic process, glycer- leading to dysregulation of factors and cells such as ophospholipid metabolism, cell cycle, and the EGFR VEGF and macrophages, making prognosis of patients Liu et al. BMC Gastroenterol (2020) 20:415 Page 15 of 23

a

b

Fig. 10 a Correlation network for Glypican2 and its co-expression genes in The Cancer Genome Atlas database. The pink nodes are genes correlated positively. b Function enrichment analysis of Gene Ontology for Glypican2 and its co-expression genes

unfavorable [66–68]. Te EGFR pathway contributes of the cancer as a result of lnc00976 overexpression, to pancreatic cancer growth and accelerates invasion which can deteriorate the outcome of patients [69, 70]. Liu et al. BMC Gastroenterol (2020) 20:415 Page 16 of 23

a

b

c

Fig. 11 a Correlation network for Glypican3 and its co-expression genes in The Cancer Genome Atlas database. The pink nodes are genes correlated positively and the blue nodes are genes correlated negatively. b Function enrichment analysis of Gene Ontology for Glypican3 and its co-expression genes. c Function enrichment analysis of Kyoto Encyclopedia of Genes and Genomes for Glypican3 and its co-expression genes Liu et al. BMC Gastroenterol (2020) 20:415 Page 17 of 23

a

b

c

Fig. 12 a Correlation network for Glypican5 and its co-expression genes in The Cancer Genome Atlas database. The pink nodes are genes correlated positively. b Function enrichment analysis of Gene Ontology for Glypican5 and its co-expression genes. c Function enrichment analysis of Kyoto Encyclopedia of Genes and Genomes for Glypican5 and its co-expression genes Liu et al. BMC Gastroenterol (2020) 20:415 Page 18 of 23

ab c

de f

Fig. 13 Gene Set Enrichment Analysis (GSEA) results of Glypican2 in The Cancer Genome Atlas database. a-f GSEA results of c6 reference for the group of low Glypican2 expression. NES, normalized enrichment score; FDR, false discovery rate

Te present study had several limitations. First, clini- Despite these limitations, we identifed GPC2, GPC3, cal data acquired from TCGA and GEO databases did and GPC5 as biomarkers for prognosis of PDAC not include all the relevant information, and there may patients and showed that joint efects analysis was more be some factors that needed to be adjusted. Second, efective for prediction of prognosis. We also explored because the study included PDAC patients who under- possible mechanisms of survival-signifcant genes went pancreaticoduodenectomy, the sample size was afecting PDAC prognosis through genome-wide analy- relatively small. Tird, the results of genome-wide anal- sis and GSEA analysis. Tese results could all improve ysis and GSEA analysis were based on online databases prognostic prediction for PDAC and provide informa- to predict potential processes infuencing prognosis, tion valuable for the management of PDAC patients and further studies at molecular and genomic levels are and making better clinical decisions in this population. necessary to confrm the results. Liu et al. BMC Gastroenterol (2020) 20:415 Page 19 of 23

ab c

de f

Fig. 14 Gene Set Enrichment Analysis (GSEA) results of Glypican3 in The Cancer Genome Atlas database. a-c GSEA results of c2 reference for the group of high Glypican3 expression; d-f GSEA results of c6 reference for the group of high Glypican3 expression. NES, normalized enrichment score; FDR, false discovery rate

Conclusions angiogenesis, infammatory response, signaling path- We identifed GPC2, GPC3, and GPC5 as potential ways such as Ras, Rap1, PI3K-Akt, chemokine, and prognostic indicators for PDAC patients and showed GPCR, and signatures including cyclin D1, p53, and that combination of these genes was more efec- PTEN. GPC5 may be involved in the transcription fac- tive for prognosis prediction. Possible mechanisms of tor complex TFRC1, the oncogenic signatures HOXA9 GPC2 infuencing prognosis may involve sequence- and BMI1, gene methylation, the phospholipid meta- specifc DNA binding, protein transport, cell diferen- bolic process, glycerophospholipid metabolism, cell tiation and oncogenic signatures (KRAS, RAF, STK33, cycle, and the EGFR pathway. and VEGFA). GPC3 may be related to cell adhesion, Supplementary Information The online version contains supplementary material available at https​://doi. org/10.1186/s1287​6-020-01560​-0.

Additional fle 1: Table 1. Gene Ontology terms of Glypican family genes. Additional fle 2: Table 2. Basic characteristics of pancreatic ductal adenocarcinoma patients in The Cancer Genome Atlas database. Additional fle 3: Table 3. Basic characteristics of PDAC patients in Gene Expression Omnibus database. Liu et al. BMC Gastroenterol (2020) 20:415 Page 20 of 23

ab c

de f

Fig. 15 Gene Set Enrichment Analysis (GSEA) results of Glypican5 in The Cancer Genome Atlas database. a-d GSEA results of c2 reference for the group of low Glypican 5 expression; e-f GSEA results of c6 reference for the group of low GPC5 expression. NES, normalized enrichment score; FDR, false discovery rate

Additional fle 4: Table 4. Genome-wide co-expression genes of Glypi- Additional fle 12: Table 12. Gene Set Enrichment Analysis results of c6 can2 in pancreatic ductal adenocarcinoma in The Cancer Genome Atlas enrichment for low Glypican2 expression in The Cancer Genome Atlas database. database. Additional fle 5: Table 5. Gene Ontology terms of Glypican2 and its co- Additional fle 13: Table 13. Gene Set Enrichment Analysis results of c2 expression genes in The Cancer Genome Atlas database. enrichment for high Glypican3 expression in The Cancer Genome Atlas database. Additional fle 6: Table 6. Genome-wide co-expression genes of Glypi- can3 in pancreatic ductal adenocarcinoma in The Cancer Genome Atlas Additional fle 14: Table 14. Gene Set Enrichment Analysis results of c6 database. enrichment for high Glypican3 expression in The Cancer Genome Atlas database. Additional fle 7: Table 7. Gene Ontology terms of Glypican3 and its co- expression genes in The Cancer Genome Atlas database. Additional fle 15: Table 15. Gene Set Enrichment Analysis results of c2 enrichment for low Glypican5 expression in The Cancer Genome Atlas Additional fle 8: Table 8. Kyoto Encyclopedia of Genes and Genomes database. pathways of Glypican3 and its co-expression genes in The Cancer Genome Atlas database. Additional fle 16: Table 16. Gene Set Enrichment Analysis results of c6 enrichment for low Glypican5 expression in The Cancer Genome Atlas Additional fle 9: Table 9. Genome-wide co-expression genes of Glypi- database. can5 in pancreatic ductal adenocarcinoma in The Cancer Genome Atlas database. Additional fle 10: Table 10. Gene Ontology terms of Glypican5 and its Acknowledgements co-expression genes in The Cancer Genome Atlas database. We would like to sincerely thank the contributors of The Cancer Genome Atlas and Gene Expression Omnibus for sharing the data with the public. Additional fle 11: Table 11. Kyoto Encyclopedia of Genes and Genomes pathways of Glypican5 and its co-expression genes in The Cancer Genome Authors’ contributions Atlas database. JL, XL, XY and TP designed this study. JL, WX, CY, XZ, ZL, QH, and TF analyzed the data. JL, GZ, CH, HS, JH, GR, and LY prepared fgures and Tables. JL, XL, XW Liu et al. BMC Gastroenterol (2020) 20:415 Page 21 of 23

and XZ fnished the manuscript. XY and TP guided and supervised the whole 5. Zeng H, Zheng R, Guo Y, Zhang S, Zou X, Wang N, et al. Cancer research and TP revised the fnal manuscript. All authors read and approved survival in China, 2003-2005: a population-based study. Int J Cancer. the fnal manuscript. 2015;136(8):1921–30. 6. Zhang L, Sanagapalli S, Stoita A. Challenges in diagnosis of pancreatic Funding cancer. World J Gastroenterol. 2018;24(19):2047–60. This work was supported in part by the National Natural Science Founda‑ 7. Vincent A, Herman J, Schulick R, Hruban RH, Goggins M. Pancreatic tion of China (No.: 81560535, 81802874, 81072321, 30760243, 30460143 and cancer. Lancet (London, England). 2011;378(9791):607–20. 30560133), Natural Science Foundation of Guangxi Province of China (Grant 8. Li N, Gao W, Zhang YF, Ho M. Glypicans as Cancer therapeutic targets. No.2017JJB140189y, 2018GXNSFAA050119), 2009 Program for New Century Trends Cancer. 2018;4(11):741–54. Excellent Talents in University (NCET), Guangxi Natural Sciences Foundation 9. Kaur SP, Cummings BS. Role of glypicans in regulation of the tumor (No.: GuiKeGong 1104003A-7), and Guangxi Health Ministry Medicine Grant microenvironment and cancer progression. Biochem Pharmacol. (Key-Scientifc Research-Grant Z201018). The present study was also partly 2019;168:108–18. supported by Scientifc Research Fund of the Health and Family Planning 10. Filmus J, Capurro M, Rast J. Glypicans. Genome Biol. 2008;9(5):224. Commission of Guangxi Zhuang Autonomous Region (Z2016318, Z2016307), 11. Kayed H, Kleef J, Keleg S, Jiang X, Penzel R, Giese T, et al. Correlation the Guangxi Key R & D Program (GKEAB18221019), The Basic Ability Improve‑ of glypican-1 expression with TGF-beta, BMP, and activin receptors in ment Project for Middle-aged and Young Teachers in Colleges and Universities pancreatic ductal adenocarcinoma. Int J Oncol. 2006;29(5):1139–48. in Guangxi (2018KY0110), Innovation Project of Guangxi Graduate Education 12. Hara H, Takahashi T, Serada S, Fujimoto M, Ohkawara T, Nakatsuka R, (JGY2018037), and 2018 Innovation Project of Guangxi Graduate Education et al. Overexpression of glypican-1 implicates poor prognosis and their (YCBZ2018036). In addition, the present study was also partly supported by chemoresistance in oesophageal squamous cell carcinoma. Br J Cancer. Guangxi Key Laboratory for the Prevention and Control of Viral Hepatitis (No. 2016;115(1):66–75. GXCDCKL201902) and Research Institute of Innovative Think-tank in Guangxi 13. Suhovskih AV, Mostovich LA, Kunin IS, Boboev MM, Nepomnyashchikh GI, Medical University (The gene-environment interaction in hepatocarcinogen‑ Aidagulova SV, et al. Proteoglycan expression in normal human prostate esis in Guangxi HCCs and its translational applications in the HCC prevention). tissue and prostate cancer. ISRN Oncol. 2013;2013:680136. We would also like to acknowledge the support from the Key laboratory of 14. Li J, Chen Y, Zhan C, Zhu J, Weng S, Dong L, et al. Glypican-1 promotes High-Incidence-Tumor Prevention & Treatment (Guangxi Medical University), tumorigenesis by regulating the PTEN/Akt/beta-catenin signal‑ Ministry of Education (GKE2018–01, GKE2019–11). ing pathway in esophageal squamous cell carcinoma. Dig Dis Sci. 2019;64(6):1493–502. Availability of data and materials 15. Zhou S, O’Gorman MR, Yang F, Andresen K, Wang L. as a serum Datasets generated and analyzed during the current study are available from marker for Hepatoblastoma. Sci Rep. 2017;7:45932. The Cancer Genome Atlas, https​://porta​l.gdc.cance​r.gov and Gene Expression 16. Tangkijvanich P, Chanmee T, Komtong S, Mahachai V, Wisedopas N, Omnibus, https​://www.ncbi.nlm.nih.gov/geo/query​/acc.cgi?acc GSE62​452. Pothacharoen P, et al. Diagnostic role of serum glypican-3 in diferentiat‑ = ing hepatocellular carcinoma from non-malignant chronic disease Ethics approval and consent to participate and other liver cancers. J Gastroenterol Hepatol. 2010;25(1):129–37. Not applicable. 17. Jia X, Liu J, Gao Y, Huang Y, Du Z. Diagnosis accuracy of serum glypican-3 in patients with hepatocellular carcinoma: a systematic review with Consent for publication meta-analysis. Arch Med Res. 2014;45(7):580–8. Not applicable. 18. Sajorda BJ, Gonzalez-Gandolf CX, Hathaway ER, Kalish JM. Simpson- Golabi-Behmel Syndrome Type 1. In: Adam MP, Ardinger HH, Pagon RA, Competing interests Wallace SE, Bean LJH, Stephens K, et al., editors. GeneReviews((R)). Seattle The authors declare that they have no competing interests. (WA): University of Washington, Seattle. University of Washington, Seattle. GeneReviews is a registered trademark of the University of Washington, Author details Seattle. All rights reserved.; 1993. 1 Department of Hepatobiliary Surgery, The First Afliated Hospital of Guangxi 19. Veugelers M, Cat BD, Muyldermans SY, Reekmans G, Delande N, Frints Medical University, Shuang Yong Rd. 6#, Nanning 530021, Guangxi Zhuang S, et al. Mutational analysis of the GPC3/GPC4 glypican gene cluster on Autonomous Region, People’s Republic of China. 2 Department of Hepato‑ Xq26 in patients with Simpson-Golabi-Behmel syndrome: identifca‑ biliary Surgery, The Third Afliated Hospital of Guangxi Medical University, tion of loss-of-function mutations in the GPC3 gene. Hum Mol Genet. Nanning 530031, Guangxi Zhuang Autonomous Region, People’s Republic 2000;9(9):1321–8. of China. 3 Department of Colorectal and Anal Surgery, The First Afliated 20. Yu W, Inoue J, Imoto I, Matsuo Y, Karpas A, Inazawa J. GPC5 is a possible Hospital of Guangxi Medical University, Nanning 530021, Guangxi Zhuang target for the 13q31-q32 amplifcation detected in lymphoma cell Autonomous Region, People’s Republic of China. lines. J Hum Genet. 2003;48(6):331–5. 21. Li F, Shi W, Capurro M, Filmus J. Glypican-5 stimulates rhabdomyosar‑ Received: 23 July 2020 Accepted: 24 November 2020 coma cell proliferation by activating hedgehog signaling. J Cell Biol. 2011;192(4):691–704. 22. Okamoto K, Tokunaga K, Doi K, Fujita T, Suzuki H, Katoh T, et al. Common variation in GPC5 is associated with acquired nephrotic syndrome. Nat Genet. 2011;43(5):459–63. References 23. Seuferlein T, Bachet JB, Van Cutsem E, Rougier P. Pancreatic adenocar‑ 1. Kamisawa T, Wood LD, Itoi T, Takaori K. Pancreatic cancer. Lancet (London, cinoma: ESMO-ESDO Clinical Practice Guidelines for diagnosis, treat‑ England). 2016;388(10039):73–85. ment and follow-up. Annals Oncol. 2012;23(Suppl 7):vii33–40. 2. Rahib L, Smith BD, Aizenberg R, Rosenzweig AB, Fleshman JM, Matrisian 24. Anders S, McCarthy DJ, Chen Y, Okoniewski M, Smyth GK, Huber W, LM. Projecting cancer incidence and deaths to 2030: the unexpected et al. Count-based diferential expression analysis of RNA sequencing burden of thyroid, liver, and pancreas cancers in the United States. Cancer data using R and bioconductor. Nat Protoc. 2013;8(9):1765–86. Res. 2014;74(11):2913–21. 25. Liao X, Huang K, Huang R, Liu X, Han C, Yu L, et al. Genome-scale 3. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2019. CA Cancer J Clin. analysis to identify prognostic markers in patients with early-stage 2019;69(1):7–34. pancreatic ductal adenocarcinoma after pancreaticoduodenectomy. 4. Chen H, Kong Y, Yao Q, Zhang X, Fu Y, Li J, et al. Three hypomethylated OncoTargets Therapy. 2017;10:4493–506. genes were associated with poor overall survival in pancreatic cancer 26. The Genotype-Tissue Expression (GTEx) project. Nat Genet. patients. Aging. 2019;11(3):885–97. 2013;45(6):580–5. 27. Human genomics. The genotype-tissue expression (GTEx) pilot analysis: multitissue gene regulation in humans. Science (New York, NY). 2015;348(6235):648–60. Liu et al. BMC Gastroenterol (2020) 20:415 Page 22 of 23

28. Tang Z, Li C, Kang B, Gao G, Li C, Zhang Z. GEPIA: a web server for cancer 49. Qi ZH, Xu HX, Zhang SR, Xu JZ, Li S, Gao HL, et al. RIPK4/PEBP1 axis pro‑ and normal gene expression profling and interactive analyses. Nucleic motes pancreatic cancer cell migration and invasion by activating RAF1/ Acids Res. 2017;45(W1):W98–w102. MEK/ERK signaling. Int J Oncol. 2018;52(4):1105–16. 29. Dennis G Jr, Sherman BT, Hosack DA, Yang J, Gao W, Lane HC, et al. DAVID: 50. Veit C, Genze F, Menke A, Hoefert S, Gress TM, Gierschik P, et al. Activa‑ database for annotation, visualization, and integrated discovery. Genome tion of phosphatidylinositol 3-kinase and extracellular signal-regulated Biol. 2003;4(5):P3. kinase is required for glial cell line-derived neurotrophic factor-induced 30. Said-Salman IH, Jebaii FA, Yusef HH, Moustafa ME. Global gene expression migration and invasion of pancreatic carcinoma cells. Cancer Res. analysis of Escherichia coli K-12 DH5alpha after exposure to 2.4 GHz wire‑ 2004;64(15):5291–300. less fdelity radiation. Sci Rep. 2019;9(1):14425. 51. Zhang Y, Tang X, Shi M, Wen C, Shen B. MiR-216a decreases MALAT1 31. Maere S, Heymans K, Kuiper M. BiNGO: a Cytoscape plugin to assess expression, induces G2/M arrest and apoptosis in pancreatic cancer cells. overrepresentation of gene ontology categories in biological networks. Biochem Biophys Res Commun. 2017;483(2):816–22. Bioinformatics (Oxford, England). 2005;21(16):3448–9. 52. Zhao X, Liu Y, Li Z, Zheng S, Wang Z, Li W, et al. Linc00511 acts as a 32. Munoz-San Martin M, Reverter G, Robles-Cedeno R, Buxo M, Ortega FJ, competing endogenous RNA to regulate VEGFA expression through Gomez I, et al. Analysis of miRNA signatures in CSF identifes upregulation sponging hsa-miR-29b-3p in pancreatic ductal adenocarcinoma. J Cell of miR-21 and miR-146a/b in patients with multiple sclerosis and active Mol Med. 2018;22(1):655–67. lesions. J Neuroinfammation. 2019;16(1):220. 53. Yu SL, Xu LT, Qi Q, Geng YW, Chen H, Meng ZQ, et al. Serum lactate dehy‑ 33. Mandelboum S, Manber Z, Elroy-Stein O, Elkon R. Recurrent functional drogenase predicts prognosis and correlates with systemic infammatory misinterpretation of RNA-seq data caused by sample-specifc gene response in patients with advanced pancreatic cancer after gemcitabine- length bias. PLoS Biol. 2019;17(11):e3000481. based chemotherapy. Sci Rep. 2017;7:45194. 34. Li Y, Xue Z, Dong X, Liu Q, Liu Z, Li H, et al. Mitochondrial dysfunction 54. Li S, Xu HX, Wu CT, Wang WQ, Jin W, Gao HL, et al. Angiogenesis in and oxidative stress in bone marrow stromal cells induced by daunoru‑ pancreatic cancer: current research status and clinical implications. bicin leads to DNA damage in hematopoietic cells. Free Radic Biol Med. Angiogenesis. 2019;22(1):15–36. 2020;146:211–21. 55. Zhang Z, Ji S, Zhang B, Liu J, Qin Y, Xu J, et al. Role of angiogenesis in 35. Frost HR. Computation and application of tissue-specifc gene set pancreatic cancer biology and therapy. Biomed Pharmacotherapy. weights. Bioinformatics (Oxford, England). 2018;34(17):2957–64. 2018;108:1135–40. 36. Li J, Chen Y, Guo X, Zhou L, Jia Z, Peng Z, et al. GPC1 exosome and its 56. Hobbs GA, Baker NM, Miermont AM, Thurman RD, Pierobon M, Tran TH, regulatory miRNAs are specifc markers for the detection and target et al. Atypical KRASG12R mutant is impaired in PI3K signaling and macro‑ therapy of colorectal cancer. J Cell Mol Med. 2017;21(5):838–47. pinocytosis in pancreatic cancer. Cancer Discovery. 2020;10(1):104–23. 37. Bosse KR, Raman P, Zhu Z, Lane M, Martinez D, Heitzeneder S, et al. Iden‑ 57. Mottini C, Tomihara H, Carrella D, Lamolinara A, Iezzi M, Huang JK, et al. tifcation of GPC2 as an Oncoprotein and candidate immunotherapeutic Predictive signatures inform the efective repurposing of Decitabine to target in high-risk neuroblastoma. Cancer Cell. 2017;32(3):295–309.e12. treat KRAS-dependent pancreatic ductal adenocarcinoma. Cancer Res. 38. Li N, Fu H, Hewitt SM, Dimitrov DS, Ho M. Therapeutically targeting 2019;79(21):5612–25. glypican-2 via single-domain antibody-based chimeric antigen recep‑ 58. Hashimoto S, Furukawa S, Hashimoto A, Tsutaho A, Fukao A, Sakamura Y, tors and immunotoxins in neuroblastoma. Proc Natl Acad Sci U S A. et al. ARF6 and AMAP1 are major targets of KRAS and TP53 mutations to 2017;114(32):E6623–e31. promote invasion, PD-L1 dynamics, and immune evasion of pancreatic 39. Xu N, Wu YP, Yin HB, Xue XY, Gou X. Molecular network-based identifca‑ cancer. Proc Natl Acad Sci U S A. 2019;116(35):17450–9. tion of competing endogenous RNAs and mRNA signatures that predict 59. Yu Y, Liang S, Zhou Y, Li S, Li Y, Liao W. HNF1A/CASC2 regulates pancreatic survival in prostate cancer. J Transl Med. 2018;16(1):274. cancer cell proliferation through PTEN/Akt signaling. J Cell Biochem. 40. Valsechi MC, Oliveira AB, Conceicao AL, Stuqui B, Candido NM, Provazzi 2019;120(3):2816–27. PJ, et al. GPC3 reduces cell proliferation in renal carcinoma cell lines. BMC 60. Zhang H, Feng X, Zhang M, Liu A, Tian L, Bo W, et al. Long non-coding Cancer. 2014;14:631. RNA CASC2 upregulates PTEN to suppress pancreatic carcinoma cell 41. Fernandez D, Guereno M, Lago Huvelle MA, Cercato M, Peters MG. Signal‑ metastasis by downregulating miR-21. Cancer Cell Int. 2019;19:18. ing network involved in the GPC3-induced inhibition of breast cancer 61. Zhang Q, Lou Y, Zhang J, Fu Q, Wei T, Sun X, et al. Hypoxia-inducible progression: role of canonical Wnt pathway. J Cancer Res Clin Oncol. factor-2alpha promotes tumor progression and has crosstalk with Wnt/ 2018;144(12):2399–418. beta-catenin signaling in pancreatic cancer. Mol Cancer. 2017;16(1):119. 42. Zheng Y, Kan M, Yu L, Niu X, Zhou D, He L, et al. GPC5 rs2352028 62. Li W, Chen C, Zhao X, Ye H, Zhao Y, Fu Z, et al. HIF-2alpha regulates polymorphism and risk of lung cancer in Han Chinese. Cancer Investig. non-canonical glutamine metabolism via activation of PI3K/mTORC2 2012;30(1):13–9. pathway in human pancreatic ductal adenocarcinoma. J Cell Mol Med. 43. Li Y, Sheu CC, Ye Y, de Andrade M, Wang L, Chang SC, et al. Genetic vari‑ 2017;21(11):2896–908. ants and risk of lung cancer in never smokers: a genome-wide associa‑ 63. Kong L, Liu P, Zheng M, Xue B, Liang K, Tan X. Multi-omics analysis based tion study. Lancet Oncol. 2010;11(4):321–30. on integrated genomics, epigenomics and transcriptomics in pancreatic 44. Sun Y, Xu K, He M, Fan G, Lu H. Overexpression of Glypican 5 (GPC5) cancer. Epigenomics. 2020. inhibits prostate Cancer cell proliferation and invasion via suppressing 64. Yokoyama S, Hamada T, Higashi M, Matsuo K, Maemura K, Kurahara H, Sp1-mediated EMT and activation of Wnt/beta-catenin signaling. Oncol et al. Predicted prognosis of patients with pancreatic cancer by machine Res. 2018;26(4):565–72. learning. Clin Cancer Res. 2020;26(10):2411–21. 45. Zhang C, Liu Z, Wang L, Qiao B, Du E, Li L, et al. Prognostic signifcance 65. Schneider J, Chromik AM, Uhl W, Mugge A, Bulut D. Apoptosis in esopha‑ of GPC5 expression in patients with prostate cancer. Tumour Biol. gus and pancreas carcinoma cells induced by circulating microparticles 2016;37(5):6413–8. is related to phosphatidyl serine and microparticle-associated caspases. 46. Mohamadkhani A, Naderi E, Sharafkhah M, Fazli HR, Moradzadeh M, Medical Oncol. 2012;29(2):962–9. Pourshams A. Detection of TP53 R249 mutation in Iranian patients with 66. Sharma B, Kanwar SS. Phosphatidylserine: a cancer cell targeting bio‑ pancreatic Cancer. J Oncol. 2013;2013:738915. marker. Semin Cancer Biol. 2018;52(Pt 1):17–25. 47. Subekti DRG, Murata A, Itoh Y, Fukuchi S, Takahashi H, Kanbayashi S, et al. 67. Wang H, Zhan M, Yang R, Shi Y, Liu Q, Wang J. Elevated expression of The disordered linker in p53 participates in nonspecifc binding to and NFE2L3 predicts the poor prognosis of pancreatic cancer patients. Cell one-dimensional sliding along DNA revealed by single-molecule fuores‑ Cycle (Georgetown, Tex). 2018;17(17):2164–74. cence measurements. Biochemistry. 2017;56(32):4134–44. 68. Di Caro G, Cortese N, Castino GF, Grizzi F, Gavazzi F, Ridolf C, et al. Dual 48. Liu Y, Shao L, Chen K, Wang Z, Wang J, Jing W, et al. GDF11 restrains prognostic signifcance of tumour-associated macrophages in human tumor growth by promoting apoptosis in pancreatic cancer. OncoTargets pancreatic adenocarcinoma treated or untreated with chemotherapy. Therapy. 2018;11:8371–9. Gut. 2016;65(10):1710–20. 69. Lei S, He Z, Chen T, Guo X, Zeng Z, Shen Y, et al. Long noncoding RNA 00976 promotes pancreatic cancer progression through OTUD7B by Liu et al. BMC Gastroenterol (2020) 20:415 Page 23 of 23

sponging miR-137 involving EGFR/MAPK pathway. J Experimental Clin Publisher’s Note Cancer Res. 2019;38(1):470. Springer Nature remains neutral with regard to jurisdictional claims in pub‑ 70. Kim J, Jo YH, Jang M, Nguyen NNY, Yun HR, Ko SH, et al. PAC-5 gene lished maps and institutional afliations. expression signature for predicting prognosis of patients with pancreatic adenocarcinoma. Cancers. 2019;11(11):1749.

Ready to submit your research ? Choose BMC and benefit from:

• fast, convenient online submission • thorough peer review by experienced researchers in your field • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations • maximum visibility for your research: over 100M website views per year

At BMC, research is always in progress.

Learn more biomedcentral.com/submissions