Detection of early pancreatic ductal adenocarcinoma with ... · pluripotent stem cell (iPSC)–like cell line (25). This iPSC-like cell line (designated as 10-22) can be propagated

SC I ENCE TRANS LAT IONAL MED I C I N E | R E S EARCH ART I C L E

CANCER

1Institute for Regenerative Medicine, Department of Cell and Developmental Biology,Abramson Cancer Center (Tumor Biology Program), Perelman School of Medicine,University of Pennsylvania, 9-131 Smilow Center for Translational Research, 3400Civic Center Boulevard, Philadelphia, PA 19104–5157, USA. 2Division of BiomedicalStatistics and Informatics, Department of Health Sciences Research, Mayo Clinic,Rochester, MN 55905, USA. 3Epigenetics Program, Department of Biochemistryand Biophysics, Perelman School of Medicine, University of Pennsylvania, Philadelphia,PA 19104, USA. 4Division of Gastroenterology and Hepatology, Department of Medicine,Mayo Clinic, Rochester, MN 55905, USA. 5Division of Epidemiology, Department ofHealth Sciences Research, Mayo Clinic, Rochester, MN 55905, USA.*Corresponding author. Email: [email protected]

Kim et al., Sci. Transl. Med. 9, eaah5583 (2017) 12 July 2017

Copyright © 2017

The Authors, some

rights reserved;

exclusive licensee

American Association

for the Advancement

of Science. No claim

to original U.S.

Government Works

hD

ownloaded from

Detection of early pancreatic ductal adenocarcinomawith thrombospondin-2 and CA19-9 blood markersJungsun Kim,1 William R. Bamlet,2 Ann L. Oberg,2 Kari G. Chaffee,2 Greg Donahue,1 Xing-Jun Cao,3

Suresh Chari,4 Benjamin A. Garcia,3 Gloria M. Petersen,5 Kenneth S. Zaret1*

Markers are needed to facilitate early detection of pancreatic ductal adenocarcinoma (PDAC), which is often di-agnosed too late for effective therapy. Starting with a PDAC cell reprogramming model that recapitulated theprogression of human PDAC, we identified secreted proteins and tested a subset as potential markers of PDAC.We optimized an enzyme-linked immunosorbent assay (ELISA) using plasma samples from patients with variousstages of PDAC, from individuals with benign pancreatic disease, and from healthy controls. A phase 1 discoverystudy (n = 20), a phase 2a validation study (n = 189), and a second phase 2b validation study (n = 537) revealedthat concentrations of plasma thrombospondin-2 (THBS2) discriminated among all stages of PDAC consistently.The receiver operating characteristic (ROC) c-statistic was 0.76 in the phase 1 study, 0.84 in the phase 2a study,and 0.87 in the phase 2b study. The plasma concentration of THBS2 was able to discriminate resectable stage Icancer as readily as stage III/IV PDAC tumors. THBS2 plasma concentrations combined with those for CA19-9, apreviously identified PDAC marker, yielded a c-statistic of 0.96 in the phase 2a study and 0.97 in the phase 2bstudy. THBS2 data improved the ability of CA19-9 to distinguish PDAC from pancreatitis. With a specificity of 98%,the combination of THBS2 and CA19-9 yielded a sensitivity of 87% for PDAC in the phase 2b study. A THBS2 andCA19-9 blood marker panel measured with a conventional ELISA may improve the detection of patients at highrisk for PDAC.

ttp:/
by guest on M
ay 27, 2020/stm

.sciencemag.org/

INTRODUCTIONPancreatic ductal adenocarcinoma (PDAC) is projected to become thesecond leading cause of cancer death in the United States by 2020 (1).Most of the PDAC patients are diagnosed at an advanced stage of dis-ease, and their tumors are not surgically resectable, contributing to anoverall 5-year survival rate of 7% (2). The lack of early diagnostics hasmade it challenging to develop therapeutics to slow or reverse PDAC(3). The CA19-9 serum marker is used to assess disease progression inPDAC patients (4, 5) but is not recommended for general screening(5, 6) because it is elevated in nonmalignant pancreatic conditions, suchas chronic pancreatitis (7), and can produce false negatives in individ-uals who do not express Lewis blood group antigens (8). Other se-creted markers have been reported for PDAC (9–12) including bloodor urine proteins (13–15), exosomes (11), microRNAs (16), and epige-netic marks in circulating nucleosomes (17). However, challenges in-clude lack of translation to the clinic, small sample sizes precludingstatistical robustness, lack of blinded design, or inappropriate construc-tion of data sets from development to validation (15–19). Most bio-markers were discovered in advanced PDAC or cell lines that are notrepresentative of earlier stages, when detection would be most relevant,although recent candidates have been tested or discovered in prediag-nostic samples of PDAC (20–22). When agnostic biomarker panels areassessed in validation samples, the need to aggregate samples frommultiple sources hampers achieving statistical power (23).

We reasoned that proteins released from progressing precursor le-sions, such as pancreatic intraepithelial neoplasia (PanIN) stages 2 and3 (PanIN2 and PanIN3) (24), might provide an effective opportunityfor discovering diagnostic markers for PDAC. We previously repro-grammed recurrent, advanced human PDAC cells into an inducedpluripotent stem cell (iPSC)–like cell line (25). This iPSC-like cell line(designated as 10-22) can be propagated indefinitely yet preferentiallygenerates PanIN2 and PanIN3 ductal lesions after growing for 3months as teratomas in immunodeficient mice. The lesions progressto invasive PDAC by 6 to 9 months. Proteomic analysis of conditionedmedia from 10-22 cell–derived precursor PanINs cultured as organoids,compared to control media from 10-22 cells grown under pluripo-tency conditions, revealed 107 secreted human proteins specific to thePanIN2/3 organoids (25). Of these, 43 proteins fell into interconnectedtransforming growth factor–b (TGFb) and integrin networks for PDACprogression (26, 27), and 25 proteins were within a network for thetranscription factor hepatocyte nuclear factor 4a (HNF4a), which alsoshowed an increase in expression (25). Here, we report an analysis ofproteins secreted or released from the 10-22 cell–derived PanIN orga-noids using a phased cancer marker development design thatincorporated criteria for prospective specimen collection and retro-spective blinded evaluation (PRoBE) (28, 29).

RESULTSDiscovery of candidate marker proteins for PDACOf the 107 proteins secreted and released selectively by human PanINorganoids (25), we focused on 53 proteins with a low abundance(≤2 nmol) in the healthy human plasma proteome and RNA sequencing(RNA-seq) databases (table S1) (30–32). Enzyme-linked immuno-sorbent assays (ELISAs) from validated sources (33) were not availablefor most of these rarely expressed proteins. Of the proteins for whichreliable ELISA kits were available and were not implicated as markersin other diseases, we focused on matrix metalloproteinase 2 (MMP2),

1 of 13

http://stm.sciencemag.org/


MMP10, and thrombospondin-2 (THBS2) because they occur inintegrated networks for TGFb and integrin signaling, which drivePDAC progression (25). We investigated these three candidate markersin a screen of human plasma samples. All procedures were performedusing a recommended biomarker phased design following the PRoBEcriteria (28, 29). De-identified human plasma samples were providedby the Mayo Clinic pancreas research biospecimen repository. We thenperformed ELISA analyses blinded to disease status and then re-turned coded data to the Mayo Clinic team for statistical analysis andinterpretation.


Phase 1 validation of candidate markers for PDACWe examined whether MMP2, MMP10, or THBS2 could discriminatebetween cancer cases (n = 10) and controls (n = 10) with an area un-der the receiver operating characteristic (ROC) curve (AUC) analysisof the sensitivity and specificity of the markers. All cancer cases for thephase 1 study were selected to have CA19-9 concentrations equal to orabove 55 U/ml. MMP2 was unable to discriminate effectively betweencancer cases and controls, and MMP10 signals were undetectable in allplasma samples (Fig. 1A). By contrast, THBS2 exhibited a c-statistic of0.76 considering all cases versus controls (n = 10) and 0.886 when

by guest on May 27, 2020

http://stm.sciencem

ag.org/D

ownloaded from

considering resectable and locally ad-vanced PDAC (n = 7). Given that humanTHBS2 has 80% amino acid sequencehomology with THBS1, we demonstratedthe specificity of each of the reagents inthe THBS2 ELISA (figs. S1 and S2).

After the phase 1 validation analysis,we performed a mass spectrometry studyof the pooled cancer plasma samples (n =10) and the pooled control plasma samples(n = 10) after the plasma samples were in-dividually depleted of the 14 most abun-dant plasma proteins (for example,serum albumin). At 5% false discovery rate(FDR), four unique peptides for THBS2were identified, of which two were fromsequences specific to THBS2 and the othertwo were from sequences that are con-served between THBS1 and THBS2. Oneof two peptides specific to THBS2 waspresent at a threefold greater concentrationin the cancer plasma pool compared to thecontrol plasma pool, and another peptidespecific to THBS2 was detected only in thecancer pool and not in the control pool(table S2, A and B). At 1% FDR, a THBS2-specific peptide was detectable only in thecancer pool (table S2C). Computationalanalysis of RNA expression data depositedin The Cancer Genome Atlas (TCGA)database (https://cancergenome.nih.gov/)showed that, of all cancers tested, PDAC(n = 134) was second only to mesothelio-ma for expression of THBS2 mRNA (Fig.1B; medians are denoted by vertical blackbars within the red boxes). Taking togetherthe phase 1 validation study by ELISA,mass spectrometry data, and TCGA RNA-seq data, we concluded that THBS2 meritedfurther study.

Phase 2a validation of THBS2 andCA19-9 markers for PDACFurther validation of THBS2 was per-formed on human plasma samples in aphase 2a study (Table 1) that containedCA19-9–negative and CA19-9–positivecases. Themedian ELISA value for THBS2at all PDAC stages (n = 81) in the phase

log(THBS2 mRNA expression)

Mesothelioma (87)Pancreatic ductal adenocarcinoma (134)

Sarcoma (263)Breast (1097)

Lung squamous (501)Head & neck (522)

Lung adenocarcinoma (517)Cholangiocarcinoma (36)

Stomach & esophageal (600)Bladder urothelial (404)

Rectum adenocarcinoma (95)Colorectal adencarcinoma (279)

Kidney papillary (291)Brain lower grade glioma (529)

Kidney clear cell carcinoma (534)Cutaneous melanoma (472)

Glioma (166)Uterine carcinoma (57)

Ovarian cystadenocarcinoma (308)Cervical squamous, endocerivcal (304)

Testicular germ cell (139)Large B lymphoma (48)

Hepatocellular carcinoma (373)Pheochromocytoma (184)

Prostate (498)Thyroid (509)

Adrenocortical (79)Uterine corpus endometrial carc. (177)

Kidney (66)Thymoma (120)

Uveal melanoma (80)Acute myeloid leukemia (173)

5 10

All stages (N = 10) Stage I/II/III (N = 7) Stage I/II (N = 6)AUC 95% CI AUC 95% CI AUC 95% CI

MMP2 0.57 0.43 0.81 0.557 0.45 0.80 0.533 0.43 0.80MMP10 No signal No signal No signalTHBS2 0.76 0.56 0.95 0.886 0.71 1.00 0.867 0.64 1.00

A

B

15

Fig. 1. Phase 1 validation studies and THBS2 expression in PDAC and other human tumors. (A) AUC analysis ofblinded ELISA data for the proteins MMP2, MMP10, and THBS2 in plasma samples from 10 patients with PDAC at variousstages of disease compared to 10 healthy controls. CI, confidence interval. (B) Box plots of THBS2 mRNA expressionmeasured in various human tumors (sample sizes in parentheses) assessed by RNA-seq. Tumors are sorted in orderof decreasing median expression of THBS2 mRNA. Of the pancreatic cancer samples from the TCGA database (n =179), we analyzed only PDAC (n = 134). All expression values are log2(RSEM values = 1)–transformed.

2 of 13

https://cancergenome.nih.gov/



Table 1. Demographic and clinical characteristics of patients. Continuous variables (age, body mass index, and CA19-9 concentration) are presented asmean (SD); categorical variables (male gender, personal history of diabetes, and stage of disease) are presented as frequency (%). IPMN, intraductal papillarymucinous neoplasm; PNET, pancreatic neuroendocrine tumor.

Kim
et al., Sci. Transl. Med. 9, eaah
Adenocarcinoma stage I/II

5583 (2017) 12 July 2017

Adenocarcinoma stage III/IV
Controls IPMN, no
adenocarcinoma
PNET Pancreatitis
Discovery phase 1
n = 6 n = 4 n = 10
Age
56.8 (7.5) 65.0 (13.0) 62.2 (15.4)
Male gender
4 (66.7%) 2 (50.0%) 5 (50.0%)
Body mass index (kg/m2)
31.2 (8.6) 26.3 (5.4) 26.0 (3.9)
Personal history of diabetes
1 (16.7%) 0 2 (20.0%)
CA19-9
20770.8 (47060.9) 111224.0 (217927.2) 12.0 (6.4)
D
Stage of disease
o
wn II 1 (16.7%)
lo
ad IIA 1 (16.7%)
e
d fr IIB 4 (66.7%)
o
h
m
III 1 (25.0%)
t
tp:/ IV 3 (75.0%)
/s
tm. Validation phase 2a n = 58 n = 23 n = 80 n = 28
s
cie Age 67.6 (9.4) 68.6 (10.9) 67.4 (9.8) 61.5 (9.4)
n
cem Male gender 43 (74.1%) 12 (52.2%) 54 (67.5%) 19 (67.9%)
a
g.o Body mass index (kg/m2) 28.9 (5.2) 28.5 (5.7) 27.2 (4.7) 26.3 (4.8)
r
b
g/
Personal history of diabetes 15 (25.9%) 3 (13.0%) 12 (15.0%) 3 (10.7%)
y
gu CA19-9 305.3 (411.1) 2137.5 (2983.7) 10.6 (6.9) 68.9 (165.5)
e
st o Stage of disease
n M
I 1 (1.7%)
a
y 2 IA 1 (1.7%)
7
, 20 IB 6 (10.3%)
2
0 II 16 (27.6%)
IIA
13 (22.4%)
IIB
21 (36.2%)
III
10 (43.5%)
IV
13 (56.5%)
Validation phase 2b
n = 88 n = 109 n = 140 n = 115 n = 30 n = 55
Age
66.5 (11.3) 64.4 (11.0) 65.8 (10.8) 68.8 (8.7) 63.2 (7.1) 55.9 (17.7)
Male gender
45 (51.1%) 62 (56.9%) 70 (50.0%) 58 (50.4%) 22 (73.3%) 27 (50.0%)
Body mass index (kg/m2)
28.4 (5.4) 29.1 (6.0) 27.1 (4.4) 26.5 (4.0) 29.0 (4.9) 27.8 (5.0)
Personal history of diabetes
34 (38.6%) 25 (22.9%) 15 (10.7%) 20 (17.4%) 10 (33.3%) 7 (13.0%)
CA19-9
633.8 (1665.9) 2399.3 (3481.1) 12.0 (14.5) 15.4 (12.3) 45.3 (98.0) 35.9 (66.0)
continued on next page

3 of 13




http://stm.sciencem

ag.org/D

ownloaded from

2a group, 29.7 ng/ml, was 12.2 ng/ml higher than that observed incontrols (n = 80) (Fig. 2A), consistent with the mass spectrometry datafor the phase 1 study. THBS2 exhibited a c-statistic of 0.842 for allPDAC samples compared to controls (n = 161; Table 2, all PDACstages). In the same sample set, CA19-9 had a comparable c-statisticof 0.846 for all PDAC samples compared to controls (Table 2, all PDACstages).

The data for the THBS2 analyses were reproducible using threedifferent lot numbers for ELISA kits testing the same subset of phase2a human plasma samples over a 2-year period, with an average 10%coefficient of variation (CV) across the samples (fig. S3). The samplesincluded four plasma samples that were refrozen and thawed twice andthree plasma samples that were refrozen and thawed three times. We con-cluded that the THBS2 assay was robust from the point of view of differ-ences in plasma sample handling and assessment.

To determine whether CA19-9 and THBS2 together could consti-tute a more discriminatory panel than either marker alone, we per-formed logistic regression to estimate the combined probability oftheir discriminatory ability. A combination of CA19-9 and THBS2 forall cases versus controls with the phase 2a data yielded a c-statistic of0.956 (95% CI, 0.93 to 0.98) (Fig. 2C, all PDAC stages, and Table 2),indicating the utility of the two-marker panel.

Phase 2b validation of THBS2 and CA19-9 markers for PDACWe performed an independent phase 2b validation study (see Table 1 forspecimens) with an increased sample size. We accomplished temporalvalidation (18) because the phase 2b analysis was conducted morethan 1 year after the phase 2a study. The distribution of THBS2 valuesacross the phase 2a and 2b studies is shown in fig. S4, and the rangeand median values of THBS2 and CA19-9 are shown in table S3.

The c-statistics for CA19-9 and THBS2 alone, 0.881 and 0.875, re-spectively, slightly improved with the larger sample size of the phase2b study (n = 337), compared to the phase 2a study (n = 161), and thecombination of the two markers yielded a c-statistic of 0.970 (95% CI,0.96 to 0.98) (Table 2, all PDAC stages; Fig. 2, B and D). With regardto the distribution variability, the 75th percentile of the control valuesfell below the 25th percentile of the case values. Furthermore, the 95thpercentile of the controls fell below the median measure observed inthe case samples. The fact that 50% of the case values exceeded 95% ofthe control values was likely driving the AUC we observed for THBS2with regard to being able to discern between cases and controls. We


compared individual and combined marker performance in the phase2a and 2b studies at the stages of resectable PDAC (stages I and II)and locally advanced and metastatic PDAC (stages III and IV). Nota-bly, the combination panel of CA19-9 and THBS2 performed wellacross all stages of PDAC (Table 2).

More detailed analysis of the distribution of ELISA signals providedinsight into how the combination of CA19-9 and THBS2 performed. Asobserved in the scatter plots in Fig. 2 (E and F), various cases (red plus)had essentially zero CA19-9 signal (that is, along the bottom of theplot), consistent with the likelihood that they were from PDAC pa-tients who were Lewis antigen–negative; however, many of thesecases had elevated THBS2 concentrations. Similarly, several cases ex-hibited THBS2 concentrations that overlapped with the upper rangeof the group of controls, and these cases exhibited high CA19-9.Thus, the two markers appeared to be complementary in their abilityto detect PDAC.

Although stages I, IIA, and IIB are classified as resectable tumors inthe sixth edition of the American Joint Committee on Cancer Pancre-atic Cancer Staging System (34), only stages I and IIA are considered tobe early. Therefore, we directly compared the AUCs for combinationsof stages: I + IIA + IIB + II (unspecified), I + IIA + II (unspecified), and I +IIA in our phase 2a and 2b studies. The AUC and 95% CI values werecomparable for the two-marker combination across these three subsets,indicating that the exclusion of the questionable early-stage IIB sampleshad limited impact on marker performance (table S4).

We evaluated the relationship between THBS2 plasma values andage, sex, and the presence of diabetes mellitus in the cohort (table S5, Aand B). We observed no apparent associations between these parametersfor any of the diagnosis groups of PDAC adenocarcinoma stages I/II,adenocarcinoma stages III/IV, pancreatitis, intraductal papillary muci-nous neoplasm, insulinoma (islet cell), and healthy controls. Given theoverall lack of associations, we did not include any of these factors asadjustor variables in subsequent modeling analyses.

Establishing a provisional cutoff plasma concentrationfor THBS2To determine a THBS2 plasma concentration to use as a cutoffpoint for discriminating healthy controls fromPDACcases in the clinic,we first considered the distribution of THBS2 values based on the 230healthy controls from our combined phase 1, 2a, and 2b studies. Fromthis distribution, we chose six cutoffs that represented a range of

Adenocarcinoma stage I/II
Adenocarcinoma stage III/IV Controls IPMN, no
adenocarcinoma
PNET Pancreatitis
Stage of disease

I
4 (4.5%)
IA
2 (2.3%)
IB
4 (4.5%)
II
37 (42.0%)
IIA
15 (17.0%)
IIB
26 (29.5%)
III
41 (37.6%)
IV
68 (62.4%)
4 of 13



approximate false-positive rates (FPRs) from 0 to 5%. These cutoffswere then evaluated for their sensitivity in detecting PDAC in thephase 2a and 2b plasma samples. As seen in Table 3 for the phase 2bstudy, a concentration of THBS2 at or above 42 ng/ml detected abouthalf of the PDACcases (sensitivity) with 99% specificity. Combining theconventional CA19-9 cutoff of ≥55 U/ml and a cutoff of 42 ng/ml forTHBS2 in the phase 2b samples, we observed 98% specificity and 87%sensitivity (Table 3).



http://stm.sciencem

ag.org/D

ownloaded from

Testing the THBS2/CA19-9 panel against other benignpancreatic conditionsWhen considering all PDAC cases versus chronic pancreatitis (phase2a, n = 109; phase 2b, n = 252), c-statistics including CA19-9increased from 0.774 or 0.816 (alone) to 0.842 or 0.867 (with THBS2)with the phase 2a or 2b data, respectively (Table 4 and Fig. 3A). TheTHBS2/CA19-9 panel was able to discriminate all PDAC cases tested(stages I to IV) versus intraductal papillary mucinous neoplasms (n =312) with a c-statistic of 0.952 (Table 4 and Fig. 3B). Thus, theTHBS2/CA19-9 panel was able to distinguish PDAC from intraductalpapillary mucinous neoplasms and helped to distinguish PDAC frompancreatitis compared to CA19-9 alone.

THBS2 lacked the ability to discriminate between all PDAC casesand PNETs and hindered, rather than enhanced, the c-statistic ofCA19-9 (Table 4 and Fig. 3C). Considering a lack of markers availablefor PNETs and the poor performance of THBS2 in the discrimination ofPDAC from PNETs, we examined whether THBS2 could discriminatePNETs (n = 30) from healthy normal controls (n = 149). CA19-9 alonedid not discriminate PNET samples, as previously reported (35). However,THBS2 alone could discriminate PNETs from healthy normal controls witha c-statistic of 0.751 (Table 4 and Fig. 3D).

PDAC can result in obstructive jaundice that can confound plasmaassays (20, 36).Of the 288 adenocarcinoma cases included in these studies,we retrieved clinical total serum bilirubin information for 279 cases(96.9%) (table S6A). Of the 279 samples with this information, 70(25.1%)were inferred to have obstructive jaundice based on total bilirubinconcentrations of ≥3.5 mg/dl. Slightly lower median CA19-9 concentra-tions (208.5 U/ml versus 220 U/ml) as well as elevated median THBS2concentrations (56.4 ng/ml versus 33.0 ng/ml) were observed in PDACsubjects with jaundice when compared to those without jaundice, indi-cating that obstructive jaundice influenced both CA19-9 and THBS2concentrations. Yet, 14 of 55 (25%) PDAC patients with normalCA19-9 and without jaundice had elevated THBS2 concentrations(≥42 ng/ml) (table S6B). Also, 8 of 13 (62%) patients with normalCA19-9 and with jaundice showed elevated THBS2 concentrations(≥42 ng/ml) (table S6B). Therefore, the THBS2 concentration in theplasma identified a subset of nonjaundiced adenocarcinoma cases withnormal CA19-9 concentrations. Furthermore, stratifying the markerpanel performance by overall PDAC or PDACwithout jaundice, versuscontrols, in the phase 2a and 2b studies affected the AUCs by less than0.01 (table S6C). Because of limited availability of benign biliary dis-ease samples, we did not compare THBS2 and CA19-9 concentrationsbetween benign biliary disease, nonjaundice PDAC, and jaundicePDAC.

Cross-validation of THBS2 measurements indifferent laboratoriesAfter these analyses, an independent biomarker development labora-tory at the University of Pennsylvania tested a subset of the phase 2bplasma samples for THBS2 concentrations. Thirty-eight samples wererandomly selected to cover the entire range of THBS2 concentrations,focusing on those around the cutoff value. The samples were de-identifiedand provided without communication other than the manufacturer’sinstructions for the ELISA and Materials and Methods. The ELISAsfor THBS2 were performed more than a year after the original studyand with reagents with different lot numbers. The THBS2 concentra-tions in the original and cross-validated assays were highly concordantand yielded Pearson and Spearman correlation coefficients of 0.95 and0.968, respectively (fig. S5A).

A

C

E

THB

S2

(ng/

ml)

Sen

sitiv

ity (%

)

100 – specificity (%)

ChanceCA19-9THBS2CA19-9 & THBS2

THBS2 (ng/ml)

Phase 2a(n = 161)

Controls + PDAC

THB

S2

(ng/

ml)

Sen

sitiv

ity (%

)

THBS2 (ng/ml)

Phase 2b(n = 337)

Controls + PDAC

100 – specificity (%)


B

D

F

CA

19-9

(U/m

l)

CA

19-9

(U/m

l)

Fig. 2. THBS2 and CA19-9 concentrations in plasma samples from patientswith PDAC versus healthy controls. (A and B) Scatter plots of THBS2 concentra-tions in plasma samples from patients at all stages of PDAC versus controls for thephase 2a (A) and phase 2b (B) validation studies. (C and D) ROC curves for THBS2,CA19-9, and THBS2 + CA19-9 concentrations in plasma samples from patientswith all stages of PDAC versus healthy controls for phase 2a (PDAC, n = 81;controls, n = 80) (C) and phase 2b (PDAC, n = 197; controls, n = 140) (D) studies.(E and F) Scatter plots showing THBS2 and CA19-9 concentrations in plasmasamples from patients with all stages of PDAC versus healthy controls for phase2a (E) and phase 2b (F) studies.

5 of 13




http://stm.sciencem

ag.org/D

ownloaded from

We noticed that the THBS2 signals were slightly lower in the cross-validated data, including those for the normal human control plasmasamples used on each plate. The original studies yielded an averagevalue of 17 ng/ml for the normal control plasma samples, whereasthe cross-validation study yielded a value of 13.25 ng/ml for thenormal control plasma samples. The lower overall values of unknownscaused 4 of the 38 samples that were just over the cutoff of 42 ng/ml tofall below the cutoff (table S7A).

To accommodate operational differences, we created a scalar whereour original cutoff of 42 ng/ml to detect cancer was divided by theoriginal average value of 17 ng/ml (the normal plasma control value)to yield a scalar cutoff of 2.47. We therefore divided the THBS2 resultfor each unknown in the cross-validation study by the value (13.25 ng/ml)of the normal control plasma, where the cutoff value would be 2.47 timesthe control value. Scaling did not affect the correlation coefficient(fig. S5B). With the data scaled in this fashion, two samples that werebelow the cutoff of 42 ng/ml in the original samples were now abovethe cutoff in the cross-validation data (table S7B). Thus, although thescaling method improved the outcome of the cross-validation assay,careful calibration was needed to ensure consistency in the assayresults over time and with different batches of reagent, once a cutofffor clinical practice was determined.

THBS2 expression in different stages of human PDACWe sought to determine the cells expressing THBS2 in a total of 42 casesof human PDAC and 4 cases of incidental PanIN and intraductal pap-illary mucinous neoplasms by immunohistochemistry. All 42 cases ofPDAC and all 4 cases of incidental PanIN/intraductal papillary mucinousneoplasms exhibited detectable THBS2 (Fig. 4, fig. S6, and table S8).Two different antibodies detected THBS2 in PanIN2 epithelia found in-cidentally in the PDAC tumor, but little THBS2 was found in PanIN1epithelia (Fig. 4, A and B). Both antibodies also detected THBS2 in stageII and stage III PDAC. A 10-fold excess of peptide specific to the secondantibody blocked staining by this antibody (Fig. 4, C to K). Epithelial


cells, but not stromal cells, predominantly stained positive for THBS2expression in PanIN/intraductal papillary mucinous neoplasm tissue(four of four) (Fig. 4, A and B, and fig. S6, B and C). In PDAC tumortissue, 32 cases were labeled with THBS2 in epithelial cells and 21 caseswere labeled in both epithelial and stromal cells, and in 8 cases of poor-ly differentiated PDAC tissue, the staining was mostly in stromal cells(table S8).

DISCUSSIONWith a 5-year survival rate for patients with stage I PDAC being atleast four times the overall survival rate for PDAC (34, 37), theTHBS2/CA19-9 marker panel may help to detect early-stage tumorsthat are resectable and should improve the prognosis of PDAC. Theperformance of THBS2 in early-stage cancer may be a consequence ofour discovery that it is secreted or released from human precursorPanIN organoids (25), reflecting the ability of the iPSC-like 10-22PDAC cell line to recapitulate human pancreatic cancer progression.Although most patients with PDAC are diagnosed at advanced stages,the time from the occurrence of the initiating mutation to the birth ofPDAC founder cells (38) can be a decade, suggesting that there maybe a time period to identify progressing disease before PDAC can beclinically imaged. PanINs have been identified in the pancreas up to10 years before the development of infiltrating PDAC (39), underscor-ing the importance of early diagnosis. However, we note that PanINscan also be observed in the absence of PDAC.

We propose that high specificity outweighs considerations ofincreased sensitivity because of heightened anxiety in patients oversuspected pancreatic cancer plus the costs of subsequent diagnosticevaluation. We found that, with a THBS2 concentration cutoff of42 ng/ml, THBS2 could discriminate PDAC patients from healthyprimary care controls with a specificity of 99% (1% FPR) and a sen-sitivity of 52%. Impressively, combining CA19-9 (>55 U/ml) withTHBS2 (>42 ng/ml) showed a specificity of 98% and a sensitivity of

Table 2. AUC calculations of ELISA results for PDAC at different stages, bootstrapped (1000 repetitions) at 95% CIs.

PDAC versus healthy controls
CA19-9 (≥55) THBS2 CA19-9 (≥55) + THBS2
n
AUC 95% CI AUC 95% CI AUC 95% CI P
Validation phase 2a

All stages
81/80 0.846 0.81 0.89 0.842 0.80 0.90 0.956 0.93 0.98 0.0003
Stage I/II
58/80 0.845 0.80 0.89 0.832 0.78 0.89 0.946 0.92 0.98 0.0067
Stage I
8/80 0.938 0.83 1.00 0.839 0.69 0.98 0.977 0.93 1.00 0.5038
Stage II
50/80 0.830 0.78 0.88 0.830 0.78 0.89 0.940 0.91 0.98 0.0080
Stage III/IV
23/80 0.846 0.79 0.90 0.842 0.78 0.90 0.956 0.92 0.99 0.0003
Validation phase 2b

All stages
197/140 0.881 0.86 0.91 0.875 0.85 0.90 0.970 0.96 0.98 <0.0001
Stage I/II
88/140 0.834 0.79 0.87 0.887 0.85 0.92 0.960 0.94 0.98 <0.0001
Stage I
10/140 0.793 0.66 0.93 0.896 0.83 0.97 0.952 0.92 0.99 0.0574
Stage II
78/140 0.839 0.80 0.88 0.885 0.85 0.92 0.961 0.94 0.98 <0.0001
Stage III/IV
109/140 0.919 0.89 0.95 0.866 0.83 0.90 0.980 0.97 0.99 0.0028
6 of 13




http://stm.sciencem

ag.org/D

ownloaded from

87% in our larger phase 2b study. An important strength of this studywas the ability to use large defined plasma samples obtained from asingle institution that followed standardized processing protocols (23).

Decreased THBS1 concentrations by mass spectroscopy analysishave been reported in plasma samples from PDAC patients and insamples obtained before the cancer diagnosis (20, 40). Jenkinson et al.(20) found that reduced concentrations of THBS1 occurred in PDACpatients with diabetes, but not in PDAC patients without diabetes. In


contrast, we observed that elevated concentrations of THBS2 were as-sociated with PDAC, but we did not find an association of elevatedTHBS2 concentrations with diabetes mellitus, age, or sex (table S5, Aand B). THBS1 and THBS2 share 80% of their protein sequence buthave diverged in function and in their genetic regulation (41, 42). Weshowed that elevated THBS2 did not correspond to THBS1 in PDAC.First, the antibodies in the ELISA kit used for our study had negligiblecross-reactivity with THBS1 (figs. S1 and S2). Second, we confirmed by

Table 3. THBS2 concentration cutoff points based on percentiles of distribution in control plasma samples.

Marker
Phase 2a Phase 2b
Cutoff
Sensitivity Specificity Sensitivity Specificity
CA19-9 (≥55)
69.14 100 77.66 98.57
THBS2 (ng/ml)

95%
36 33.33 96.25 58.38 93.57
96%
37 33.33 97.50 57.36 95.00
97%
38 30.86 97.50 53.81 96.43
98%
40 28.40 97.50 53.30 97.86
99%
42 24.69 97.50 51.78 99.29
100%
73.4 7.41 100 23.35 100
CA19-9 (≥55) and THBS2 (ng/ml)

95%
36 74.07 96.25 88.32 92.86
96%
37 74.07 97.50 88.32 94.29
97%
38 74.07 97.50 87.82 95.71
98%
40 74.07 97.50 87.82 97.14
99%
42 72.84 97.50 87.31 97.86
100%
73.4 69.14 100 81.73 98.57
Table 4. AUC calculations of ELISA results for patients with all-stage PDAC versus patients with benign pancreatic diseases, bootstrapped (1000repetitions) at 95% CIs.

PDAC versus benign pancreatic diseases
CA19-9 (≥55) THBS2 CA19-9 (≥55) + THBS2
n
AUC 95% CI AUC 95% CI AUC 95% CI P
Validation phase 2a

PDAC versus pancreatitis
81/28 0.774 0.71 0.84 0.727 0.64 0.81 0.842 0.76 0.92 0.2740
Validation phase 2b

PDAC versus pancreatitis
197/55 0.816 0.77 0.86 0.731 0.67 0.79 0.867 0.82 0.92 0.2450
PDAC versus IPMN
197/115 0.884 0.86 0.91 0.784 0.74 0.82 0.952 0.93 0.97 0.0003
PDAC versus PNET
197/30 0.805 0.75 0.86 0.444 0.36 0.71 0.785 0.72 0.89 0.6502
PNET versus controls
30/140 0.576 0.52 0.63 0.751 0.67 0.84 0.755 0.67 0.85 0.0089
7 of 13




http://stm.sciencem

ag.org/D

ownloaded from

mass spectrometry that the peptides specific to THBS2 were moreabundant in cancer patient plasma samples than in plasma samplesfrom normal healthy controls (table S2). Thus, elevated THBS2 concen-trations in PDAC were independent of THBS1 concentrations reportedin the literature.

THBS2 is a glycoprotein that may be an angiogenesis inhibitor,and mutation of the mouse TSP-2 gene increases susceptibility tocancer (43). We found that THBS2 antigen is expressed in normalpancreas cells, but the baseline concentration of THBS2 is very lowin normal human plasma by both mass spectrometry and ELISA.We found that THBS2 antigen is robustly expressed in PDAC tumortissue, perhaps concordant with the poor vascularization associatedwith PDAC, and is also increased in the plasma of PDAC patients.THBS2 is down-regulated in gastric cancer cells (44). Further workis needed to understand how the release of THBS2 into the plasmais increased in the patient cohorts studied here.

A group of scientists has initiated the STARD (Standards for Re-porting of Diagnostic Accuracy) with guidelines to improve the re-porting of diagnostic accuracy (45). It will be useful to follow thesestandard guidelines in the clinic by reporting imprecision as the CV(%CV) and precision as 95% CIs near clinical decision points obtainedby repeating the test over several independent days. Also, to reduceeven small differences in the assay occurring between different labora-tories, presenting the likelihood ratio with 95% CIs along with speci-ficity and sensitivity at several cutoff points is recommended. Ourcross-validation study was an initial attempt to address these issues,and more work is needed for a determination of clinical decisionpoints with confidence.


The combination ofTHBS2 andCA19-9improved the discrimination of patientswithPDAC from those with chronic pancreatitis.Longitudinal studies areneeded todeterminewhether a subset of the pancreatitis patientswho scored positive for THBS2 but wereclinically assessed to be PDAC-negativeharbored early-stage PDAC. Likewise, largerstudies are needed todetermine the effective-ness of THBS2 for diagnosing PNETswhereCA19-9 is not applicable, and other cancersshowing high THBS2 mRNA expression(Fig. 1B). Further research with larger num-bers of pancreatic cancer cases without jaun-dice as well as patients without cancer butwith jaundice will be necessary to quantifythe value of THBS2 and CA19-9 for detect-ing nonjaundice pancreatic cancer.

There are limitations to our study.The prevalence of PDAC in different po-pulations affects the positive predictivevalue (PPV) and negative predictive val-ue (NPV) for determining the utility of abiomarker in a population. The PPV isthe probability that subjects with a posi-tive screening test have the disease, andthe NPV is the probability that subjectswith a negative screening test do not havethe disease. Given the low prevalence of 4to 12.4 cases of pancreatic cancer per100,000 in the general population (https://

seer.cancer.gov/statfacts/html/pancreas.html), our marker panel witha combined 98% specificity and 87% sensitivity would have a PPV of0.002, with an NPV of 1.0 (2). Yet, when viewed in terms of the 1.5%lifetime risk of PDAC in the general population (2), the PPV becomesabout 0.4 with an NPV of 0.99. For patients older than 55 years who arenewly diagnosed with diabetes (46), with a prevalence of 1% in thegeneral population for PDAC, the PPV is 0.31 and the NPV is 1.0. Forfirst-degree relatives of PDAC patients and smokers in the general popu-lation, each group with a lifetime risk of 3.75% (47), the PPV is 0.63 andthe NPV is 0.99. For carriers with relevant germline mutations (in ag-gregate, including BRCA1, BRCA2, CDKN2A, and PALB2), the lifetimerisk is 40% (47), the PPV rises to 0.97, and the NPV is 0.92. On the basisof these considerations, we suggest that the THBS2/CA19-9 markerpanel could serve as a low-cost, nonintervention screening tool inasymptomatic individuals who have a high risk of developing PDAC(3, 47, 48) and also in patients who are newly diagnosed with diabetesmellitus that developed as a result of pancreatic injury (49), but not inthe general population.

Another limitation of our work is that our histological analysis ofTHBS2 expression at different stages of pancreatic cancer was limitedby the portion of tissue available from each of the resections. Further-more, it is unclear how expression of THBS2 in cells under normal orpathological conditions may relate to the extent to which the protein issecreted or released into the plasma and how stable it is in the plasma.Also, further work is needed to refine clinical decision points for high-risk individuals, to determine the panel’s utility for detecting earlier-stageprogression to PDAC, and to determine the specificity for pancreaticcancer versus cancers of other tissue types and other disease states.

A

100 − specificity (%)

Sen

sitiv

ity (

%)

Sen

sitiv

ity (

%)

Sen

sitiv

ity (

%)

Sen

sitiv

ity (

%)

PDAC vs pancreatitis PDAC vs IPMN

PDAC vs PNET PNET vs healthy controls


100 − specificity (%)

100 − specificity (%) 100 − specificity (%)

C

B

D

Fig. 3. THBS2 and CA19-9 concentrations in plasma samples from all-stage PDAC cases versus benign pan-creatic disease cases. (A to D) ROC curves for THBS2, CA19-9, and THBS2 + CA19-9 concentrations in plasmasamples from patients with PDAC in the phase 2b study (n = 197) versus pancreatitis (n = 55) (A), PDAC versusintraductal papillary mucinous neoplasm (IPMN) (n = 115) (B), PDAC versus PNET (n = 30) (C), and PNET (n = 30)versus healthy controls (n = 140) (D) are shown.

8 of 13

https://seer.cancer.gov/statfacts/html/pancreas.html

https://seer.cancer.gov/statfacts/html/pancreas.html




http://stm.sciencem

ag.org/D

ownloaded from

MATERIALS AND METHODSStudy design and populationsAll procedures were performed using a recommended biomarkerphased design following the PRoBE criteria (28, 29). De-identified humanplasma samples from the Mayo Clinic pancreas research biospecimen re-pository were shipped to our laboratory, where ELISA analyses were per-formed blinded to disease status, and then coded data were returned to theMayo Clinic team for statistical analysis and interpretation.


Collection of plasma samples was ap-proved by the Mayo Clinic Institutional Re-view Board (IRB). After rapid case finding(50) and informed consent, participantswith PDAC provided venous blood sam-ples before initiation of cancer therapy.Samples were frozen at −80°C until used.Similarly, blood samples were obtainedfrom theMayo Clinic through primary care(healthy controls) and gastroenterologyclinics (participants diagnosedwith chronicpancreatitis, intraductal papillarymucinousneoplasms, and PNETs). An aliquot of theserum was assayed for CA19-9 at the MayoClinic Immunochemical Core Laboratoryas recommended by the ELISA kit manu-facturer (Cobas/Roche). Demographic andclinical characteristics in each group areshown in Table 1.Exploratory set (phase 1)Plasma samples from 20 non-HispanicCaucasian subjects recruited at the MayoClinic included 10 healthy primary carecontrols and 10 [6 early stage (I/II) and 4late stage (III/IV)] patients with clinicallyor histologically proven PDAC. All cancercases for phase 1 were selected to haveCA19-9 concentrations above 55 U/ml.Validation set (phase 2a)Plasma samples from 189 non-HispanicCaucasian subjects recruited at the MayoClinic included 81 (58 early stage and 23late stage) patients with clinically or histo-logically proven PDAC, 80 healthy primarycare controls, and 28 patients with a person-al history of chronic pancreatitis; patientswith hereditary pancreatitis were excludedgiven their increased risk for PDAC. Thecontrols were matched to the cases by ageand sex. About 15% of the healthy controlsself-reported a personal history of diabetes.Validation set (phase 2b)Plasma samples collected from 537 non-Hispanic Caucasian subjects recruited atthe Mayo Clinic included 197 (88 earlystage and 109 late stage) patients withclinically or histologically proven PDAC,140 healthy primary care controls, 115patients with intraductal papillary muci-nous neoplasm without PDAC, 30 patientswith PNET, and 55 patients with a self-

reported personal history of chronic pancreatitis; patients with hered-itary pancreatitis were excluded. About 11% of the controls self-reporteda personal history of diabetes.

Measurement of markers in human plasmaAfter the ELISAs for the phase 1 study were completed, the remainingPDAC(n=10) and control samples (n=10)were each separately depletedof abundant serumproteins by filtration and thenhigh-performance liquid

PDAC stage 3

THBS2 Ab#1 Ab#2 + peptide block

THBS2 Ab#1 THBS2 Ab#2Incidental PanIN in PDAC

PanIN2(positive)

PanIN1(weak/negative)

PDAC stage 2

100 µm

A B

C

GF

ED

H

JI K

THBS2 Ab#2

Fig. 4. Expression of THBS2 in human PanIN tissue and PDAC tumor tissue. (A and B) Representative THBS2immunohistochemistry analysis of incidental PanIN1/2 tissue derived from the head and neck of a pancreas from apatient with pancreatic periampullary cancer using two different antibodies (Ab). The arrows indicate that PanIN2tissue stained positive for THBS2; dotted arrows indicate weak or negative staining of PanIN1 tissue. (C to K) THBS2expression, designated by arrows, was also confirmed in stage II (C to E) and stage III (F to K) PDAC pancreaticcancer tissue arrays. Competitive assays were performed for antibody #2 by preincubating the antibody with a10-fold excess of antigen peptide (E, H, and K) to confirm target specificity. Brown color indicates THBS2 staining,and blue color indicates hematoxylin nuclear staining. THBS2 was detected in the epithelial cells of noninvasivelesions (PanINs and intraductal papillary mucinous neoplasms) and poorly differentiated PDAC tissue as well as infibroblasts in invasive PDAC tissue (see table S8 and fig. S6).

9 of 13




http://stm.sciencem

ag.org/D

ownloaded from

chromatography using a Seppro IgY14 LC10 column (Sigma-Aldrich).The resulting 10 samples of cancer plasmas, depleted of abundant pro-teins, were pooled separately from a pool of the controls, and the twopools were subjected to two-dimensional strong cation exchange (SCX)chromatography/tandem mass spectrometry analysis as previously de-scribed (51). Inbrief, anSCXtip columnwasmadewitha200-ml tippackedwith 20-ml PolySULFOETHYL resin (Nest Group). The SCX tip was pre-washed with buffer B [500mMKCl, 10mMNaH2PO4, and 30% acetoni-trile (pH 2.6)], followed by equilibration with buffer A [10 mMNaH2PO4

and 30% acetonitrile (pH 2.6)]. The lyophilized digested peptides (100 mg)were reconstituted in 50-ml buffer A. The reconstituted digested peptidesolutionwas loaded into the SCX tip column twice, followed by washingwith 50-ml buffer A. All flow-through fractions were combined (“flow-through”). The following 100-ml KCl concentration buffers, made bymixing the different proportions of buffer A and B, were used to suc-cessively wash the column: 30, 40, 50, 60, 70, 85, 100, 150, and 500mM.A total of 10 fractionswere dried and desalted using the homemadeC18Stage Tips. About 3 mg of the digested peptides was injected into a75-mm inner diameter × 25-cm C18 column with a pulled tip. EASY-nLC 1000 was run at a flow rate of 300 nl/min for a 180-min gradient.Online nanospray was used to spray the separated peptides into an Or-bitrap Fusion Tribrid mass spectrometer (Thermo Electron). The rawdata were acquired with Xcalibur, and pFind2.8 software was used tosearch UniProt Human database. A 5% FDR for the protein spectrummeasurement was used initially to filter the peptide search results. TableS2 shows the results for a total of four THBS2 peptides thatweredetectedin the pooled cancer plasma samples. The pFind2.8 search engine revealedtwo unique peptides in each of the pooled plasmas; the cancer pool had apeptide specific to THBS2 (VCNSPEPQYGGK) and a peptide shared be-tween THBS1 and THBS2 (NALWHTGNTPGQVR), and the normal poolhad a peptide specific to THBS2 (TRNMSACWQDGR) and a peptideshared between THBS1 and THBS2 (FYVVMWK) (table S2A). Wequantified THBS2 levels in each of the pooled plasmas by measuringthe area under the peptide signals based on mass and retention timein original MS1 andMS2 windows.We then normalized the total spec-tral counts in each sample (table S2A).

To more stringently assess THBS2 peptide levels, we searched with1 and 5% FDR settings against the UniProt Human database (89,796entries in total) with the updated pFind3.0 search engine (52). Searchparameters were set for a precursor mass tolerance of ±7 parts permillion, fragment mass tolerance of ±0.4 Da, trypsin cleaving after lysineand arginine with up to two miscleavages, carbamidomethyl (C)/+57.021as the fixed modification, and acetyl (protein N-term)/+42.011, de-amidated (NQ)/+0.984, and oxidation (M)/+15.995 as the variablemodifications. The target-decoy approach was used to filter the searchresults, in which the FDR was less than 1 or 5% at both the peptideand protein level. At both 1 and 5% FDR, the peptide specific toTHBS2 (VCNSPEPQYGGK) and the peptide shared between THBS1and THBS2 (NALWHTGNTPGQVR) were seen in the cancer pooledsample, consistent with the original pFind2.8 search (table S2, B andC). However, at either 1 or 5% FDR with the pFind3.0 search, no pep-tides specific to THBS2 sequence were identified, and only peptidesshared between THBS1 and THBS2 were identified in the normalpooled sample (table S2, B and C). Therefore, more stringent analysisverified THBS2 sequence in cancer pooled sample.

ELISAkits forhumanMMP2(Millipore),humanMMP10(RayBiotech),and human Thrombospondin-2 Quantikine (DTSP20, R&D Systems)were used as described by the manufacturers’ instructions. Duplicate5-ml plasma samples were diluted 10-fold with calibrator diluent


RD5P buffer and all 50 ml used for THBS2.Marker concentrations weredetermined from standard curves of positive control proteins from thekits with a four-parameter logistic nonlinear regression model usingSoftMax Pro Software (Molecular Devices). Normal pooled humanplasma (IPLA-N, Innovative Research) was tested in duplicate on eachELISA plate. Across 15 independent ELISA plates, THBS2 in duplicatecontrol samples of commercial normal pooled human plasma rangedbetween 15 and 21 ng/ml, with a CV of 13%. Also, the inclusion (orexclusion) of occasional plasma samples that were orange or reddishin color, indicating hemolysis, had a negligible impact on the data.

RNA-seq analysis from TCGATHBS2 mRNA amounts were assessed in TCGA RNA-seq data sets(http://cancergenome.nih.gov/) using the cBioPortal for Cancer Genomics(53, 54). Data were downloaded from the University of California, SantaCruz Xena data hub and sample IDs curated using the Broad Institute’sGenome Data Analysis Center Firehose. THBS2mRNA values were es-timated by the RNA-Seq by Expectation Maximization (RSEM)algorithm (55) and log2(RSEM + 1)–transformed for Fig. 1B, as parsedand plotted using scripts in Python, R.

Western blot and ELISA for validation of THBS2 ELISA kitsfor cross-reactivity with THBS1The recombinant THBS proteins were obtained from R&D Systemsand performed Western blot with polyclonal goat anti-THBS2 (detec-tion antibody; working concentration, 0.15 nM) and monoclonalmouse anti-THBS2 (capture antibody; working concentration, 3 nM)to check the cross-reactivity. A detection antibody and a 100-fold molarexcess of recombinant THBS1 or THBS2 proteins were incubated in 5%nonfat milk for 30 min at room temperature for competition assay. Theincubated solution was centrifuged at 10,000 rpm for 15 min to removeany immunocomplexes before applying onto a polyvinylidene difluoride(PVDF) membrane; a total of two 10-ng proteins were transferred fordetection antibodies. For competition assay of capture antibody, a 10-foldmolar excess of recombinant THBS1 or THBS2 proteins was incu-bated with capture antibody in 5% bovine serum albumin for 30 minat room temperature. The incubated solution was centrifuged at15,000 rpm for 15 min to remove any immunocomplexes before ap-plying onto a PVDF membrane; a total of 10 50-ng proteins weretransferred for detection antibodies. The presence of THBS2 in a gelwas confirmed by silver staining or reprobing membranes with detec-tion antibody in THBS2-competed membranes. To determine wheth-er the presence of THBS1 interferes with the THBS2 ELISA, we spikeda recombinant THBS1 protein (200 ng/ml) into various concentrationsof recombinant THBS2 proteins (0 to 20 ng/ml) or human plasma ofwide range of THBS2 in THBS2 ELISA.

Immunostaining of THBS2 in human pancreaticcancer tissueThe pancreatic tumor tissue sections were obtained from US Biomax(catalog #PA1002), and each tissue spot was individually examined bytheir own pathologists certified according to World Health Organiza-tion published standardizations of diagnosis, classification, and path-ological grade. Incidental PanIN1/2 tissue section was derived fromthe head and neck of the pancreas of a pancreatic periampullary cancerpatient at the Fox Chase Cancer Center (FCCC) under IRB 09-801 toK.S.Z., and its histology was confirmed by pathologist J. Anderson atFCCC. These tissue blocks do not correspond to plasma samples wherewe measured plasma THBS2 concentrations. The paraffin-embedded

10 of 13

http://cancergenome.nih.gov/




http://stm.sciencem

ag.org/D

ownloaded from

tissues were antigen-retrieved by boiling in citric acid buffer (pH 6.0)after deparaffination. Next, the endogenous peroxidase activity in tissueslides was quenched in hydrogen peroxide solution for 15 min at roomtemperature. Tissues were blocked with avidin/biotin blocking (VectorLaboratories) for 15 min each, followed by nonprotein blocker (ThermoFisher Scientific) for 30 min at room temperature. Primary antibodieswere applied and incubated for 12 to 16 hours at 4°C. Two primaryantibodies for THBS2 were used for our study: goat polyclonal THBS2antibody (dilution 1:25; sc-7655, Santa Cruz Biotechnology) and rabbitpolyclonal THBS2 antibody (dilution 1:100; TA590658, OriGene). It isnot clear where TA590658 antibody recognizes and whether it detectssecreted THBS2. Yet, sc-7655 antibody can detect both secreted and cy-toplasmic THBS2 because it targets the epitopes of 15 to 20 aminoacids in length that are located within the first 50 amino acids ofthe peptide sequence for THBS2, whose signal peptides are locatedbetween 1 and 18 amino acids. Only two amino acids of the epitopeare overlapped to the signal peptides, and the remaining amino acidsare overlapped over the main body of the peptides. A peptide was avail-able for sc-7655 from Santa Cruz Biotechnology, thus the sc-7655antibodies were incubated with the corresponding peptides in 10-foldexcess for 30 min before being applied onto tissue section to confirmthe specificity of signals. Also, no primary antibody controls for sc-7655and TA590658 antibodies were used for negative controls. After wash-ing twice, tissues were incubated with biotinylated anti-goat immuno-globulin G (IgG) or rabbit IgG (Vector Laboratories) at 37°C for 30 min.Tissue sections were conjugated with avidin–horseradish peroxidaseby using VECTASTAIN Elite ABC kit (Vector Laboratories) at 37°Cfor 30 min, followed by developing with DAB (3,3′-diaminobenzidine)peroxidase substrate kit (Vector Laboratories) for peroxidase for 4 to5 min. Developed tissue sections were stained with hematoxylin fornucleus, dehydrated, andmounted.We confirmed theTHBS2 sequenceof the peptide by mass spectrometry.

Statistical analysisThe primary comparison for this study was defined as PDAC cases(all stages) versus healthy controls. To explore any relationship betweenpatient demographic information and THBS2 plasma concentrations,we calculated a Spearman correlation coefficient for continuous varia-bles (age) and median expression concentrations for categorical varia-bles (sex, male versus female; presence or absence of diabetes mellitus).On the basis of the data obtained from our phase 1 and 2 studies, weobserved no apparent associations between age, sex, diabetes mellitus,jaundice, and THBS2. Given this lack of association, any concernsregarding the potential for confounding were mitigated, and theseclinical factors were not included in subsequent multivariablemodeling.

Univariate and multivariable logistic regression models were devel-oped to consider each candidate marker (THBS2) alone and combinedwith CA19-9. The response variable was coded as 1 to indicate the pres-ence of cancer (0 for controls). Candidate markers (THBS2) were enteredas continuous variables. CA19-9 was dichotomized as 0 (normal) (<55U/ml)or 1 (elevated) (≥55 U/ml). The AUC was calculated for each modelconsidered. To assess whether the difference observed between AUCsfrom the CA19-9 and THBS2 and the CA19-9 alone models was sta-tistically significantly different from 0, we considered a test statisticT (T = AUCCa199 − AUCCa199+THBS2)

2/(s2Ca199 + s2Ca199+THBS2)(56), which looks at the difference in AUC between the two modelsdivided by the sum of the variances from the two models. The factthat this test statistic followed a c² distribution with 1 degree of free-


dom under the null hypothesis was used to calculate a resulting P val-ue. A bootstrap percentile CI approach was used to estimate a 95% CIfor the AUC. This approach resampled the data set (1000 times) andthen ran the logistic regression models to calculate the AUC on eachbootstrapped data set to approximate the sampling distribution of theAUC. The 2.5th and 97.5th percentiles from this distribution of AUCvalues were then used as estimates of lower and upper bounds for the95% CI for the AUC.

A similar approach was considered for each of the subanalyses thatstratified by stage (early stage and late stage) and other comparisongroups (intraductal papillary mucinous neoplasms, chronic pancreatitis,or PNET). A k statistic was calculated to assess agreement (below cutoffversus above cutoff) in the THBS2 assay results from each of the twoindependent laboratories in the cross-validation study. Analyses wereperformed using SAS 9.4 on Linux.

SUPPLEMENTARY MATERIALSwww.sciencetranslationalmedicine.org/cgi/content/full/9/398/eaah5583/DC1Fig. S1. Validation of cross-reactivity of antibodies enclosed in THBS2 ELISA with THBS1 byWestern blot.Fig. S2. Validation of cross-reactivity and interference of THBS1 in THBS2 ELISA.Fig. S3. Reproducibility of the ELISA for THBS2.Fig. S4. Distribution of THBS2 values in phase 2 samples.Fig. S5. Cross-validation tests, performed 1 year apart, of THBS2 concentrations in the same setof plasmas as determined in different laboratories.Fig. S6. Representative immunohistochemistry images of THBS2 in human normal pancreas,pancreatitis, and PDAC tissue.

Table S1. List of 53 proteins secreted or released from 10-22 cell, PanIN-stagelesions that are at low abundance in healthy human plasma proteome and RNA-seqdatabases.

Table S2. Mass spectrometry assessment of THBS2 concentrations in phase 1 plasmasamples.

Table S3. Range and median values of THBS2 and CA19-9 in this study.

Table S4. Impact of excluding stage IIB (and unspecified stage II) subjects.

Table S5A. THBS2 values by sex and diabetes mellitus status.

Table S5B. Spearman correlation analysis of age and THBS2 values.

Table S6A. Obstructive jaundice cases in the PDAC cohorts.

Table S6B. THBS2 and CA19-9 values and obstructive jaundice status.

Table S6C. AUC values for CA19-9, THBS2, and combined markers by jaundice status in phases2a and 2b of PDAC cases versus controls.

Table S7A. Cross-tabulation of normal versus elevated THBS2 values, given a cutoff of42 ng/ml, for the original cross-validation THBS2 assays (k = 0.786).

Table S7B. Cross-tabulation of normal versus elevated scaled THBS2 values, given a cutoff of2.47, for the original and cross-validation THBS2 assays (k = 0.895).

Table S8. Summary of THBS2 immunohistochemistry in a total of 42 human PDAC and 4 casesof incidental PanIN and intraductal papillary mucinous neoplasm by immunohistochemistry.

REFERENCES AND NOTES1. L. Rahib, B. D. Smith, R. Aizenberg, A. B. Rosenzweig, J. M. Fleshman, L. M. Matrisian,

Projecting cancer incidence and deaths to 2030: The unexpected burden ofthyroid, liver, and pancreas cancers in the United States. Cancer Res. 74, 2913–2921(2014).

2. R. L. Siegel, K. D. Miller, A. Jemal, Cancer statistics, 2016. CA Cancer J. Clin. 66, 7–30(2016).

3. S. T. Chari, K. Kelly, M. A. Hollingsworth, S. P. Thayer, D. A. Ahlquist, D. K. Andersen,S. K. Batra, T. A. Brentnall, M. Canto, D. F. Cleeter, M. A. Firpo, S. S. Gambhir, V. L. Go,O. J. Hines, B. J. Kenner, D. S. Klimstra, M. M. Lerch, M. J. Levy, A. Maitra, S. J. Mulvihill,G. M. Petersen, A. D. Rhim, D. M. Simeone, S. Srivastava, M. Tanaka, A. I. Vinik,D. Wong, Early detection of sporadic pancreatic cancer: Summative review. Pancreas 44,693–712 (2015).

4. K. Satake, G. Kanazawa, I. Kho, Y.-S. Chung, K. Umeyama, A clinical evaluation ofcarbohydrate antigen 19-9 and carcinoembryonic antigen in patients with pancreaticcarcinoma. J. Surg. Oncol. 29, 15–21 (1985).

11 of 13

http://www.sciencetranslationalmedicine.org/cgi/content/full/9/398/eaah5583/DC1




http://stm.sciencem

ag.org/D

ownloaded from

5. M. J. Duffy, C. Sturgeon, R. Lamerz, C. Haglund, V. L. Holubec, R. Klapdor, A. Nicolini,O. Topolcan, V. Heinemann, Tumor markers in pancreatic cancer: A European Group onTumor Markers (EGTM) status report. Ann. Oncol. 21, 441–447 (2010).

6. G. Y. Locker, S. Hamilton, J. Harris, J. M. Jessup, N. Kemeny, J. S. Macdonald,M. R. Somerfield, D. F. Hayes, R. C. Bast Jr.; American Society of Clinical Oncology, ASCO2006 update of recommendations for the use of tumor markers in gastrointestinalcancer. J. Clin. Oncol. 24, 5313–5327 (2006).

7. K. S. Goonetilleke, A. K. Siriwardena, Systematic review of carbohydrate antigen (CA 19-9)as a biochemical marker in the diagnosis of pancreatic cancer. Eur. J. Surg. Oncol. 33,266–270 (2007).

8. M. A. Tempero, E. Uchida, H. Takasaki, D. A. Burnett, Z. Steplewski, P. M. Pour, Relationshipof carbohydrate antigen 19-9 and Lewis antigens in pancreatic cancer. Cancer Res.47, 5501–5503 (1987).

9. H. C. Harsha, K. Kandasamy, P. Ranganathan, S. Rani, S. Ramabadran, S. Gollapudi,L. Balakrishnan, S. B. Dwivedi, D. Telikicherla, L. D. Selvan, R. Goel, S. Mathivanan,A. Marimuthu, M. Kashyap, R. F. Vizza, R. J. Mayer, J. A. Decaprio, S. Srivastava,S. M. Hanash, R. H. Hruban, A. Pandey, A compendium of potential biomarkers ofpancreatic cancer. PLOS Med. 6, e1000046 (2009).

10. J. R. Mayers, C. Wu, C. B. Clish, P. Kraft, M. E. Torrence, B. P. Fiske, C. Yuan, Y. Bao,M. K. Townsend, S. S. Tworoger, S. M. Davidson, T. Papagiannakopoulos, A. Yang,T. L. Dayton, S. Ogino, M. J. Stampfer, E. L. Giovannucci, Z. R. Qian, D. A. Rubinson, J. Ma,H. D. Sesso, J. M. Gaziano, B. B. Cochrane, S. Liu, J. Wactawski-Wende, J. E. Manson,M. N. Pollak, A. C. Kimmelman, A. Souza, K. Pierce, T. J. Wang, R. E. Gerszten, C. S. Fuchs,M. G. Vander Heiden, B. M. Wolpin, Elevation of circulating branched-chain aminoacids is an early event in human pancreatic adenocarcinoma development. Nat. Med. 20,1193–1198 (2014).

11. S. A. Melo, L. B. Luecke, C. Kahlert, A. F. Fernandez, S. T. Gammon, J. Kaye, V. S. LeBleu,E. A. Mittendorf, J. Weitz, N. Rahbari, C. Reissfelder, C. Pilarsky, M. F. Fraga,D. Piwnica-Worms, R. Kalluri, Glypican-1 identifies cancer exosomes and detects earlypancreatic cancer. Nature 523, 177–182 (2015).

12. K. H. Yu, C. G. Barry, D. Austin, C. M. Busch, V. Sangar, A. K. Rustgi, I. A. Blair, Stable isotopedilution multidimensional liquid chromatography-tandem mass spectrometry forpancreatic cancer serum biomarker discovery. J. Proteome Res. 8, 1565–1576 (2009).

13. A. Chan, I. Prassas, A. Dimitromanolakis, R. E. Brand, S. Serra, E. P. Diamandis, I. M. Blasutig,Validation of biomarkers that complement CA19.9 in detecting early pancreaticcancer. Clin. Cancer Res. 20, 5787–5795 (2014).

14. S. Makawita, A. Dimitromanolakis, A. Soosaipillai, I. Soleas, A. Chan, S. Gallinger, R. S. Haun,I. M. Blasutig, E. P. Diamandis, Validation of four candidate pancreatic cancer serologicalbiomarkers that improve the performance of CA19.9. BMC Cancer 13, 404 (2013).

15. T. P. Radon, N. J. Massat, R. Jones, W. Alrawashdeh, L. Dumartin, D. Ennis, S. W. Duffy,H. M. Kocher, S. P. Pereira, L. Guarner (posthumous), C. Murta-Nascimento, F. X. Real,N. Malats, J. Neoptolemos, E. Costello, W. Greenhalf, N. R. Lemoine, T. Crnogorac-Jurcevic,Identification of a three-biomarker panel in urine for early detection of pancreaticadenocarcinoma. Clin. Cancer Res. 21, 3512–3521 (2015).

16. N. A. Schultz, C. Dehlendorff, B. V. Jensen, J. K. Bjerregaard, K. R. Nielsen, S. E. Bojesen,D. Calatayud, S. E. Nielsen, M. Yilmaz, N. H. Holländer, K. K. Andersen, J. S. Johansen,MicroRNA biomarkers in whole blood for detection of pancreatic cancer. JAMA 311,392–404 (2014).

17. M. Bauden, D. Pamart, D. Ansari, M. Herzog, M. Eccleston, J. Micallef, B. Andersson,R. Andersson, Circulating nucleosomes as epigenetic biomarkers in pancreatic cancer.Clin. Epigenetics 7, 106 (2015).

18. M. E. Miller, S. L. Hui, W. M. Tierney, Validation techniques for logistic regression models.Stat. Med. 10, 1213–1226 (1991).

19. R. B. Rao, G. Fung, R. Rosales, On the dangers of cross-validation: An experimentalevaluation, in Proceedings of the 2008 SIAM International Conference on Data Mining(2008), pp. 588–598.

20. C. Jenkinson, V. L. Elliott, A. Evans, L. Oldfield, R. E. Jenkins, D. P. O’Brien, S. Apostolidou,A. Gentry-Maharaj, E.-O. Fourkala, I. J. Jacobs, U. Menon, T. Cox, F. Campbell, S. P. Pereira,D. A. Tuveson, B. K. Park, W. Greenhalf, R. Sutton, J. F. Timms, J. P. Neoptolemos, E. Costello,Decreased serum thrombospondin-1 levels in pancreatic cancer patients up to24 months prior to clinical diagnosis: Association with diabetes mellitus. Clin. Cancer Res.22, 1734–1743 (2016).

21. J. E. Mirus, Y. Zhang, C. I. Li, A. E. Lokshin, R. L. Prentice, S. R. Hingorani, P. D. Lampe, Cross-species antibody microarray interrogation identifies a 3-protein panel of plasmabiomarkers for early diagnosis of pancreas cancer. Clin. Cancer Res. 21, 1764–1771 (2015).

22. B. M. Nolen, R. E. Brand, D. Prosser, L. Velikokhatnaya, P. J. Allen, H. J. Zeh, W. E. Grizzle,Y. Huang, A. Lomakin, A. E. Lokshin, Prediagnostic serum biomarkers as earlydetection tools for pancreatic cancer in a large prospective cohort study. PLOS ONE 9,e94928 (2014).

23. M. Capello, L. E. Bantis, G. Scelo, Y. Zhao, P. Li, D. S. Dhillon, N. J. Patel, D. L. Kundnani,H. Wang, J. L. Abbruzzese, A. Maitra, M. A. Tempero, R. Brand, L. Brennan, E. Feng,I. Taguchi, V. Janout, M. A. Firpo, S. J. Mulvihill, M. H. Katz, S. M. Hanash, Sequential


validation of blood-based protein biomarker candidates for early-stage pancreaticcancer. J. Natl. Cancer Inst. 109 (2017).

24. A. Maitra, R. H. Hruban, Pancreatic cancer. Annu. Rev. Pathol. 3, 157–188 (2008).25. J. Kim, J. P. Hoffman, R. K. Alpaugh, A. D. Rhim, M. Reichert, B. Z. Stanger, E. E. Furth,

A. R. Sepulveda, C.-X. Yuan, K.-J. Won, G. Donahue, J. Sands, A. A. Gumbs, K. S. Zaret, AniPSC line from human pancreatic ductal adenocarcinoma undergoes early to invasivestages of pancreatic cancer progression. Cell Rep. 3, 2088–2099 (2013).

26. N. Bardeesy, K.-h. Cheng, J. H. Berger, G. C. Chu, J. Pahler, P. Olson, A. F. Hezel, J. Horner,G. Y. Lauwers, D. Hanahan, R. A. DePinho, Smad4 is dispensable for normal pancreasdevelopment yet critical in progression and tumor biology of pancreas cancer. Genes Dev.20, 3130–3146 (2006).

27. S. Jones, X. Zhang, D. W. Parsons, J. C.-H. Lin, R. J. Leary, P. Angenendt, P. Mankoo,H. Carter, H. Kamiyama, A. Jimeno, S.-M. Hong, B. Fu, M.-T. Lin, E. S. Calhoun, M. Kamiyama,K. Walter, T. Nikolskaya, Y. Nikolsky, J. Hartigan, D. R. Smith, M. Hidalgo, S. D. Leach, A. P. Klein,E. M. Jaffee, M. Goggins, A. Maitra, C. Iacobuzio-Donahue, J. R. Eshleman, S. E. Kern,R. H. Hruban, R. Karchin, N. Papadopoulos, G. Parmigiani, B. Vogelstein, V. E. Velculescu,K. W. Kinzler, Core signaling pathways in human pancreatic cancers revealed by globalgenomic analyses. Science 321, 1801–1806 (2008).

28. M. S. Pepe, Z. Feng, H. Janes, P. M. Bossuyt, J. D. Potter, Pivotal evaluation of the accuracyof a biomarker used for classification or prediction: Standards for study design. J. Natl.Cancer Inst. 100, 1432–1438 (2008).

29. M. S. Pepe, R. Etzioni, Z. Feng, J. D. Potter, M. L. Thompson, M. Thornquist, M. Winget,Y. Yasui, Phases of biomarker development for early detection of cancer. J. Natl. CancerInst. 93, 1054–1061 (2001).

30. E. Kolker, R. Higdon, W. Haynes, D. Welch, W. Broomall, D. Lancet, L. Stanberry, N. Kolker,MOPED: Model organism protein expression database. Nucleic Acids Res. 40,D1093–D1099 (2012).

31. V. Nanjappa, J. K. Thomas, A. Marimuthu, B. Muthusamy, A. Radhakrishnan, R. Sharma,A. Ahmad Khan, L. Balakrishnan, N. A. Sahasrabuddhe, S. Kumar, B. N. Jhaveri, K. V. Sheth,R. Kumar Khatana, P. G. Shaw, S. M. Srikanth, P. P. Mathur, S. Shankar, D. Nagaraja,R. Christopher, S. Mathivanan, R. Raju, R. Sirdeshmukh, A. Chatterjee, R. J. Simpson,H. C. Harsha, A. Pandey, T. S. Prasad, Plasma Proteome Database as a resource for proteomicsresearch: 2014 update. Nucleic Acids Res. 42, D959–D965 (2014).

32. C. Wu, C. Orozco, J. Boyer, M. Leglise, J. Goodale, S. Batalov, C. L. Hodge, J. Haase, J. Janes,J. W. Huss III, A. I. Su, BioGPS: An extensible and customizable portal for queryingand organizing gene annotation resources. Genome Biol. 10, R130 (2009).

33. I. Prassas, D. Brinc, S. Farkona, F. Leung, A. Dimitromanolakis, C. C. Chrystoja, R. Brand,V. Kulasingam, I. M. Blasutig, E. P. Diamandis, False biomarker discovery due to reactivityof a commercial ELISA for CUZD1 with cancer antigen CA125. Clin. Chem. 60,381–388 (2014).

34. K. Y. Bilimoria, D. J. Bentrem, C. Y. Ko, J. Ritchey, A. K. Stewart, D. P. Winchester,M. S. Talamonti, Validation of the 6th edition AJCC pancreatic cancer staging system:Report from the National Cancer Database. Cancer 110, 738–744 (2007).

35. V. Molina, L. Visa, C. Conill, S. Navarro, J. M. Escudero, J. M. Auge, X. Filella, M. A. Lopez-Boado,J. Ferrer, L. Fernandez-Cruz, R. Molina, CA 19–9 in pancreatic cancer: Retrospectiveevaluation of patients with suspicion of pancreatic cancer. Tumour Biol.33, 799–807 (2012).

36. L. Yan, S. Tonack, R. Smith, S. Dodd, R. E. Jenkins, N. Kitteringham, W. Greenhalf,P. Ghaneh, J. P. Neoptolemos, E. Costello, Confounding effect of obstructive jaundicein the interpretation of proteomic plasma profiling data for pancreatic cancer.J. Proteome Res. 8, 142–148 (2009).

37. M. Wagner, C. Redaelli, M. Lietz, C. A. Seiler, H. Friess, M. W. Büchler, Curative resection isthe single most important factor determining outcome in patients with pancreaticadenocarcinoma. Br. J. Surg. 91, 586–594 (2004).

38. S. Yachida, S. Jones, I. Bozic, T. Antal, R. Leary, B. Fu, M. Kamiyama, R. H. Hruban,J. R. Eshleman, M. A. Nowak, V. E. Velculescu, K. W. Kinzler, B. Vogelstein,C. A. Iacobuzio-Donahue, Distant metastasis occurs late during the genetic evolution ofpancreatic cancer. Nature 467, 1114–1117 (2010).

39. D. J. Brat, K. D. Lillemoe, C. J. Yeo, P. B. Warfield, R. H. Hruban, Progression of pancreaticintraductal neoplasias to infiltrating adenocarcinoma of the pancreas. Am. J. Surg.Pathol. 22, 163–169 (1998).

40. S. Nie, A. Lo, J. Wu, J. Zhu, Z. Tan, D. M. Simeone, M. A. Anderson, K. A. Shedden,M. T. Ruffin, D. M. Lubman, Glycoprotein biomarker panel for pancreaticcancer discovered by quantitative proteomics analysis. J. Proteome Res. 13,1873–1884 (2014).

41. P. Bornstein, S. Devarayalu, P. Li, C. M. Disteche, P. Framson, A second thrombospondingene in the mouse is similar in organization to thrombospondin 1 but does not respondto serum. Proc. Natl. Acad. Sci. U.S.A. 88, 8636–8640 (1991).

42. D. J. Good, P. J. Polverini, F. Rastinejad, M. M. Le Beau, R. S. Lemons, W. A. Frazier,N. P. Bouck, A tumor suppressor-dependent inhibitor of angiogenesis is immunologicallyand functionally indistinguishable from a fragment of thrombospondin. Proc. Natl. Acad.Sci. U.S.A. 87, 6624–6628 (1990).

12 of 13



http://stm.sciencem

agD

ownloaded from

43. T. Hawighorst, P. Velasco, M. Streit, Y.-K. Hong, T. R. Kyriakides, L. F. Brown, P. Bornstein,M. Detmar, Thrombospondin-2 plays a protective role in multistep carcinogenesis: Anovel host anti-tumor defense mechanism. EMBO J. 20, 2631–2640 (2001).

44. R. Sun, J. Wu, Y. Chen, M. Lu, S. Zhang, D. Lu, Y. Li, Down regulation ofThrombospondin2 predicts poor prognosis in patients with gastric cancer.Mol. Cancer 13, 225 (2014).

45. P. M. Bossuyt, J. B. Reitsma, D. E. Bruns, C. A. Gatsonis, P. P. Glasziou, L. M. Irwig, D. Moher,D. Rennie, H. C. W. de Vet, J. G. Lijmer; Standards for Reporting of Diagnostic Accuracy,The STARD statement for reporting studies of diagnostic accuracy: Explanation andelaboration. Clin. Chem. 49, 7–18 (2003).

46. S. T. Chari, C. L. Leibson, K. G. Rabe, J. Ransom, M. de Andrade, G. M. Petersen, Probabilityof pancreatic cancer following diabetes: A population-based study. Gastroenterology129, 504–511 (2005).

47. G. M. Petersen, Familial pancreatic adenocarcinoma. Hematol. Oncol. Clin. North Am. 29,641–653 (2015).

48. A. K. Rustgi, Familial pancreatic cancer: Genetic advances. Genes Dev. 28, 1–7 (2014).49. G. Aggarwal, K. G. Rabe, G. M. Petersen, S. T. Chari, New-onset diabetes in pancreatic cancer:

A study in the primary care setting. Pancreatology 12, 156–161 (2012).50. R. R. McWilliams, W. R. Bamlet, M. de Andrade, D. N. Rider, J. M. Cunningham,

G. M. Petersen, Nucleotide excision repair pathway polymorphisms and pancreaticcancer risk: Evidence for role of MMS19L. Cancer Epidemiol. Biomarkers Prev. 18,1295–1302 (2009).

51. W.-J. Qian, D. T. Kaleta, B. O. Petritis, H. Jiang, T. Liu, X. Zhang, H. M. Mottaz, S. M. Varnum,D. G. Camp II, L. Huang, X. Fang, W.-W. Zhang, R. D. Smith, Enhanced detection of lowabundance human plasma proteins using a tandem IgY12-SuperMix immunoaffinityseparation strategy. Mol. Cell. Proteomics 7, 1963–1973 (2008).

52. H. Chi, K. He, B. Yang, Z. Chen, R.-X. Sun, S.-B. Fan, K. Zhang, C. Liu, Z.-F. Yuan, Q. H. Wang,S. Q. Liu, M. Q. Dong, S.-M. He, pFind–Alioth: A novel unrestricted database searchalgorithm to improve the interpretation of high-resolution MS/MS data. J. Proteomics125, 89–97 (2015).

53. E. Cerami, J. Gao, U. Dogrusoz, B. E. Gross, S. O. Sumer, B. A. Aksoy, A. Jacobsen, C. J. Byrne,M. L. Heuer, E. Larsson, Y. Antipin, B. Reva, A. P. Goldberg, C. Sander, N. Schultz, ThecBio cancer genomics portal: An open platform for exploring multidimensional cancergenomics data. Cancer Discov. 2, 401–404 (2012).

54. J. Gao, B. A. Aksoy, U. Dogrusoz, G. Dresdner, B. Gross, S. O. Sumer, Y. Sun, A. Jacobsen,R. Sinha, E. Larsson, E. Cerami, C. Sander, N. Schultz, Integrative analysis of complexcancer genomics and clinical profiles using the cBioPortal. Sci. Signal. 6, pl1 (2013).


55. B. Li, C. N. Dewey, RSEM: Accurate transcript quantification from RNA-Seq data with orwithout a reference genome. BMC Bioinformatics 12, 323 (2011).

56. Mithat Gonen, Analyzing Receiver Operating Characteristic Curves with SAS (SAS Institute,2007).

Acknowledgments: We thank S. Vedula for preparing plasma samples for proteomics;H. Collins in the Penn Diabetes Center Biomarkers Core; Z.-F. Yuan for peptide searches usingpFind3.0; D. Balli for PDAC data curation from the TCGA database; P. Kanuparthi forimmunohistochemistry; R. Vonderheide, A. Rustgi, and J. Becker for comments on themanuscript; and E. Hulme for help with manuscript preparation. Funding: This work wassupported by NIH grant no. R37GM36477; the Abramson Cancer Center Pancreatic CancerTranslational Center for Excellence; the Institute for Regenerative Medicine at the University ofPennsylvania; NIH grant nos. P30DK050306 and its cell culture core (to K.S.Z.), U01CA21038(to G.M.P. and K.S.Z.), and P50CA102701 Mayo Clinic Specialized Programs of ResearchExcellence in Pancreatic Cancer (to G.M.P.); Department of Defense grant no. BC123187P1 (toB.A.G.); and NIH grant no. P30DK19525 supporting the Penn Diabetes Research Center BioassayCore. Author contributions: J.K. and K.S.Z. curated biomarker candidates from a previous study.W.R.B. and G.M.P. selected the patient and control populations, analyzed the data, and preparedfigures and tables. J.K. and G.D. analyzed TCGA data. J.K. performed the ELISAs blinded to sampleidentity. A.L.O., K.G.C., and S.C. advised on patient selection, statistics, and data analysis. J.K., X.-J.C.,and B.A.G. obtained mass spectrometry data. W.R.B., G.D., X.-J.C., and G.M.P. obtained andstatistically analyzed the mass spectrometry data. J.K. performed the immunohistochemistry.Competing interests: K.S.Z. has consulted for BetaLogics/J&J and RaNA Therapeutics. J.K. andK.S.Z. have a patent pending (application no. 61/837,358) for the dual marker panel entitled“Methods for Diagnosing Pancreatic Cancer.” The other authors declare that they have nocompeting interests.

Submitted 13 July 2016Resubmitted 16 December 2016Accepted 27 April 2017Published 12 July 201710.1126/scitranslmed.aah5583

Citation: J. Kim, W. R. Bamlet, A. L. Oberg, K. G. Chaffee, G. Donahue, X.-J. Cao, S. Chari, B. A. Garcia,G. M. Petersen, K. S. Zaret, Detection of early pancreatic ductal adenocarcinoma withthrombospondin-2 and CA19-9 blood markers. Sci. Transl. Med. 9, eaah5583 (2017).

.or

13 of 13


g/


blood markersDetection of early pancreatic ductal adenocarcinoma with thrombospondin-2 and CA19-9

A. Garcia, Gloria M. Petersen and Kenneth S. ZaretJungsun Kim, William R. Bamlet, Ann L. Oberg, Kari G. Chaffee, Greg Donahue, Xing-Jun Cao, Suresh Chari, Benjamin

DOI: 10.1126/scitranslmed.aah5583, eaah5583.9Sci Transl Med

with the marker CA19-9, boosts detection of the early stages of PDAC in high-risk human populations.cancer and control human plasma samples in a multiphase study. The authors report that, THBS2, in combinationcandidate markers of early-stage disease. The protein thrombospondin-2 (THBS2) was screened against 746 like state, enabling the reprogrammed cells to recapitulate human PDAC progression and revealing secreted

−. genetically reprogrammed late-stage human PDAC cells to a stem cellet aldetecting early-stage disease. Kim Pancreatic ductal adenocarcinoma (PDAC) has a dismal prognosis due to a lack of diagnostics for

Getting a head start on pancreatic cancer

ARTICLE TOOLS http://stm.sciencemag.org/content/9/398/eaah5583

MATERIALSSUPPLEMENTARY http://stm.sciencemag.org/content/suppl/2017/07/10/9.398.eaah5583.DC1

CONTENTRELATED

http://science.sciencemag.org/content/sci/368/6491/589.fullhttp://science.sciencemag.org/content/sci/364/6446/1156.fullhttp://science.sciencemag.org/content/sci/364/6446/1132.full

REFERENCES

http://stm.sciencemag.org/content/9/398/eaah5583#BIBLThis article cites 53 articles, 19 of which you can access for free

PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions

Terms of ServiceUse of this article is subject to the

registered trademark of AAAS. is aScience Translational MedicineScience, 1200 New York Avenue NW, Washington, DC 20005. The title

(ISSN 1946-6242) is published by the American Association for the Advancement ofScience Translational Medicine

of Science. No claim to original U.S. Government WorksCopyright © 2017 The Authors, some rights reserved; exclusive licensee American Association for the Advancement


http://stm.sciencem

ag.org/D

ownloaded from

http://stm.sciencemag.org/content/9/398/eaah5583

http://stm.sciencemag.org/content/suppl/2017/07/10/9.398.eaah5583.DC1

http://science.sciencemag.org/content/sci/364/6446/1132.full



http://stm.sciencemag.org/content/9/398/eaah5583#BIBL

http://www.sciencemag.org/help/reprints-and-permissions

http://www.sciencemag.org/about/terms-service


Documents

Detection of early pancreatic ductal adenocarcinoma with ... · pluripotent stem cell (iPSC)–like cell line (25). This iPSC-like cell line (designated as 10-22) can be propagated