desmoplasia in primary tumors and metastatic lesions of ... · biology of human tumors desmoplasia...

Biology of Human Tumors

Desmoplasia in Primary Tumors and MetastaticLesions of Pancreatic CancerClifford J.Whatcott1, Caroline H. Diep1, Ping Jiang2, Aprill Watanabe3, Janine LoBello3,Chao Sima4, Galen Hostetter5, H. Michael Shepard2, Daniel D.Von Hoff1, and Haiyong Han1

Abstract

Purpose: Pancreatic ductal adenocarcinoma (PDAC) is char-acterized by high levels of fibrosis, termed desmoplasia, which isthought to hamper the efficacy of therapeutics treating PDAC.Ourprimary focus was to evaluate differences in the extent of desmo-plasia in primary tumors and metastatic lesions. As metastaticburden is a primary cause for mortality in PDAC, the extent ofdesmoplasia in metastases may help to determine whether des-moplasia targeting therapeutics will benefit patients with late-stage, metastatic disease.

Experimental Design: We sought to assess desmoplasia inmetastatic lesions of PDAC and compare it with that of primarytumors. Fifty-three patients' primaries and 57 patients' metastaseswere stained using IHC staining techniques.

Results:Weobserved a significant negative correlation betweenpatient survival and extracellular matrix deposition in primary

tumors. Kaplan–Meier curves for collagen I showed mediansurvival of 14.6 months in low collagen patients, and 6.4 monthsin high-level patients (log rank, P < 0.05). Low-level hyaluronanpatients displayed median survival times of 24.3 monthsas compared with 9.3 months in high-level patients (log rank,P < 0.05). Our analysis also indicated that extracellular matrixcomponents, such as collagen and hyaluronan, are found in highlevels in both primary tumors and metastatic lesions. The differ-ence in the level of desmoplasia between primary tumors andmetastatic lesions was not statistically significant.

Conclusions: Our results suggest that both primary tumors andmetastases of PDAC have highly fibrotic stroma. Thus, stromal tar-geting agents have thepotential tobenefit PDACpatients, even thosewith metastatic disease. Clin Cancer Res; 21(15); 3561–8. �2015 AACR.

See related commentary by Olive, p. 3366

IntroductionPancreatic ductal adenocarcinoma (PDAC) is the fourth

leading cause of cancer-related death in the United States. Withthe 5-year survival rate at approximately 6%, novel therapeuticstrategies are desperately needed to enhance patient outcomes(1). PDAC is characterized by the development of extensivefibrosis at primary tumor sites (2). This fibrosis is termeddesmoplasia. In normal tissues, mechanisms common to des-moplasia may serve to aid in inflammation and resolution ofgranulation tissue in the wound-healing process. However, intumor tissues such as in PDAC, desmoplasia promotes tumordevelopment and inhibits drug penetration and uptake (3, 4).Metastatic disease burden, on the other hand, is one of the

primary causes of death in patients with PDAC. Metastaticlesions disrupt organ function, leading to hepatic failure, forexample, when pancreatic metastases colonize the liver (5, 6).Recent reports have suggested that targeting components of thedesmoplastic reaction may enhance the activity of therapeuticstargeting tumor cells (7–11). The concern remains, however, ofwhether desmoplasia-targeting therapeutics will offer clinicalbenefit for patients with metastatic disease. To this end, wesought to determine the impact of desmoplasia on patientoutcomes. In this report, we also sought to determine whethermetastatic lesions of PDAC are desmoplastic. To assess theextent of desmoplasia in these tissues, we looked at the expres-sion of multiple extracellular matrix components and proteins,as well as cellular morphology in tissues taken from a diversityof patients. To our knowledge, this is the first report on thedesmoplastic nature of pancreatic metastases to various sites.

In desmoplasia, the pancreatic stellate cell (PSC) transformsfrom a quiescent, lipid droplet-storing cell to that of an activated,smoothmuscle actin (SMA) expressing cell via TGFb and platelet-derived growth factor-mediated transcriptional programs, elicit-ing ECM synthesis and deposition (12, 13). TGFb plays a key rolein this activation through both autocrine and paracrine signalingmechanisms (14). Reports have suggested that PSC activation anddepositionof ECMs lead to increases in both tissue solid stress andtissue interstitial fluid pressures, both of which may mediatevascular compression and dysfunction (8, 15). Expansion of thestromal compartment and ECM deposition is thought to result inthe poor drug penetration observed in PDAC. Indeed, Olive andcolleagues observed marked decreases in the delivery of doxoru-bicin to tumor tissues in a genetically engineered mouse model

1Clinical Translational Research Division, The Translational GenomicsResearch Institute, Phoenix, Arizona. 2Halozyme Therapeutics, SanDiego, California. 3Integrated Cancer Genomics Division, The Transla-tionalGenomicsResearch Institute, Phoenix,Arizona. 4ComputationalBiology Division, The Translational Genomics Research Institute,Phoenix, Arizona. 5Laboratory of Analytical Pathology,The Van AndelResearch Institute, Grand Rapids, Michigan.

Note: Supplementary data for this article are available at Clinical CancerResearch Online (http://clincancerres.aacrjournals.org/).

Corresponding Authors: Clifford J. Whatcott, Translational Genomics ResearchInstitute, 445North Fifth Street, Suite 400. Phoenix, AZ85004. Phone: 602-343-8739; Fax: 602-343-8740; E-mail: [email protected]; and Haiyong Han,E-mail: [email protected]

doi: 10.1158/1078-0432.CCR-14-1051

�2015 American Association for Cancer Research.

ClinicalCancerResearch

www.aacrjournals.org 3561

on January 31, 2020. © 2015 American Association for Cancer Research. clincancerres.aacrjournals.org Downloaded from

Published OnlineFirst February 18, 2015; DOI: 10.1158/1078-0432.CCR-14-1051

http://clincancerres.aacrjournals.org/

(KPC) for pancreatic cancer, relative to adjacent normal tissues(16). Stromal expansion is accompanied by significant tissuehypovascularity—with as much as a 75% decrease in tissue vesseldensity—observed in both KPC tumor and human tumor tissuesections (16). The combination of these factors reduces both thecapability and availability of drug penetration in PDAC. Desmo-plasia is well documented in the pathogenesis of primary tumorsof PDAC. However, metastatic burden is also considered a majorcause of patient death (17).

Metastatic disease burden results in organ failure and patientdeath. In a study of over 500 autopsies performed at Roswell ParkMemorial Institute (Buffalo, NY) of patients who had succumbedto various cancer types, organ invasion resulting in organ failurewas found in 15%of patients (18). In a subsequent study, hepaticfailure resulting in patient death was associated with primarytumors in the pancreas (6). Systemic delivery of chemotherapeu-tics attempts to redress metastatic disease burden both in thepalliative and the neoadjuvant settings. It is not altogether obvi-ous, however, whether metastatic lesions are also desmoplastic,and thus physiologically chemoresistant. Duda and colleaguessuggest that malignant cells facilitate colonization of new tissuesby conscripting activated fibroblasts, tending the notion thatmetastatic lesions may be desmoplastic (19). Consistent with thenotion that desmoplasia inhibits drug uptake in primary tumorsof the pancreas, desmoplastic metastases then would also beexpected to be resistant to drug uptake by virtue of this samemechanism. If so, it is important that we understand the nature ofdesmoplasia in metastatic lesions if we are to enhance patientoutcomes.

Recent reports suggest that concomitant stromal targeting mayenhance therapeutic outcomes in PDAC patients. In the presentstudy, we observed a negative correlation betweenpatient survivaland components of the ECM. In addition, we found similar levelsof fibrosis inmetastatic lesions compared with unmatched tumorprimaries, as assessed by tissue levels of ECM components. Wetherefore conclude that stromal targeting approaches may also bebeneficial to patients with metastatic disease. These findingssuggest that research in stromal targeting may yield added ther-apeutic benefit in PDAC patients.

Materials and MethodsCase selection

PDAC tumor samples were obtained from formalin-fixed andparaffin-embedded (FFPE) tissue blocks collected for clinicalpurposes between the years 1999 and 2009 at Scottsdale Health-care. Tumor data regarding histologic subtype, grade, and TNMclassification were obtained from deidentified pathology reports.Demographic information, such as age at diagnosis, gender,histology, stage of disease, smoking history, and overall survival(OS), was also noted fromdeidentified, retained records at Scotts-dale Healthcare. The PDAC diagnosis, histologic subtype, andTNM staging were determined by a board-certified staff pathol-ogist at Scottsdale Healthcare. In addition, histomorphologicconfirmation was made by a second board-certified pathologistat TGen (G. Hostetter). Additional patient-matched primary andmetastasis samples were obtained fromYonsei University, CancerMetastasis Research Center (Seoul, South Korea) for use in thisreport. The PDAC diagnosis was confirmed on these samples atYonsei University by a staff pathologist before use in study.

Tissue microarray constructionTissue microarrays (TMA) were constructed with 1 mm diam-

eter cores punched from two to three distinct regions of each FFPEtumor block, compiled from both primary tumors andmetastaticlesions (43 samples are represented with 3 cores, and 7 samplesare represented with 2 cores). Tumor regions to be sampled werereviewed by study pathologist (G. Hostetter) and TMA construc-tion was performed using a Veridiam Semi-Automated TissueArrayer (VTA-100).

IHC staining and assessmentTMAblockswere sectioned at 5mmandattached to Fisherbrand

Superfrost plus slides (Fisher Scientific) by water flotation andovernight drying. Slides were subsequently deparaffinized withxylene, rehydrated through a series of graded ethanol baths, andantigen retrieved on-line using a BondMax autostainer (LeicaMicrosystems). All antibodies used in this study were obtainedfromAbcam. Slideswere visualized using either theBondPolymerRefine Detection kit (Leica) using 3,30-diaminobenzidine tetra-hydrochloride chromogen as substrate or 4plus goat anti-mouseIgG biotinylated secondary antibody (Biocare Medical) using3,30-diaminobenzidine tetrahydrochloride chromogen as sub-strate. Percent area scoring was performed using Adobe Photo-shop software (Adobe Systems). Slides were digitally scannedusing an Aperio Scanscope (Leica Microsystems) and importedinto Photoshop. Images were then converted to a CYMK colorformat, and the yellow channel was assessed for its intensity inmonochrome grayscale. Assessment of the yellow channelallowed for quantitation of the development of brown chromo-gen on our slides, without also inadvertently measuring blue orred coloring. A threshold was selected and used which identifiedpositive pixels on our slides. Percent area was calculated as apercentage of positive pixels relative to the total number of pixelsin view. Staining intensity was assessed independently by two ofthe authors (C.J. Whatcott and H. Han) upon review of multiplerandom snapshots taken from the digitally scanned slides. Hya-luronan (HA) and Collagen I (Col I) staining was assessed by useof a combination score, where percent area was multiplied by theintensity of the stain, for a maximum possible score of 300. Thiscombination scorewas used as a surrogate readout of the extent of

Translational Relevance

Pancreatic desmoplasia is thought to impede proper drugdelivery in pancreatic ductal adenocarcinoma therapeuticinterventions. Although desmoplasia in primary tumors hasbeen studied, the presence or absence of desmoplasia atmetastatic sites has not been investigated. The present studyexamines extracellular matrix deposition in sections fromclinical samples of both primary pancreatic tumors and fromvarious sites of metastasis. On the basis of Collagen I levels,our findings indicate that primary tumors and metastaticlesions show comparable levels of desmoplasia. Our findingsalso indicate a negative correlation between patient survivaland extracellular matrix (Collagen I or hyaluronan) deposi-tion. As stromal targeting has gained traction as a potentialapproach to enhancing chemotherapeutic interventions, ouranalysis provides further support for use of the approach, evenin patients with late-stage, metastatic disease.

Whatcott et al.

Clin Cancer Res; 21(15) August 1, 2015 Clinical Cancer Research3562




fibrosis, where indicated by HA and Col I positivity. Patients with"low HA" were defined where tissues exhibited a total HA histo-chemical score of 112 or less. "High HA" patient samples weredefinedwhere total HA histochemical staining was 113 or greater.Patients with "high Col I" had a total 185 or greater immuno-histochemical score, based on multiple images taken from FFPEslides. "LowCol I"wasdefinedaspatientswith a total 184or lowerIHC score. Total combination cutoff scores for both HA and Col Iwere selected following ROC curve analysis (20). Total scoresyielding the highest AUC with ROC analysis were used as thecutoff for the survival curves. ROC analysis in our cohort yieldedAUC for HA and Col I of 0.75 and 0.70, respectively. Total scoresused to separate low and high groups for a-SMA, Col III, and ColIV, were 77, 128, and 53, respectively. ROC analysis yielded AUCfora-SMA, Col III, andCol IV of 0.59, 0.79, and 0.68, respectively.

Pentachrome staining was achieved by following the manu-facturer's protocol (American Master Tech). Briefly, slides werefirst deparaffinized with xylene, rehydrated through a series ofgraded ethanol baths, and stained with Verhoeff's Elastic Stain.Following rinsing and differentiation steps, slides were placed inglacial acetic acid, and then in a 1% Alcian Blue solution. Fol-lowing similar rinsing procedures, slides were also exposed toCrocein Scarlet – Acid Fuchsin, 5% Phosphotungstic Acid, and anAlcoholic Saffron Solution.

Hyaluronan staining was achieved using the hyaluronan bind-ing protein (HABP) as has been described previously (21). Briefly,sectionswere deparaffinized, followedby rehydration. Slideswerefirst incubatedwith an avidin and biotin solution and then placedin a 0.3% hydrogen peroxide, diluted in methanol. Slides werethen incubated with 1% BSA as a blocking solution and detectedusing biotinylated HABP (2 mg/mL) on our autostainer. Afterwashing with PBS, HRP-streptavidin was applied at room tem-perature. Slides were then visualized using diaminobenzidinetetrahydrochloride treatment. Hyaluronan-specific staining wasconfirmed using recombinant human hyaluronidase (HalozymeTherapeutics) pretreatment before the above protocol on controlslides.

Statistical analysisSurvival and IHC staining data were analyzed using GraphPad

Prism 5 software (GraphPad Software). Survival data are pre-sented in Kaplan–Meier curves, in which the statistical signifi-cance of the median survival times was assessed using a log-ranktest. Statistical significance in our IHC data was analyzed with astandard, unpaired t test. Cox regression analyses were performedusing R, version 3.0.1, on clinical data to assess significance of thecorrelation between pathologic parameters and patient survival.Univariate analysis was followed by multivariate analysis wherewe assessed the statistical significance of our observations, adjust-ing for the confounding influence of these clinicopathologicfactors on tissue staining, by first selecting parameters that hadmet a threshold P value of less than 0.2. Parameters that had metthis threshold were analyzed using multivariate Cox regressionanalysis. P values of less than 0.05 were considered significant.

ResultsComponents of the extracellular matrix and their correlationwith survival in patients with pancreatic cancer

Considering recent reports on the impact of HA in the che-moresistance of tumors in KPC mice, we evaluated HA and other

ECMs in a cohort of human patient samples (8, 22).We sought todetermine whether any correlation existed between ECM depo-sition and patient survival. A total of 50 patients with pathologicdiagnoses of PDAC and with survival data were identified andincluded in this study.Median age of the patients includedwas 69years, with male and female individuals comprising 52% and48% of the cohort, respectively. Patients included were predom-inantly of Caucasian race. These and other demographic data aresummarized online (Supplementary Table S1). TNM staging andOS ranges are shown as well (Supplementary Table S2). Mediansurvival for all patients was 13.3 months. Individual patientfollow-up was maintained for over 100 months. In analyses ofcorrelation between survival and other clinicopathalogical para-meters by Cox regression analysis, statistical significance wasidentified with lymph node status and combined AJCC stagingstatus (Col I andCol IV), aswell aswell as age and combinedAJCCstaging status (HA). No statistically significant correlation wasobserved between survival and gender, smoking status, or histo-logic grading (Supplementary Table S3). It should be noted thatsignificant correlations have been observed in prior studies withlymph node status and patient survival (23). Furthermore, fulltreatment history was not available for analysis in this context.Multivariate analyses were also performed on IHC staining andother clinicopathological parameters versus survival. As shown inSupplementary Table S4, Col I, Col IV, and HA retain a significantassociation with survival (P ¼ 0.015, 0.034, and 0.033,respectively).

To quantify the tissue staining of HA and Collagen I in theprimary tumors of patients, we analyzed tissue sections accordingto percent area staining as well as staining intensity (24). Thesurvival curves of the high and low HA also demonstrated astatistically significant difference (log rank, P ¼ 0.037) with acalculated HR of 2.6. The difference in median survival betweenthe high and low HA curves was 15.0 months. Curves werestatistically compared using a log-rank test, demonstratingthat the curves for high and lowCol I (P¼ 0.013)were statisticallydifferent, with a calculated HR of 2.3, for higher levels of Col I.The difference in median survival between the high and low Col Icurves was 8.2months. Kaplan–Meier curves displayingOS basedon the expression of HA and Col I are shown in Fig. 1. Forcomparison, survival analysis for a-SMA (a marker for myofibro-blasts), Col III, or Col IV is also shown in Supplementary Fig. S1.Patients whose tumors have high a-SMA or Col III did notexhibit statistically significant correlations toward shorter survival(P ¼ 0.97 and 0.09 for a-SMA and Col III, respectively; Supple-mentary Fig. S1). A statistically significant difference betweenhigh and low Col IV patient groups was observed (P < 0.05).Because of the negative correlations that we observed, we decidedto expand our analysis and subsequently evaluated markers ofdesmoplasia histochemically in both primary tumors and meta-static lesions.

Desmoplastic features inprimary tumors andmetastatic lesionsThe desmoplasia and dense extracellular matrix of PDAC

primary tumors have been well documented (2). To comparethedesmoplastic features betweenprimary andmetastatic lesions,we first stained the tissues with Movat's pentachrome stain.Movat's pentachrome staining, which was originally developedto stain connective tissue, indicates the presence of collagens inyellow. As can be seen in Fig. 2, except for small areas ofinterlobular staining, the yellow collagen staining was largely

Fibrosis in Pancreatic Metastases

www.aacrjournals.org Clin Cancer Res; 21(15) August 1, 2015 3563




absent in adjacent normal tissues (Fig. 2, pentachrome stain). Onthe contrary, intense collagen staining was observed throughoutprimary tumor tissues. Hematoxylin and eosin (H&E) stainingshows acinar microarchitecture loss accompanied by basementmembrane degradation and escape of neoplastic cells fromductalcompartments in the primary tumor. Individual tumor glands aresurrounded by dense stromal cells, which correspond to intenseand diffuse collagen expression in the pentachrome staining,consistent with the notion that stromal cells are responsible forthe production of collagens. Intriguingly, the liver metastasis alsoshowed significant quantities of stromal cells (Fig. 2, H&E,metas-tasis). This stromal cell population also corresponded to increasedcollagen expression, as indicated in pentachrome-stained section.This observation indicates that PDACmetastases are also desmo-plastic, and appear to be very similar in the extent to that of theprimary tumor. To further examine desmoplasia in PDACmetas-tases and primary tumors, we next evaluated additional markersof desmoplasia.

Levels of extracellular matrix components in both primarytumors and metastatic lesions

Activated stellate cells (or myofibroblasts) are the primarysource of ECMs in pancreatic desmoplasia (25). We stained tissuesections witha-SMA and cytokeratin (an epithelial cellmarker) todistinguish cell types. As shown inFig. 2, acinar and epithelial cellsstained strongly for cytokeratins in all three tissue locations,adjacent normal, primary tumor, and liver metastases. PSCs andother stromal cells did not stain for cytokeratin. a-SMA stainingwasnot observed in acinar or epithelial cells, butwas observed in amajority of the stromal cells of primary tumors and metastases(Fig. 2, a-SMA). Similar to that seen in H&E staining, the tumorglandular structures are interwoven with a-SMA–positive stromalcells. Next, we sought to examine the expression levels of indi-vidualmajor ECMcomponents.We stained a tumor TMAcontain-ing 107 evaluable and unique patient primary tumors andmetas-tases sections, with antibodies specific to collagen I, III, IV, andHA. Figure 3 shows the representative images of the ECMproteins

Adjacent normal

Tumor

Metastasis

H&E Pentachrome Cytokeratins α-SMA

Figure 2.Histochemical staining for stromalcontent of adjacent normal, primarytumor, and metastatic lesions.Representative images are shown ofpatient tissue sections stained by bothstandard H&E and Movat'spentachrome staining techniques.Representative images of cytokeratinand a-SMA staining are also shown.Significant desmoplasia is seen in bothtumor and liver metastases, butnot in adjacent normal tissues.(scale bar ¼ 100 mm).

Figure 1.Kaplan–Meier survival curves for patientswith high and low levels of hyaluronan (HA) andCollagen I (Col I) in their primary tumors. Average staining positivity for HAin patient tissues was plotted in Kaplan–Meier curves which allowed us to compare "low" and "high" HA level groups. Median survival in the "high" HA groupwas 9.3 months, as opposed to 24.3 months for the "low" HA group. This difference amounted to a separation of 15.0 months in median survival between the"low" HA and "high" HA groups. Median survival in the "high" Col I group was 6.4 months, as opposed to 14.6 months for the "low" Col I group. The mediandifference was 8.2 months.

Whatcott et al.





in PDAC tissues. For collagen I, we observed significantly strongerstaining in the stromal compartment of both tumor and metas-tases (Fig. 3, Collagen I). Adjacent normal tissues appeared toexpress collagen I in the peri-acinar cell space. For collagen III, weobserved similar staining patterns in tumor and liver tissues tothat seen with collagen I. However, staining in adjacent normaltissues appeared to be limited to the peri-lobular space. Thestaining patterns and intensity of collagen IV andHAwere similarto those of collagen I.

To assess what differences, if any, exist in the desmoplasticreaction between primary tumors and metastatic lesions, wefirst sought to compare collagen I expression in patient matchedsamples from both sites (representative images shown in Fig. 4).In a small set of 7 patients, we saw comparable levels of collagen Iexpressed in both primary tumors and in tissues excised from thecorresponding metastases. For further analysis, we next sought tomeasure Collagen I expression in our TMA datasets. For quanti-tation, we assessed Collagen I staining by percentage area where

positive staining occurred. Using unmatched primary tumors andmetastatic lesions from our TMA, we measured the percentagearea inwhichCollagen I stained positively in a given snapshot of atumor section. All metastatic sites were pooled, and comparedwith the percentage positive staining observed in pooled primarytumors. The percentage areawas assessed frommultiple images ofeach patient tumor by two independent observers. Figure 4 (right)shows the distribution of the percentage area of positive Col Istaining in all primary and metastasis cases. The average percent-age area for the primary tumor cases is 74.0 and the averagepercentage area for the metastasis cases is 68.0. There is nostatistically significant difference in percentage area between thetwo groups. Col III, Col IV, HA, and a-SMA staining was alsoassessed in a similar fashion (Supplementary Fig. S2). Like Col I,Col III levels were not different in the metastatic lesions whencompared with primary tumors (64.5 vs. 58.9, n.s.). Col IV, HA,and a-SMA, however, were found to be present at comparable,though statistically different, levels in primary tumors versus

Tumor Metastasis

Sample 1

Sample 2

Sample 3

% A

rea

Figure 4.Representative images of patient matched pancreatic primary and metastatic lesions stained for collagen I, and quantitative assessment of stromal content of allprimary tumors and metastatic lesions. Patient tissue sections (left) from confirmed cases of ductal adenocarcinoma were stained for collagen I. Tumorsand theirmatchedmetastatic lesions from3patients are displayed. Primary tumors andmetastatic lesions show similar levels of collagen I staining. Metastatic lesionsshown were excised from the lymph node (sample 1), liver (sample 2), and mesentery (sample 3). (scale bar ¼ 100 mm) For comparison, sample scoringis shown in Table 2 (corresponding to samples 2, 3, and 4). In our quantitative assessment (right), the percent area of positive collagen I stainingwas plotted for eachcase and grouped into primary tumor and metastases for comparison.

Adjacent normal

Tumor

Metastasis

Collagen I Collagen III Collagen IV HA

Figure 3.IHC assessment of collagen and HAexpression in PDAC tissues. Patienttissues sections were stained withantibodies for collagens I, III, IV, andHA. Representative images are shownof adjacent normal, primary tumor,and metastatic lesions. (scalebar ¼ 100 mm).






metastatic lesions (42.4 vs. 33.9, 78.2 vs. 69.5, and 42.9 vs. 49.8,respectively). These results indicate that in PDAC, the metastaticlesions have similar levels of the desmoplastic reaction to those ofprimary tumors.

DiscussionIn our efforts to understand the role of the stroma in cancer, we

have endeavored to determine whether stromal targeting therapyhas the potential to increase patient survival in pancreatic cancer.However, one lingering concern has been a limited understandingof desmoplasia in metastatic lesions. Indeed, "will stroma target-ing therapy have any effect in late-stage, metastatic disease ifmetastatic lesions are not also desmoplastic?" To begin to addressthese concerns, in this study, we sought to answer the question,"what role, if any, does desmoplasia play in the outcomes ofpatients with pancreatic cancer?" Our results demonstrate thatECMs such as hyaluronan and collagen, surrogates for desmo-plastic activity, correlate negativelywith patient survival. Also, ourfindings show that there is no significant difference between theextent of desmoplasia in primary tumors and metastatic lesions.

To the question, "does the presence of desmoplastic ECMscorrelate with poor survival in pancreatic cancer," our data dem-onstrated that survival correlated negatively with the amount ofhyaluronan in tumors. Hyaluronan is a linear polysaccharide thatplays an important role in a range of cellular processes. Hyalur-onan-related signaling pathways (including CD44 mediated) arethought to be involved in biologic functions including cell pro-liferation, tissue hydration, cell motility, inflammation, angio-genesis, andmalignancy. Increases inhyaluronan levels have beenobserved in many different cancer types, including pancreatic,breast, gastric, and bladder cancer (26–29). As hyaluronan canact as a molecular sieve, enhancing residence time and reducingpenetration of macromolecules, we hypothesized that the levelsof ECMs like hyaluronan correlate with survival in pancreaticcancer. HABP interrogation of hyaluronan levels yielded histo-chemical images of rather uniform staining intensities. Weassessed patient tissues using positive staining area in a givenimage as a measure of the synthesis of hyaluronan. Here also,hyaluronan staining served in large part as a surrogate stain for theextent of desmoplasia in our tissue samples. Because of itsincreased levels in cancer, however, hyaluronan has already beenproposed as a suitable drug target (30). Indeed, interventions areunder current investigation for targeting stromal components likehyaluronan (clinicaltrials.gov, NCT01839487).

In addition to hyaluronan, Collagens I, III, and IV are expressedat high levels in pancreatic cancer. Receptor signalingmechanismsmediated by integrin subunits have been shown to be dysregu-lated. Though the subunit expression trends are not altogetherclear, integrin receptor combinations such asavb3,avb6, anda6b4,have been shown to enhance proliferation and tumorigenesis(31–33). However, as with hyaluronan, collagen levels arethought to negatively correlate with penetration of macromole-cules into tumor tissues as well. We hypothesized and interro-gated collagen I for correlations with patient survival. In a priorreport, Erkan and colleagues found a positive correlation withcollagen levels and patient survival (24). One possible explana-tion for this discrepancy betweenourfindingsmaybe found in thedifferent staining methods that were used for collagen. Erkan andcolleagues used the amphoteric dye, aniline blue, whereas weused collagen I-specific immunostaining. Nonspecific for the

collagen subtypes, their data suggest, taken together with ourfindings, that additional alterations in other collagen subtypesmay be occurring. Indeed, it may suggest significant loss of othercollagen subtypes to account for the overall loss Erkan andcolleagues observed in association with the poorly survivingcohort. An additional explanation for the discrepancy might alsobe found in prior treatments the patients received, resulting inoverall changes in collagen deposition. Further investigation willbe needed to make this determination.

To the question, "aremetastatic lesions also desmoplastic," ourfindings suggest that primary tumors and metastatic lesions bothshow significant levels of desmoplasia (Fig. 4). Our data suggestthat metastatic lesions of multiple sites, including liver, lung, andthe peritoneal cavity, display significant and comparable levels ofdesmoplasia, including high levels of collagens I, III, and IV, tothat found in primary pancreatic tumors (Fig. 4, right). Althoughexpanded cohorts of eachmetastatic site are needed to confirm theorgan site-specific consequences offibrosis, we have observed thattogether, the metastatic lesions are statistically indistinguishablein their percent area of fibrosis, as measured by Collagen I, fromprimary pancreatic tumors regardless of metastatic site (Table 1).Indeed, our small sample of matched pairs of patient tumorssupports these findings (Fig. 4, Table 2). Because we do not havedetailed treatment history from our patient cohort, it is possiblethat prior treatment could potentially bias the results as reported.Nonetheless, our data suggest that ECM retention in the context ofchemotherapeutic treatment correlates negatively with patientsurvival for HA, Col I, and Col IV. Together with studies showingnegative correlation between collagen levels and macromoleculepenetration, these findings suggest that patients with metastaticdisease may potentially benefit from stroma targeting therapies(34). The origin of demosplasia-inducing myofibroblasts in met-astatic lesions is debated, but appears to originate from theprimary tumor in somemodels (19). In the liver, where pancreaticcancer cells commonly metastasize, desmoplasia is also recog-nized as a physiologic response to tumors arising in situ. Whethermyofibroblasts are brought to the site of metastasis, or whetherthey are resident myofibroblasts of the liver, significant desmo-plasia in metastatic lesions suggests that metastasizing tumorsmaintain preestablished intercellular signaling mechanisms withstromal cells postmetastasis.

It is clear from prior studies that PDAC patients may die frommetastatic disease burden and subsequent organ failure(6, 17, 35). Are stromal targeted therapies expected to have atherapeutic benefit in metastatic disease? Or, would they insteadonly be expected to be effectivewhere the primary tumor is knownto be desmoplastic? Recent studies have pointed to the tumormicroenvironment as a potential area by which to achieve greaterpatient outcomes via chemotherapeutics. By targeting tumor–stromal interactions, the hope has been to enhance current regi-mens, or to identify newpotential strategies. It is clear that stromalinteractions promote primary tumor growth and pathogenesis.The current study demonstrates that metastatic lesions, too, are

Table 1. Breakdown of metastatic lesions, by site

Site of metastasis Cases % Total

Liver 26 46Lung 3 5Lymph node 7 12Other 21 37

Whatcott et al.





desmoplastic, suggesting that patients with metastatic diseasemay also benefit from stromal targeted therapies.

Disclosure of Potential Conflicts of InterestH.M. Shepard reports receiving commercial research grants from and has

ownership interest (including patents) in Halozyme Therapeutics, Inc. Nopotential conflicts of interest were disclosed by the other authors.

Authors' ContributionsConception and design: C.J. Whatcott, P. Jiang, D.D. Von Hoff, H. HanDevelopment of methodology: C.J. Whatcott, P. Jiang, H.M. ShepardAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): C.J. Whatcott, C.H. Diep, A. Watanabe, J. LoBello,G. Hostetter, D.D. Von HoffAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): C.J. Whatcott, P. Jiang, J. LoBello, C. Sima,H.M. Shepard, H. HanWriting, review, and/or revision of the manuscript: C.J. Whatcott, C.H. Diep,P. Jiang, A. Watanabe, J. LoBello, H.M. Shepard, D.D. Von Hoff, H. HanAdministrative, technical, or material support (i.e., reporting or organizingdata, constructing databases): C.H. DiepStudy supervision: D.D. Von Hoff, H. Han

AcknowledgmentsThe authors thank Dr. Sun Young Rha, Chief of the Division of Genomics

and Translational Research, Cancer Metastasis Research Center, at YonseiUniversity, and her team, for their assistance and contribution of tissuesamples for our study, and Amol Tembe at TGen, and Sha Tao at the VanAndel Research Institute, for their assistance in the statistical analysis of ourdata.

Grant SupportThis work is supported by a Stand Up To Cancer Dream Team Translational

Research Grant (SUC2-AACR-DT0509). StandUp To Cancer is a Program of theEntertainment Industry Foundation administered by the American Associationfor Cancer Research. This work was also supported in part by R01 (CA169281)and U01 (CA128454) grants from the NIH/NCI, by financial contributionsfrom the Katz Family Foundation, and by a grant from theNational Foundationfor Cancer Research (NFCR).

The costs of publication of this articlewere defrayed inpart by the payment ofpage charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Received May 7, 2014; revised January 28, 2015; accepted January 29, 2015;published OnlineFirst February 18, 2015.

References1. Siegel R, Ma J, Zou Z, Jemal A. Cancer statistics, 2014. CA Cancer J Clin

2014;64:9–29.2. Mahadevan D, Von Hoff DD. Tumor-stroma interactions in pancreatic

ductal adenocarcinoma. Mol Cancer Ther 2007;6:1186–97.3. Minchinton AI, Tannock IF. Drug penetration in solid tumours. Nat Rev

Cancer 2006;6:583–92.4. Netti PA, Berk DA, Swartz MA, Grodzinsky AJ, Jain RK. Role of extracellular

matrix assembly in interstitial transport in solid tumors. Cancer Res2000;60:2497–503.

5. Chiang AC, Massague J. Molecular basis of metastasis. N Engl J Med2008;359:2814–23.

6. Houten L, Reilley AA. An investigation of the cause of death from cancer.J Surg Oncol 1980;13:111–6.

7. Neesse A, Frese KK, Bapiro TE, Nakagawa T, Sternlicht MD, Seeley TW, et al.CTGF antagonism with mAb FG-3019 enhances chemotherapy responsewithout increasing drug delivery in murine ductal pancreas cancer. ProcNatl Acad Sci U S A 2013;110:12325–30.

8. Provenzano PP, Cuevas C, Chang AE, Goel VK, Von Hoff DD, HingoraniSR. Enzymatic targeting of the stroma ablates physical barriers totreatment of pancreatic ductal adenocarcinoma. Cancer Cell 2012;21:418–29.

9. Von Hoff DD, Ramanathan RK, Borad MJ, Laheru DA, Smith LS, WoodTE, et al. Gemcitabine plus nab-paclitaxel is an active regimen in patients

with advanced pancreatic cancer: a phase I/II trial. J Clin Oncol 2011;29:4548–54.

10. Jain RK. Normalizing tumor microenvironment to treat cancer: bench tobedside to biomarkers. J Clin Oncol 2013;31:2205–18.

11. Von Hoff DD, Ervin T, Arena FP, Chiorean EG, Infante J, Moore M, et al.Increased survival in pancreatic cancer with nab-paclitaxel plus gemcita-bine. N Engl J Med 2013;369:1691–703.

12. Yen TW, Aardal NP, Bronner MP, Thorning DR, Savard CE, Lee SP, et al.Myofibroblasts are responsible for the desmoplastic reaction surroundinghuman pancreatic carcinomas. Surgery 2002;131:129–34.

13. Faouzi S, Le Bail B, Neaud V, Boussarie L, Saric J, Bioulac-Sage P, et al.Myofibroblasts are responsible for collagen synthesis in the stroma ofhuman hepatocellular carcinoma: an in vivo and in vitro study. J Hepatol1999;30:275–84.

14. Akhurst RJ, Hata A. Targeting the TGFbeta signalling pathway in disease.Nat Rev Drug Discov 2012;11:790–11.

15. Stylianopoulos T, Martin JD, Chauhan VP, Jain SR, Diop-Frimpong B,Bardeesy N, et al. Causes, consequences, and remedies for growth-inducedsolid stress in murine and human tumors. Proc Natl Acad Sci U S A 2012;109:15101–8.

16. Olive KP, JacobetzMA,DavidsonCJ,GopinathanA,McIntyreD,HonessD,et al. Inhibition ofHedgehog signaling enhances delivery of chemotherapyin a mouse model of pancreatic cancer. Science 2009;324:1457–61.

Table 2. Comparison of Col I staining in matched samples

Sample Site Pathology % Area Col I Intensity Total score

1 Pancreas Ductal adenocarcinoma, moderately differentiated 38.3 3 114.9Lung Adenocarcinoma 10.2 3 30.7

2 Pancreas Ductal adenocarcinoma, well differentiated 23.4 3 70.3Lymph node Metastatic adenocarcinoma 11.5 3 34.6

3 Pancreas Ductal adenocarcinoma, well differentiated 17.6 3 52.7Liver Metastatic adenocarcinoma 18.0 3 54.1

4 Pancreas Ductal adenocarcinoma, moderately differentiated 46.5 3 139.5Mesentery Adenocarcinoma 53.8 3 161.5

5 Pancreas Ductal adenocarcinoma, moderately differentiated 31.4 3 94.3Lymph node Metastatic adenocarcinoma 26.6 3 79.7

6 Pancreas Ductal adenocarcinoma, moderately differentiated 68.1 3 204.2Umbilicus Metastatic adenocarcinoma 54.2 3 162.6

7 Pancreas Adenocarcinoma, unspecified 35.1 3 105.4Liver Metastatic adenocarcinoma 44.2 3 132.6






17. Yachida S, Jones S, Bozic I, Antal T, Leary R, Fu B, et al. Distant metastasisoccurs late during the genetic evolution of pancreatic cancer. Nature2010;467:1114–7.

18. Ambrus JL, Ambrus CM, Mink IB, Pickren JW. Causes of death in cancerpatients. J Med 1975;6:61–4.

19. Duda DG, Duyverman AM, Kohno M, Snuderl M, Steller EJ, Fukumura D,et al. Malignant cells facilitate lung metastasis by bringing their own soil.Proc Natl Acad Sci U S A 2010;107:21677–82.

20. Hanley JA, McNeil BJ. The meaning and use of the area under a receiveroperating characteristic (ROC) curve. Radiology 1982;143:29–36.

21. Jiang P, Li X, Thompson CB, Huang Z, Araiza F, Osgood R, et al. Effectivetargeting of the tumor microenvironment for cancer therapy. AnticancerRes 2012;32:1203–12.

22. Jacobetz MA, Chan DS, Neesse A, Bapiro TE, Cook N, Frese KK, et al.Hyaluronan impairs vascular function and drug delivery in amousemodelof pancreatic cancer. Gut 2013;62:112–20.

23. Slidell MB, ChangDC, Cameron JL, Wolfgang C, Herman JM, Schulick RD,et al. Impact of total lymph node count and lymph node ratio on stagingand survival after pancreatectomy for pancreatic adenocarcinoma: a large,population-based analysis. Ann Surg Oncol 2008;15:165–74.

24. Erkan M, Michalski CW, Rieder S, Reiser-Erkan C, Abiatari I, Kolb A, et al.The activated stroma index is a novel and independent prognostic markerin pancreatic ductal adenocarcinoma. Clin Gastroenterol Hepatol 2008;6:1155–61.

25. Apte MV, Haber PS, Darby SJ, Rodgers SC, McCaughan GW, Korsten MA,et al. Pancreatic stellate cells are activated by proinflammatory cytokines:implications for pancreatic fibrogenesis. Gut 1999;44:534–41.

26. Theocharis AD, Tsara ME, Papageorgacopoulou N, Karavias DD, Theo-charis DA. Pancreatic carcinoma is characterized by elevated content of

hyaluronan and chondroitin sulfate with altered disaccharide composi-tion. Biochim Biophys Acta 2000;1502:201–6.

27. Anttila MA, Tammi RH, TammiMI, Syrjanen KJ, Saarikoski SV, Kosma VM.High levels of stromal hyaluronan predict poor disease outcome inepithelial ovarian cancer. Cancer Res 2000;60:150–5.

28. Auvinen P, Tammi R, Parkkinen J, Tammi M, Agren U, Johansson R, et al.Hyaluronan in peritumoral stroma and malignant cells associates withbreast cancer spreading and predicts survival. Am J Pathol 2000;156:529–36.

29. Setala LP, Tammi MI, Tammi RH, Eskelinen MJ, Lipponen PK, AgrenUM, et al. Hyaluronan expression in gastric cancer cells is associatedwith local and nodal spread and reduced survival rate. Br J Cancer1999;79:1133–8.

30. Whatcott CJ, Han H, Posner RG, Hostetter G, Von Hoff DD. Targeting thetumor microenvironment in cancer: why hyaluronidase deserves a secondlook. Cancer Discov 2011;1:291–6.

31. Linder S, Castanos-Velez E, von Rosen A, Biberfeld P. Immunohistochem-ical expression of extracellular matrix proteins and adhesion molecules inpancreatic carcinoma. Hepatogastroenterology 2001;48:1321–7.

32. Grzesiak JJ, Ho JC, Moossa AR, Bouvet M. The integrin-extracellular matrixaxis in pancreatic cancer. Pancreas 2007;35:293–301.

33. Guo W, Giancotti FG. Integrin signalling during tumour progression.Nat Rev Mol Cell Biol 2004;5:816–26.

34. Magzoub M, Jin S, Verkman AS. Enhanced macromolecule diffusion deepin tumors after enzymatic digestion of extracellular matrix collagen and itsassociated proteoglycan decorin. FASEB J 2008;22:276–84.

35. Matsuda Y, Hagio M, Naito Z, Ishiwata T. Clinicopathological features of30 autopsy cases of pancreatic carcinoma. J Nippon Med Sch 2012;79:459–67.


Whatcott et al.




2015;21:3561-3568. Published OnlineFirst February 18, 2015.Clin Cancer Res Clifford J. Whatcott, Caroline H. Diep, Ping Jiang, et al. Pancreatic CancerDesmoplasia in Primary Tumors and Metastatic Lesions of

Updated version

10.1158/1078-0432.CCR-14-1051doi:

Access the most recent version of this article at:

Material

Supplementary

http://clincancerres.aacrjournals.org/content/suppl/2015/02/19/1078-0432.CCR-14-1051.DC1

Access the most recent supplemental material at:

Cited articles

http://clincancerres.aacrjournals.org/content/21/15/3561.full#ref-list-1

This article cites 35 articles, 13 of which you can access for free at:

Citing articles

http://clincancerres.aacrjournals.org/content/21/15/3561.full#related-urls

This article has been cited by 22 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected]

To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://clincancerres.aacrjournals.org/content/21/15/3561To request permission to re-use all or part of this article, use this link



http://clincancerres.aacrjournals.org/lookup/doi/10.1158/1078-0432.CCR-14-1051

http://clincancerres.aacrjournals.org/content/suppl/2015/02/19/1078-0432.CCR-14-1051.DC1

http://clincancerres.aacrjournals.org/content/21/15/3561.full#ref-list-1

http://clincancerres.aacrjournals.org/content/21/15/3561.full#related-urls

http://clincancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://clincancerres.aacrjournals.org/content/21/15/3561


desmoplasia in primary tumors and metastatic lesions of ... · biology of human tumors desmoplasia...

Documents