The American Journal of Pathology, Vol. 184, No. 3, March 2014

ajp.amjpathol.org

BIOMARKERS, GENOMICS, PROTEOMICS, AND GENE REGULATION

bIII-Tubulin Overexpression Is an Independent Predictorof Prostate Cancer Progression Tightly Linked to ERGFusion Status and PTEN DeletionMaria C. Tsourlakis,* Philipp Weigand,* Katharina Grupp,*y Martina Kluth,* Stefan Steurer,* Thorsten Schlomm,zx

Markus Graefen,z Hartwig Huland,z Georg Salomon,z Thomas Steuber,z Waldemar Wilczak,* Hüseyin Sirma,* Ronald Simon,*Guido Sauter,* Sarah Minner,* and Alexander Quaas*

From the Institute of Pathology,* the General, Visceral and Thoracic Surgery Department and Clinic,y the Martini-Clinic,z and the Prostate Cancer Center,Section for Translational Prostate Cancer Research, the Department of Urology,x University Medical Center Hamburg-Eppendorf, Hamburg, Germany

Accepted for publication

C

P

h

November 6, 2013.

Address correspondence toMaria C. Tsourlakis, M.D.,Institute of Pathology, Univer-sity Medical Center Hamburg-Eppendorf, Martinistrasse 52,20246 Hamburg, Germany.E-mail: m.tsourlakis@uke.de.

opyright ª 2014 American Society for Inve

ublished by Elsevier Inc. All rights reserved

ttp://dx.doi.org/10.1016/j.ajpath.2013.11.007

Evidence suggests that class III b-tubulin (bIII-tubulin) may represent a prognostic and predictivemolecular marker in prostate cancer. bIII-Tubulin expression was determined by IHC in 8179 prostatecancer specimens in a TMA format. Results were compared with tumor phenotype, biochemical recurrence,v-ets avian erythroblastosis virus E26 oncogene homolog (ERG) status, and deletions on PTEN, 3p13,5q21, and 6q15. bIII-Tubulin expression was detectable in 25.6% of 8179 interpretable cancers. HighbIII-tubulin expression was strongly associated with both TMPRSS2:ERG rearrangement and ERGexpression (P < 0.0001 each). High bIII-tubulin expression was tightly linked to high Gleason grade,advanced pT stage, and early prostate-specific antigen (PSA) recurrence in all cancers (P< 0.0001 each),but also in the subgroups of ERG-negative and ERG-positive cancers. When all tumors were analyzed, theprognostic role of bIII-tubulin expression was independent of Gleason grade, pT stage, pN stage, surgicalmargin status, and preoperative PSA. Independent prognostic value became even more evident if theanalysis was limited to preoperatively available features, such as biopsy specimen Gleason grade, pre-operative PSA, cT stage, and bIII-tubulin expression (P < 0.0001 each). bIII-Tubulin expression wasassociated with PTEN (P < 0.0001) when all tumors were analyzed, but also in the subgroups of ERG-negative and ERG-positive cancers. bIII-Tubulin expression is an independent prognostic parameter.The significant associations with key genomic alterations of prostate cancer, such as TMPRSS2:ERG fusionsand PTEN deletions, suggest interactions with several pivotal pathways involved in prostate cancer.(Am J Pathol 2014, 184: 609e617; http://dx.doi.org/10.1016/j.ajpath.2013.11.007)

This study was supported by German Cancer Aid grant 109505.M.C.T. and P.W. contributed equally to this work.Disclosures: None declared.

Prostate cancer is the most common malignancy in men inWestern societies.1Most prostate cancers behave in an indolentmanner, but a subset is highly aggressive.2 Despite recent ad-vances in research, the only established pretreatment prognosticparameters include Gleason grade and tumor extent on biopsyspecimens, preoperative prostate-specific antigen (PSA), andclinical stage. Because these data are statistically powerful butnot sufficient for optimal individual treatment decisions, abetter understanding of the biological features of the diseasemay eventually lead to better prognostic biomarkers.

Microtubules are multifunctional cytoskeletal proteinsinvolved in crucial cellular roles, including maintenanceof cell shape, intracellular transport, meiosis, and mitosis.

stigative Pathology.

.

Microtubules are composed of polymers of a- and b-tubulinheterodimers. Both a- and b-tubulins exist as multiple iso-types with a complex pattern of distribution among differenttissues, as previously reviewed.3 Class III b-tubulin (bIII-tubulin; alias TUBB3) is a microtubule protein, normallyexpressed in cells of neuronal origin.4 Expression of bIII-tubulin has also been described in several extraneuronal cells,such as melanocytes, spermatozoa, follicular lymphoid cells,and neuroendocrine cells of the fetal respiratory tract

Table 1 Clinicopathological Features of 11,152 Arrayed ProstateCancers

Features

No. of patients

Study cohort onTMA (n Z 11,152)

Biochemical relapse amongcategories (n Z 1824)

Follow-up (months)Mean 53.4Median 36.8Age (years)<50 318 4950e60 2768 46060e70 6548 1081>70 1439 232

Pretreatment PSA (ng/mL)<4 1407 1424e10 6735 82710e20 2159 521>20 720 309

Tumor stagepT2 7370 570pT3a 2409 587pT3b 1262 618pT4 63 49

Gleason grade�3 þ 3 2859 1933 þ 4 6183 8494 þ 3 1565 573�4 þ 4 482 208

Lymph node metastasispN0 6117 1126pNþ 561 291

Surgical marginNegative 8984 1146Positive 1970 642

Data are given as number of patients unless otherwise indicated.Numbers do not always add up to 11,152 in the different categoriesbecause of cases with missing data.

Figure 1 Representative images of bIII-tubulin immunostaining. A:Negative staining in prostate cancer and positive staining in nerves. B:Positive staining in prostate cancer.

Tsourlakis et al

epithelium.5 It was found overexpressed in several solid tu-mors, including nonesmall cell lung cancer,6 ovarian can-cer,7,8 urothelial carcinoma of the bladder,9 and head andneck squamous cell carcinoma.10 High levels of bIII-tubulinexpression were also described to be associated with poorclinical outcome in various cancers, including nonesmall celllung cancer, ovarian cancer, and urothelial carcinoma of thebladder.6,7,9 Other studies described a link between bIII-tubulin overexpression and reduced response to taxane-basedmicrotubule-targeting anticancer drugs.8,11e13

There are several lines of evidence suggesting that bIII-tubulin may represent a suitable prognostic and, perhaps,predictive molecular marker in prostate cancer. Studiesanalyzing 25814 and 28415 prostate cancers treated by radicalprostatectomy suggested associations with unfavorabletumor phenotype and biochemical relapse.14,15 In prostatecancer cell lines, functional overexpression or knockdown ofbIII-tubulin modulated sensitivity to taxane medication.14

In the present study, we used our pre-existing prostatecancer TMA16 containing 11,152 prostate cancer specimens,

610

including various molecularly defined subgroups, to inves-tigate the clinical relevance of bIII-tubulin and to comparebIII-tubulin with other molecular and prognostic features.The results of this study show that high bIII-tubulinexpression linked to PTEN loss and v-ets avian erythro-blastosis virus E26 oncogene homolog (ERG) activation is astrong and independent predictor of early PSA recurrence inprostate cancers.

Materials and Methods

Patients

Radical prostatectomy specimens were available from 11,152patients with prostatic adenocarcinomas undergoing surgerybetween 1992 and 2011 at the Department of Urology and theMartini Clinics at the University Medical Center Hamburg-Eppendorf (Hamburg, Germany). Follow-up data were avail-able for a total of 9695 patients with amedian follow-up of 36.8months (range, 1 to 228months) (Table 1). None of the patientsreceived neo-adjuvant or adjuvant therapy. The PSA valueswere measured after surgery, and PSA recurrence was definedas a postoperative PSA of 0.2 ng/mL and increasing at first ofappearance. All prostate specimens were analyzed according toa standard procedure, including a complete embedding of theentire prostate for histological analysis.17 The TMAmanufacturing process was described earlier in detail.18 Inshort, one 0.6-mm core was taken from a randomly selectedtumor block containing a particularly large tumor massreflecting the Gleason grade of the pathological diagnosis fromeach patient. The tissues were distributed among 24 TMAblocks, each containing 144 to 522 tumor samples. For internalcontrols, each TMA block also contained various control tis-sues, including normal prostate tissue. The molecular databaseattached to this TMA contained results on ERG expression in9628, ERG break-apart fluorescence in situ hybridization(FISH) analysis in 6106,19 and deletion status of CHD1 (5q21)in 3022 (blocks 1 to 7 of the TMA),20MAP3K7 (6q15) in 3528(blocks 1 to 10 of the TMA),21PTEN (10q23) in 6115 (blocks 1to 24 of the TMA),22 and FOXP1 (3p13) in 1290 cancers23

(blocks 1 to 7 of the TMA). The use of tissues and clinicaldata were according to the Hamburger Krankenhaus Gesetz(x12) and approved by our local ethical committee.

ajp.amjpathol.org - The American Journal of Pathology

Figure 2 Relationship of bIII-tubulin expression with ERG fusion probed by IHC and FISH (A). bIII-tubulin expression versus PTEN, 3p13, 6q15, and 5q21deletions probed by FISH analysis in all cancers (B), ERG-negative cancers (C), and ERG-positive cancers (D).

bIII-Tubulin in Prostate Cancer

IHC Data

Freshly cut TMA sections were immunostained on 1 dayand in one experiment. Primary antibody specific for bIII-tubulin (rabbit monoclonal antibody, dilution 1:150; Epi-tomics Inc., Burlingame, CA) was applied, and slides were

Table 2 Associations between bIII-Tubulin Expression Resultsand Cancer Phenotype in All Cancers

ParameterNo.Evaluable

IHC result (%)

P valueNegative Weak Moderate Strong

All cancers 8179 74.37 17.73 5.34 2.56Tumor stagepT2 5195 79.5 15.32 3.93 1.25pT3a 1905 69.82 20.1 6.56 3.52pT3b 988 57.49 25.2 9.82 7.49pT4 52 51.92 26.92 15.38 5.77 <0.0001

Gleason grade�3 þ 3 1893 83.94 12.89 2.8 0.373 þ 4 4647 74.26 18.31 5.23 2.194 þ 3 123 64.39 21.22 8.7 5.69�4 þ 4 360 60.28 23.61 7.78 8.33 <0.0001

Lymph node metastasisN0 4624 72.28 18.92 5.8 3.01Nþ 438 53.88 28.31 11.42 6.39 <0.0001

Surgical marginNegative 6449 75.62 16.92 5.29 2.17Positive 1582 69.6 20.73 5.69 3.98 <0.0001

The American Journal of Pathology - ajp.amjpathol.org

deparaffinized and exposed to heat-induced antigen retrievalfor 5 minutes in an autoclave at 121�C in pH 7.8 Tris-EDTA-citrate buffer. Bound antibody was then visualizedusing the EnVision Kit (Dako, Glostrup, Denmark). Stain-ing intensity seen in nerves and axons served as a positivecontrol, as previously described.14 The staining intensity (0,1, 2, 3) and the fraction of positive tumor cells wererecorded for each tissue spot. A final score was built fromthese two parameters according to the following, as previ-ously described24,25: negative scores had a staining intensityof 0; weak scores, a staining intensity of 1 in �70% oftumor cells or 2 in �30% of tumor cells; moderate scores, astaining intensity of 1 in >70% of tumor cells, a stainingintensity of 2 in >30% and �70% of tumor cells, or astaining intensity of 3 in �30% of tumor cells; and strongscores, a staining intensity of 2 in >70% of tumor cells or astaining intensity of 3 in >30% of tumor cells.

Statistics

Statistical calculations were performed with JMP 9 soft-ware (SAS Institute Inc., Cary, NC). Contingency tablesand the c2 test were performed to search for associationsbetween molecular parameters and tumor phenotype. Sur-vival curves were calculated according to Kaplan-Meier.The log-rank test was applied to detect significant sur-vival differences between groups. Cox proportional

611

Table 3 Associations between bIII-Tubulin Expression Resultsand ERG-Negative Cancer Phenotype

ParameterNo.evaluable

IHC result (%)

P valueNegative Weak Moderate Strong

All cancers 4084 84.4 11.17 3.16 1.27Tumor stage

pT2 2694 88.72 8.69 2.12 0.48pT3a 861 82.11 12.78 3.48 1.63pT3b 494 65.38 21.66 7.89 5.06pT4 22 68.18 18.18 13.64 0 <0.0001

Gleason grade�3 þ 3 883 92.53 5.44 1.81 0.233 þ 4 2291 85.38 11 2.71 0.924 þ 3 668 76.35 15.87 5.24 2.54�4 þ 4 225 66.67 21.78 6.22 5.33 <0.0001

Lymph node metastasisN0 2364 83.38 11.93 3.34 1.35Nþ 216 60.65 25.93 9.26 4.17 <0.0001

Surgical marginNegative 3226 85.21 10.57 3.07 1.15Positive 781 80.92 13.57 3.71 1.79 0.0308

Table 4 Associations between bIII-Tubulin Expression Resultsand ERG-Positive Cancer Phenotype

ParameterNo.evaluable

IHC result (%)

P valueNegative Weak Moderate Strong

All cancers 3259 60.48 26.36 8.71 4.45Tumor stagepT2 1934 65.62 24.77 7.19 2.43pT3a 880 56.82 27.61 9.77 5.8pT3b 404 45.54 30.45 13.12 10.89pT4 23 39.13 34.78 13.04 13.03 <0.0001

Gleason grade�3 þ 3 728 72.8 21.98 4.67 0.553 þ 4 1929 59.82 27.58 8.76 3.844 þ 3 468 47.65 28.21 13.68 10.47�4 þ 4 110 46.36 25.45 11.82 16.36 <0.0001

Lymph node metastasisN0 1847 57.34 27.72 9.47 5.47Nþ 185 46.49 30.27 15.14 8.11 0.0137

Surgical marginNegative 2536 62.34 25.24 8.68 3.75Positive 668 54.49 29.94 8.83 6.74 0.0002

Tsourlakis et al

hazards regression analysis was performed to test the sta-tistical independence and significance between patholog-ical, molecular, and clinical variables. Separate analyseswere performed using different sets of parameters availableeither before or after prostatectomy.

Results

bIII-Tubulin Expression in Prostate Cancer

A total of 8179 (73.3%) of tumor samples were interpretablein our TMA analysis. Reasons for noninformative cases(2973 spots; 26.7%) included lack of tissue samples orabsence of unequivocal cancer tissue in the TMA spot. bIII-Tubulin immunostaining was predominantly localized in thecytoplasm of invasive prostate cancers and, in some rarecases, in the normal-appearing prostate epithelium frompatients with cancer. Positive bIII-tubulin immunostainingwas seen in 2096 (25.6%) of our 8179 interpretable prostatecancers and was considered weak in 17.7%, moderate in5.3%, and strong in 2.5% of cancers. Representative imagesof negative and positive bIII-tubulin immunostainings aregiven in Figure 1.

Association with TMPRSS2:ERG Fusion Status and ERGProtein Expression

To evaluate whether bIII-tubulin expression is associatedwith ERG status in prostate cancers, we used data from pre-vious studies.19 Data on TMPRSS2:ERG fusion status, ob-tained by FISH, were available from 4673 patients, and byimmunohistochemistry (IHC), from 7343 patients. Data onboth ERG FISH and IHC were available from 5942 cancers,and an identical result (ERG IHC positive and break by FISH)

612

was found in 5670 (95.4%) of 5942 cancers. High-levelbIII-tubulin staining was tightly linked to TMPRSS2:ERGrearrangement and ERG expression in prostate cancers(P < 0.0001 each) (Figure 2A). For example, moderate orstrong bIII-tubulin immunostaining was seen in 14.4% ofcancers with TMPRSS2:ERG fusion detected by FISH butfound in only 4.5% of cancers without such rearrangements(P < 0.0001).

Associations with Tumor Phenotype

Increased bIII-tubulin expression was significantly linked tohigh Gleason grade, advanced pathological tumor stage, andpositive nodal status, if all tumors were jointly analyzed(P < 0.0001 each) (Table 2), but also in the subgroup ofERG-negative cancers (P < 0.0001 each) (Table 3) andERG-positive cancers [P < 0.0001 for pathological tumorstage (pT) and Gleason grade, and P Z 0.0137 for nodalstatus] (Table 4).

Associations with Other Key Genomic Deletions

Earlier studies had provided evidence for distinct molecularsubgroups of prostate cancers defined by TMPRSS2:ERGfusions and several genomic deletions. We and others haddescribed a strong link of PTEN and 3p13 deletions to ERGpositivity and of 5q21 and 6q15 deletions to ERG neg-ativity.21,22,26e28 To study, whether bIII-tubulin expressionmight be particularly linked to one of these genomic deletions,bIII-tubulin data were comparedwith pre-existing findings onPTEN (10q23), FOXP1 (3p13),MAP3K7 (6q15), and CHD1(5q21) deletions. In the analysis of all tumors,PTEN deletionswere significantly linked to high bIII-tubulin expression(P< 0.0001) (Figure 2B). These associations were retained in

ajp.amjpathol.org - The American Journal of Pathology

bIII-Tubulin in Prostate Cancer

ERG-negative (P < 0.0001) (Figure 2C) and ERG-positive(P < 0.0001) (Figure 2D) cancers. All other deletionsanalyzed were unrelated to bIII-tubulin expression both in allcancers and in subgroup analyses.

Associations with PSA Recurrence

Follow-up data were available for 7192 patients with inter-pretable bIII-tubulin immunostaining on the TMA. The suit-ability of our follow-up data is confirmed by the strongassociation between Gleason grade and early PSA recurrence(Figure 3A). There was a statistically significant associationbetween high bIII-tubulin expression and early PSA recur-rence if all tumors were analyzed (P < 0.0001) (Figure 3B),but also in the subgroups of ERG-negative, ERG-positive,PTEN-nondeleted, and PTEN-deleted cancers (P < 0.0001each) (Figure 3, CeF).

Figure 3 A: Relationship of Gleason grade with biochemical recurrence (n Zchemical recurrence in all prostate cancers (n Z 7192) (B), ERG-negative prostaPTEN-nondeleted cancers (n Z 3222) (E), and PTEN-deleted cancers (n Z 773)

The American Journal of Pathology - ajp.amjpathol.org

Multivariate Analysis

Four multivariate analyses were performed evaluating theclinical relevance of bIII-tubulin expression in different sce-narios (Table 5). Scenario 1 was using all postoperativelyavailable parameters, including pathological tumor stage,pathological lymph node status (pN), surgical margin status,preoperative PSA value, and pathological Gleason grade ob-tained after the morphological evaluation of the entire resectedprostate. Scenario 2 was using all postoperatively availableparameters, with exception of nodal status. The rationale forthis approach was that the indication and extent of lymph nodedissection is not standardized in the surgical therapy of pros-tate cancer and that excluding pN in multivariate analysis canmarkedly increase case numbers. Two additional scenarioshad to purpose to model the preoperative situation. Scenario 3included bIII-tubulin expression, preoperative PSA, clinical

9682). Association of bIII-tubulin immunostaining intensity with bio-te cancers (n Z 3557) (C), ERG-positive prostate cancers (n Z 2870) (D),(F).

613

Table 5 Multivariate Analysis

Parameter ScenarioNvalue

P value (RR)

Gleason, radicalprostatectomy

PSA,preoperative

Pathologicaltumor stage(pT)

Lymph nodestatus (pN)

Surgical margin(R)

Gleason,preoperativebiopsy

Clinical tumorstage (cT)

bIII-Tubulinexpressionon TMA

�4 þ 4 vs�3 þ 3 20 vs <4 ng/mL pT4 vs pT2 pNþ vs pN0 R1 vs R0

�4 þ 4 vs�3 þ 3 cT3a vs cT1c

Strong vsnegative

All cancers 1 4323 <0.0001 (2.4) <0.0001 (1.75) <0.0001 (4.45) <0.0001 (1.51) <0.0001 (1.36) d d 0.0008 (1.68)

2 7027 <0.0001 (6.09) <0.0001 (2.14) <0.0001 (5.21) d <0.0001 (1.44) d d <0.0001 (1.88)

3 6882 <0.0001 (12.55) <0.0001 (3.25) d d d d <0.0001 (2.09) <0.0001 (1.61)

4 6774 d <0.0001 (4.02) d d d <0.0001 (4.16) <0.0001 (1.46) <0.0001 (2.5)

ERGnegative

1 2176 <0.0001 (4.01) 0.0002 (1.73) <0.0001 (5.03) <0.0001 (1.77) 0.0296 (1.24) d d 0.1219 (1.82)

2 3475 <0.0001 (6.3) <0.0001 (2.03) <0.0001 (7.25) d 0.0021 (1.31) d d 0.0071 (2.35)

3 3435 <0.0001 (11.79) <0.0001 (3.05) d d d d <0.0001 (1.6) 0.0005 (2.9)

4 3387 d <0.0001 (3.82) d d d <0.0001 (3.77) <0.0001 (1.41) 0.0001 (3.2)

ERGpositive

1 1742 <0.0001 (7.02) 0.0079 (1.66) <0.0001 (4.74) 0.1016 (1.25) 0.0042 (1.35) d d 0.0181 (1.65)

2 2802 <0.0001 (7.48) 0.0002 (1.92) <0.0001 (5.21) d 0.0001 (1.43) d d 0.003 (1.76)

3 2719 <0.0001 (16.51) <0.0001 (3.2) d d d d <0.0001 (1.84) 0.0012 (1.79)

4 2671 d <0.0001 (3.70) d d d <0.0001 (4.58) <0.0001 (1.6) <0.0001 (2.31)

d, parameter not included in the respective scenario; RR, risk ratio.

Tsourlakis et al

tumor stage (cT stage), and Gleason grade obtained on theprostatectomy specimen. Because postoperative determinationof a tumor Gleason grade is better than the preoperativelydetermined Gleason grade (subjected to sampling errors andconsequently under-grading in more than one third of cases29),another multivariate analysis was added. Scenario 4 used thepreoperative Gleason grade obtained on the original biopsyspecimen, which was combined with preoperative PSA, cTstage, and bIII-tubulin expression.

If all tumors were analyzed, all scenarios suggest a ten-dency of bIII-tubulin to represent an independent predictorof prognosis. Separate analysis of ERG-positive and ERG-negative cancers revealed that bIII-tubulin expression hasindependent prognostic relevance in ERG-positive cancersin all scenarios. In ERG-negative cancers, bIII-tubulin wasidentified as an independent predictor of prognosis in sce-narios 2, 3, and 4.

Discussion

The results of our study identify high bIII-tubulin expres-sion as a strong prognostic feature in prostate cancers. TheIHC experiments revealed positive bIII-tubulin staining in25.6% of prostate cancers. Two earlier studies analyzing28415 and 25814 prostate cancers on a TMA by IHC hadreported bIII-tubulin positivity in 11.6%15 and 16.7%14 ofprostate cancers. These authors had used three 1-mm15 andfour 0.6-mm14 cores per cancer in their studies. Because thefrequency of bIII-tubulin expression detected in our studyusing TMAs constructed from one 0.6-mm core per patient issomewhat higher than in these studies, a methodological re-striction related to our TMA approach is unlikely. By usingthe same TMA or smaller subsets of it, we had earlierreproduced the substantial prognostic impact of all previouslywell-established prognostic markers, such as PTEN22 orTP5317 inactivation and the Ki67 labeling index.30 We had

614

also used this TMA platform for the successful identificationof various other prognostic features in prostate cancer.24,31e33

bIII-Tubulin immunostaining was significantly associatedwith unfavorable tumor phenotype and early PSA recurrencein our patients. This was in line with earlier reports describinga significant association between high bIII-tubulin expres-sion and high Gleason score,14,15 advanced pathologicaltumor stage,14 positive lymph node status,14 and high risk forbiochemical recurrence.14 Because bIII-tubulin renders themicrotubules instable and, therefore, more dynamic,34,35 itappears conceivable that bIII-tubulin contributes to tumoraggressiveness by enhancing cell motility and migrationrequired for tumor growth and metastasis. Cell culture ex-periments revealed that drugs inhibiting cell migration areineffective in the presence of high bIII-tubulin levels.36

Many patients included in our project in conjunction withan extensive molecular database attached to our cohortenabled us to further expand our analysis to molecularlydefined subgroups. More than half of prostate cancers,particularly in young patients, carry gene fusions linking theandrogen-regulated TMPRSS2 with transcription factors ofthe ETS family.37,38 As a result of this rearrangement, theexpression of ERG becomes androgen regulated andmassively overexpressed. Our data demonstrate that highbIII-tubulin expression was markedly more frequent inERG-positive than in ERG-negative cancers. Finding thisassociation by two independent approaches for ERG fusiondetection (IHC/FISH) largely excludes a false-positive as-sociation, caused by nonfunctional tissue. This associationsuggests a selection advantage for tumor cells withexpression of both bIII-tubulin and ERG and raises thepossibility that both genes might act cooperatively. In fact, arole of ERG for cell motility has been suggested before.39,40

In endothelial cells, ERG increases cell migration andlamellipodia formation by up-regulation of the histonedeacetylase HDAC6, which is required for tubulin acetyla-tion and consequent destabilization of microtubules.39 It

ajp.amjpathol.org - The American Journal of Pathology

bIII-Tubulin in Prostate Cancer

might be speculated that ERG and bIII-tubulin jointly affectcell motility through enhancement of microtubule dynamics.

A further aim of this study was to determine whether bIII-tubulin expression is linked to key genomic deletions knownto be associated with ERG fusion status. Certain chromo-somal deletions are tightly linked to either positive or nega-tive ERG status. In particular, deletions at 3p13 and 10q23were found to be associated with ERG-positive cancers,whereas 5q21 and 6q15 were linked to ERG-negative can-cers.21,22,26e28 In earlier studies, we had identified bio-markers that were significantly linked to some of thesegenomic defined subgroups. For example, CRISP3 over-expression is strongly associated with PTEN-deleted, ERG-positive prostate cancer.41

A strong link between bIII-tubulin expression and pres-ence of PTEN deletions was found in this study, which wasindependent from the known association between PTENdeletions and ERG fusions. This suggests that PTEN lossmight provide a selection advantage for tumor cells withelevated bIII-tubulin levels. A functional link between PTENloss and microtubule architecture has been shown recently.Vitolo et al42 reported that PTEN inactivation inducestubulin-based microtentacles that facilitate cell reattachmentand suggested that such cytoskeletal changes could have animpact on tumor biological features. It remains, however,speculative if bIII-tubulin could also participate in thismechanism. Alternatively, aberrant expression of bIII-tubulin in PTEN-deleted prostate cancer cells might berelated to tumor dedifferentiation. For example, a previousstudy has shown that PTEN loss in mice induces rapiddedifferentiation and results in down-regulation of prostate-specific genes, such as MSMB, TGM4, and SBP.43 Thus, itis conceivable that aberrant expression of bIII-tubulin mightreflect the transcriptional reprogramming in PTEN-deletedcancers linked to tumor dedifferentiation. By identifyingERG and PTEN as two key alterations linked to bIII-tubulinexpression, our study provides important insights into thepathways leading to aberrant bIII-tubulin expression inprostate cancer.

Our data showing strong prognostic relevance of bIII-tubulin in prostate cancers raise the possibility of bIII-tubulin representing a biomarker with a potential clinicalutility. This notion is further supported by the fact that ourapproach of analyzing molecular features on a minute TMAspecimen, measuring 0.6 mm in diameter, closely modelsthe molecular analyses of core needle biopsy specimens, inwhich comparable amounts of tissues are evaluated. Multi-variate analyses were, therefore, performed to further eval-uate the relevance of bIII-tubulin measurement.

Several issues need to be considered to comprehend dataobtained in multivariate analyses involving tissues derivedfrom prostatectomy specimens. First, it is obvious thatmultivariate analysis, including all strong prognostic featuresavailable after surgery (eg, pT, pN, surgical margin, or thevalidated Gleason grade) make it difficult for any biomarkerto be established as an independent predictor of prognosis.

The American Journal of Pathology - ajp.amjpathol.org

The inclusion of the pN category also represents a particularissue, because it can substantially reduce the number of casesincluded in a study. This is because lymph node dissection isnot routinely performed in prostate cancer surgery. Post-operative analyses use many parameters, which are unavai-lable at that moment in time, when decisions are made onpatient therapy. Competing parameters for a clinically usefulprognostic biomarker would, thus, rather include parametersthat are available before surgery, such as the (unvalidated)Gleason grade obtained on a core needle biopsy specimen,the preoperative PSA value, and the clinical T category. Inprinciple, potential prognostic biomarkers should, thus, beevaluated on preoperative needle biopsy specimens. From apractical point of view, such analyses are hardly feasible, alsobecause the precious core needle biopsy specimens would beexhausted after only a few studies. In this project, multiplemodels were applied for multivariate analyses to compensateas much as possible for these inherent limitations. That in-dependent association of high-level bIII-tubulin stainingwith early PSA recurrence in prostate cancers was found inall analyzed scenarios, including various combinations ofpreoperatively and postoperatively available parameters,strongly argues for bIII-tubulin representing a biomarkerwith clinical relevance in prostate cancers.

In summary, our study identified a strong link of highbIII-tubulin expression with early PSA recurrence in pros-tate cancers, which was independent of grade, stage, marginstatus, lymph node involvement, and preoperative PSAlevel. The even stronger independent prognostic impact ofbIII-tubulin expression in settings using only parametersthat are preoperatively available may suggest that bIII-tubulin expression analysis (either alone or in combinationwith other molecular parameters) might result in clinicallyuseful information in prostate cancer.

Acknowledgments

We thank Christina Koop, Julia Schumann, Sünje Seekamp,and Inge Brandt for excellent technical support.

References

1. Jemal A, Bray F, Center MM, Ferlay J, Ward E, Forman D: Globalcancer statistics. CA Cancer J Clin 2011, 61:69e90

2. Parkin DM, Bray F, Ferlay J, Pisani P: Global cancer statistics, 2002.CA Cancer J Clin 2005, 55:74e108

3. Orr GA, Verdier-Pinard P, McDaid H, Horwitz SB: Mechanisms ofTaxol resistance related to microtubules. Oncogene 2003, 22:7280e7295

4. Katsetos CD, Herman MM, Mork SJ: Class III beta-tubulin in humandevelopment and cancer. Cell Motil Cytoskeleton 2003, 55:77e96

5. Katsetos CD, Draberova E, Legido A, Dumontet C, Draber P: Tubulintargets in the pathobiology and therapy of glioblastoma multiforme, I:class III beta-tubulin. J Cell Physiol 2009, 221:505e513

6. Koh Y, Jang B, Han SW, Kim TM, Oh DY, Lee SH, Kang CH,Kim DW, Im SA, Chung DH, Kim YT, Kim TY, Kim YW, Kim JH,Heo DS, Bang YJ: Expression of class III beta-tubulin correlates with

615

Tsourlakis et al

unfavorable survival outcome in patients with resected non-small celllung cancer. J Thorac Oncol 2010, 5:320e325

7. Ferrandina G, Zannoni GF, Martinelli E, Paglia A, Gallotta V,Mozzetti S, Scambia G, Ferlini C: Class III beta-tubulin over-expression is a marker of poor clinical outcome in advanced ovariancancer patients. Clin Cancer Res 2006, 12:2774e2779

8. Hetland TE, Hellesylt E, Florenes VA, Trope C, Davidson B, Kaern J:Class III beta-tubulin expression in advanced-stage serous ovariancarcinoma effusions is associated with poor survival and primarychemoresistance. Hum Pathol 2011, 42:1019e1026

9. Choi JW, Kim Y, Lee JH, Kim YS: Expression of b-tubulin isotypesin urothelial carcinoma of the bladder. World J Urol 2012. http://dx.doi.org/10.1007/s00345-012-0993-z [Epub ahead of print]

10. Koh Y, Kim TM, Jeon YK, Kwon TK, Hah JH, Lee SH, Kim DW,Wu HG, Rhee CS, Sung MW, Kim CW, Kim KH, Heo DS: Class IIIbeta-tubulin, but not ERCC1, is a strong predictive and prognosticmarker in locally advanced head and neck squamous cell carcinoma.Ann Oncol 2009, 20:1414e1419

11. Zhang HL, Ruan L, Zheng LM, Whyte D, Tzeng CM, Zhou XW:Association between class III beta-tubulin expression and response topaclitaxel/vinorebine-based chemotherapy for non-small cell lungcancer: a meta-analysis. Lung Cancer 2012, 77:9e15

12. Chen X, Wu J, Lu H, Huang O, Shen K: Measuring beta-tubulin III,Bcl-2, and ERCC1 improves pathological complete remission predic-tive accuracy in breast cancer. Cancer Sci 2012, 103:262e268

13. Zheng WE, Chen H, Yuan SF, Wu LL, Zhang W, Sun HY, Chen WJ:Overexpression of bIII-tubulin and survivin associated with drugresistance to docetaxel-based chemotherapy in advanced gastric can-cer. J BUON 2012, 17:284e290

14. Ploussard G, Terry S, Maille P, Allory Y, Sirab N, Kheuang L,Soyeux P, Nicolaiew N, Coppolani E, Paule B, Salomon L, Culine S,Buttyan R, Vacherot F, de la Taille A: Class III beta-tubulinexpression predicts prostate tumor aggressiveness and patientresponse to docetaxel-based chemotherapy. Cancer Res 2010, 70:9253e9264

15. Egevad L, Valdman A, Wiklund NP, Seve P, Dumontet C: Beta-tubulin III expression in prostate cancer. Scand J Urol Nephrol 2010,44:371e377

16. Grupp K, Diebel F, Sirma H, Simon R, Breitmeyer K, Steurer S, Hube-Magg C, Prien K, Pham T, Weigand P, Michl U, Heinzer H, Kluth M,Minner S, Tsourlakis MC, Izbicki JR, Sauter G, Schlomm T,Wilczak W: SPINK1 expression is tightly linked to 6q15- and 5q21-deleted ERG-fusion negative prostate cancers but unrelated to PSArecurrence. Prostate 2013, 73:1690e1698

17. Schlomm T, Iwers L, Kirstein P, Jessen B, Kollermann J, Minner S,Passow-Drolet A, Mirlacher M, Milde-Langosch K, Graefen M,Haese A, Steuber T, Simon R, Huland H, Sauter G, Erbersdobler A:Clinical significance of p53 alterations in surgically treated prostatecancers. Mod Pathol 2008, 21:1371e1378

18. Kononen J, Bubendorf L, Kallioniemi A, Barlund M, Schraml P,Leighton S, Torhorst J, Mihatsch MJ, Sauter G, Kallioniemi OP:Tissue microarrays for high-throughput molecular profiling of tumorspecimens. Nat Med 1998, 4:844e847

19. Minner S, Enodien M, Sirma H, Luebke AM, Krohn A, Mayer PS,Simon R, Tennstedt P, Muller J, Scholz L, Brase JC, Liu AY,Schluter H, Pantel K, Schumacher U, Bokemeyer C, Steuber T,Graefen M, Sauter G, Schlomm T: ERG status is unrelated to PSArecurrence in radically operated prostate cancer in the absence ofantihormonal therapy. Clin Cancer Res 2011, 17:5878e5888

20. Burkhardt L, Fuchs S, Krohn A, Masser S, Mader M, Kluth M,Bachmann F, Huland H, Steuber T, Graefen M, Schlomm T, Minner S,SauterG,SirmaH,SimonR:CHD1 is a 5q21 tumor suppressor required forERG rearrangement in prostate cancer. Cancer Res 2013, 73:2795e2805

21. Kluth M, Hesse J, Heinl A, Krohn A, Steurer S, Sirma H, Simon R,Mayer PS, Schumacher U, Grupp K, Izbicki JR, Pantel K, Dikomey E,Korbel JO, Plass C, Sauter G, Schlomm T, Minner S: Genomic dele-tion of MAP3K7 at 6q12-22 is associated with early PSA recurrence in

616

prostate cancer and absence of TMPRSS2:ERG fusions. Mod Pathol2013, 26:975e983

22. Krohn A, Diedler T, Burkhardt L, Mayer PS, De Silva C, Meyer-KornblumM,KotschauD, Tennstedt P,Huang J,GerhauserC,MaderM,Kurtz S, Sirma H, Saad F, Steuber T, Graefen M, Plass C, Sauter G,SimonR,Minner S, SchlommT:Genomic deletion of PTEN is associatedwith tumor progression and early PSA recurrence in ERG fusion-positiveand fusion-negative prostate cancer. Am J Pathol 2012, 181:401e412

23. Krohn A, Seidel A, Burkhardt L, Bachmann F, Mader M, Grupp K,Eichenauer T, Becker A, Adam M, Graefen M, Huland H, Kurtz S,Steurer S, Tsourlakis MC, Minner S, Michl U, Schlomm T, Sauter G,Simon R, Sirma H: Recurrent deletion of 3p13 targets multiple tumoursuppressor genes and defines a distinct subgroup of aggressive ERGfusion-positive prostate cancers. J Pathol 2013, 231:130e141

24. Minner S, Wittmer C, Graefen M, Salomon G, Steuber T, Haese A,Huland H, Bokemeyer C, Yekebas E, Dierlamm J, Balabanov S,Kilic E, Wilczak W, Simon R, Sauter G, Schlomm T: High levelPSMA expression is associated with early PSA recurrence in surgicallytreated prostate cancer. Prostate 2011, 71:281e288

25. Tsourlakis MC, Walter E, Quaas A, Graefen M, Huland H, Simon R,Sauter G, Steurer S, Schlomm T, Minner S: High Nr-CAM expressionis associated with favorable phenotype and late PSA recurrence inprostate cancer treated by prostatectomy. Prostate Cancer Prostatic Dis2013, 16:159e164

26. Berger MF, Lawrence MS, Demichelis F, Drier Y, Cibulskis K,Sivachenko AY, et al: The genomic complexity of primary humanprostate cancer. Nature 2011, 470:214e220

27. Lapointe J, Li C, Giacomini CP, Salari K, Huang S, Wang P,Ferrari M, Hernandez-Boussard T, Brooks JD, Pollack JR: Genomicprofiling reveals alternative genetic pathways of prostate tumorigen-esis. Cancer Res 2007, 67:8504e8510

28. Taylor BS, Schultz N, Hieronymus H, Gopalan A, Xiao Y, Carver BS,Arora VK, Kaushik P, Cerami E, Reva B, Antipin Y, Mitsiades N,Landers T, Dolgalev I, Major JE, Wilson M, Socci ND, Lash AE,Heguy A, Eastham JA, Scher HI, Reuter VE, Scardino PT, Sander C,Sawyers CL, Gerald WL: Integrative genomic profiling of humanprostate cancer. Cancer Cell 2010, 18:11e22

29. Epstein JI, Feng Z, Trock BJ, Pierorazio PM: Upgrading anddowngrading of prostate cancer from biopsy to radical prostatectomy:incidence and predictive factors using the modified Gleason gradingsystem and factoring in tertiary grades. Eur Urol 2012, 61:1019e1024

30. Tennstedt P, Koster P, Bruchmann A, Mirlacher M, Haese A, Steuber T,Sauter G, Huland H, Graefen M, Schlomm T, Minner S, Simon R: Theimpact of the number of cores on tissue microarray studies investigatingprostate cancer biomarkers. Int J Oncol 2012, 40:261e268

31. Minner S, Jessen B, Stiedenroth L, Burandt E, Kollermann J,MirlacherM, ErbersdoblerA, EichelbergC, FischM,Brummendorf TH,Bokemeyer C, Simon R, Steuber T, Graefen M, Huland H, Sauter G,Schlomm T: Low level HER2 overexpression is associated with rapidtumor cell proliferation and poor prognosis in prostate cancer. ClinCancer Res 2010, 16:1553e1560

32. Müller J, Ehlers A, Burkhardt L, Sirma H, Steuber T, Graefen M,Sauter G, Minner S, Simon R, Schlomm T, Michl U: Loss ofpSer2448-mTOR expression is linked to adverse prognosis and tumorprogression in ERG-fusion-positive cancers. Int J Cancer 2013, 132:1333e1340

33. Minner S, Kraetzig F, Tachezy M, Kilic E, Graefen M, Wilczak W,Bokemeyer C, Huland H, Sauter G, Schlomm T: Low activatedleukocyte cell adhesion molecule expression is associated withadvanced tumor stage and early prostate-specific antigen relapse inprostate cancer. Hum Pathol 2011, 42:1946e1952

34. Ranganathan S, Benetatos CA, Colarusso PJ, Dexter DW, Hudes GR:Altered beta-tubulin isotype expression in paclitaxel-resistant humanprostate carcinoma cells. Br J Cancer 1998, 77:562e566

35. Ranganathan S, Dexter DW, Benetatos CA, Hudes GR: Cloning andsequencing of human betaIII-tubulin cDNA: induction of betaIII

ajp.amjpathol.org - The American Journal of Pathology

bIII-Tubulin in Prostate Cancer

isotype in human prostate carcinoma cells by acute exposure to anti-microtubule agents. Biochim Biophys Acta 1998, 1395:237e245

36. Ganguly A, Yang H, Cabral F: Class III beta-tubulin counteracts theability of paclitaxel to inhibit cell migration. Oncotarget 2011, 2:368e377

37. Tomlins SA, Rhodes DR, Perner S, Dhanasekaran SM, Mehra R,Sun XW, Varambally S, Cao X, Tchinda J, Kuefer R, Lee C,Montie JE, Shah RB, Pienta KJ, Rubin MA, Chinnaiyan AM:Recurrent fusion of TMPRSS2 and ETS transcription factor genes inprostate cancer. Science 2005, 310:644e648

38. Weischenfeldt J, Simon R, Feuerbach L, Schlangen K, Weichenhan D,Minner S, et al: Integrative genomic analyses reveal an androgen-driven somatic alteration landscape in early-onset prostate cancer.Cancer Cell 2013, 23:159e170

39. Birdsey GM, Dryden NH, Shah AV, Hannah R, Hall MD,Haskard DO, Parsons M, Mason JC, Zvelebil M, Gottgens B,Ridley AJ, Randi AM: The transcription factor Erg regulates expres-sion of histone deacetylase 6 and multiple pathways involved in

The American Journal of Pathology - ajp.amjpathol.org

endothelial cell migration and angiogenesis. Blood 2012, 119:894e903

40. Chow A, Amemiya Y, Sugar L, Nam R, Seth A: Whole-transcriptomeanalysis reveals established and novel associations with TMPRSS2:ERG fusion in prostate cancer. Anticancer Res 2012, 32:3629e3641

41. Grupp K, Kohl S, Sirma H, Simon R, Steurer S, Becker A, Adam M,Izbicki J, Sauter G, Minner S, Schlomm T, Tsourlakis MC: Cysteine-rich secretory protein 3 overexpression is linked to a subset of PTEN-deleted ERG fusion-positive prostate cancers with early biochemicalrecurrence. Mod Pathol 2013, 26:733e742

42. Vitolo MI, Boggs AE, Whipple RA, Yoon JR, Thompson K,Matrone MA, Cho EH, Balzer EM, Martin SS: Loss of PTEN inducesmicrotentacles through PI3K-independent activation of cofilin. Onco-gene 2013, 32:2200e2210

43. Thielen JL, Volzing KG, Collier LS, Green LE, Largaespada DA,Marker PC: Markers of prostate region-specific epithelial identitydefine anatomical locations in the mouse prostate that are molecularlysimilar to human prostate cancers. Differentiation 2007, 75:49e61

617