accurate predition of repeat prostated biospys outcomes_robinson_prostate cancer prostatic dis_2010

8/2/2019 Accurate Predition of Repeat Prostated Biospys Outcomes_Robinson_Prostate Cancer Prostatic Dis_2010

1/6

ORIGINAL ARTICLE

Accurate prediction of repeat prostate biopsy outcomes bya mitochondrial DNA deletion assay

K Robinson1, J Creed1, B Reguly1, C Powell1, R Wittock1, D Klein1, A Maggrah1, L Klotz2, RL Parr1 andGD Dakubo1

1Genesis Genomics, Thunder Bay, ON, Canada and 2Division of Urology, Department of Surgery, Sunnybrook Health SciencesCentre, University of Toronto, Toronto, ON, Canada

Several cancers are characterized by large-scale mtDNA deletions. We previously provided evidencethat one of these deletions has potential utility in resolving false from true-negative prostate needlebiopsies. This study was to assess the clinical value of this deletion in predicting re-biopsy outcomes.We used a quantitative polymerase chain reaction assay to measure the levels of the deletion inindividual negative needle biopsies from 101 patients who had a repeat biopsy within a year with

known outcomes. Using an empirically established cycle threshold (Ct) cutoff of 31, and the lowestCt for each patient as diagnostic of prostate cancer, as well as the histopathologic diagnosis onsecond biopsy, we calculated the clinical performance of the deletion. The Ct cutoff at 31 gave asensitivity and specificity of 84 and 54%, respectively, with the area under a receiver-operatingcharacteristics curve of 0.749. The negative predictive value was 91%. The assay was able to predictthe presence of a missed tumor in 17 out of 20 men a year before diagnosis. This ancillary testappears to identify men who do not require a repeat biopsy with a high degree of certainty. Theresults suggest that the majority of men with atypical small acinar proliferation have a concurrentmissed tumor and therefore require close monitoring for early detection.Prostate Cancer and Prostatic Diseases (2010) 13, 126131; doi:10.1038/pcan.2009.64; published online 19 January 2010

Keywords: Negative prostate needle biopsies; mtDNA deletion; negative predictive value; fieldcancerization

Introduction

In the developed world, prostate cancer (PCa) remainsthe most frequent noncutaneous malignancy diagnosedin men over the age of 50 years; however, the case fatalityof PCa is lower than many cancers. The decreasedmortality is attributable to early detection of curabletumors and the slow natural history of many PCas. Earlydetection is achieved by increased awareness coupledwith diligent digital rectal examination and PSA screen-ing regimes. Although the PSA assay continues to beimportant in PCa management, it is quite inaccurate as ascreening test. Several nonmalignant conditions includ-ing BPH and prostate infections can cause elevated PSA.1

The positive predictive value of an elevated PSA is35%. Between 15 and 30% of men with negative biopsiesharbor PCa. In these cases, the negative results areaccounted for by sampling error.25 Men whose biopsiesshow high-grade prostatic intraepithelial neoplasia(HGPIN) or atypical small acinar proliferation (ASAP)

have a 2441% risk of underlying PCa that is identifiedon repeat biopsy.6 Cancer has been reported in up to100% of ASAP cases.7 These findings highlight theneed to re-biopsy patients with ASAP/HGPIN within612 months.

Men with an elevated PSA are at increased risk forPCa. A single negative biopsy is not sufficient to excludethe diagnosis with certainty. Progressive rise in the PSAafter an initial negative biopsy causes further anxiety andis often considered an indication for requiring a second

biopsy.8 An ancillary diagnostic test to rule out cancerwithout a repeat biopsy would have many advantages.

We previously reported the potential clinical valueof a 3.4 kb mitochondrial DNA deletion that shows theconcept of field effect or cancerization.9 This couldresolve negative prostate biopsies into true- and false-negative categories. However, of the 96 patients in thevalidation study that provided proximal biopsy samples,only 48 had accompanying malignant biopsy cores forcomparative analysis of test performance. Though manyof these patients (30) had only a single biopsy, the 3.4 kbdeletion was accurate in predicting the presence of amalignant focus in 67% of the patients. However, theperformance of the deletion improved to 78% when weexamined 18 patients who had two or more previoussamples. This suggests that the marker could predict the

Received 10 September 2009; revised 4 December 2009; accepted 18December 2009; published online 19 January 2010

Correspondence: Dr GD Dakubo, Genesis Genomics, 1000-290 MunroStreet, Thunder Bay, ON P7A 7T1, Canada.E-mail: [email protected]

Prostate Cancer and Prostatic Diseases (2010) 13, 126131& 2010 Nature Publishing Group All rights reserved 1365-7852/10

www.nature.com/pcan
http://dx.doi.org/10.1038/pcan.2009.64mailto:[email protected]://www.nature.com/pcanhttp://www.nature.com/pcanmailto:[email protected]://dx.doi.org/10.1038/pcan.2009.64


2/6

outcome for the patient with a rising PSA and nega-tive initial biopsies with a high degree of accuracy.9 Thisstudy was designed to establish the 3.4-kb deletion

biomarker cutoff that had the highest predictive accuracyfor the results of subsequent biopsies. The findingsshould form the basis for the subsequent evaluation ofthis marker on prostate re-biopsy outcomes.

Materials and methods

Patients and samplesAll patients were consented and samples obtained inaccordance with the ethical guidelines of the TrafalgarEthics Board (Oakville, Canada) that operates in accor-dance with the Tri-Council Policy Statement on EthicalConduct of Research Involving Humans.

A total of 101 patients who have had a follow-up biopsy within 1 year of an initial negative biopsy forsuspected PCa were recruited for the study. Sections(20 mm) from the six representative anatomic regions of

the prostate (right apex, right mid, right base, left apex,left mid, left base) of the initial negative biopsies wereobtained. It should be noted that by conventionalmodern practice, each patient received between 10 and12 core biopsies; however, for logistic reasons we testedonly cores representing six anatomic areas of theprostate. In all, a total of 595 (instead of 606) biopsieswere analyzed. Eleven patients provided 5 cores each(instead of the 6 requested cores). The biopsies spana period between 1997 and 2008. On second biopsy,20 patients were found to have cancer, with the remain-ing found to be benign. Ten of the PCa patients subse-quently had prostatectomies for treatment of PCa.

DNA extraction and target amplificationDNA was extracted from 20 mm prostate needle biopsysections using a QIAamp DNA Mini Kit (Qiagen,Mississauga, Ontario, Canada). Negative extraction con-trols were included and were monitored for amplifica-tion. Samples were quantified, diluted to 2 ngml1 anddistributed to master template 96-well plates. Polymerasechain reaction (PCR) amplification was performed withPerfeCta Sybr Green Supermix (Quanta Biosciences,Gaithersburg, MD, USA). The following reaction condi-tions were used: 1 PerfeCta Sybr Green Supermix,0.25 mM forward primer, 0.25 mM reverse primer (primersequences available upon request), 20 ng template DNAin a 25 ml reaction volume. PCR was performed using thefollowing cycling parameters: 95 1C for 2 min to activatethe polymerase, followed by 45 cycles of 95 1C for 30s,30 s annealing at optimized primer temperatures: 61.5 1Cfor total mtDNA and tumor necrosis factor, 64.4 1Cfor deletion-specific primer and extension at 72 1C for30 s. Plate reads were performed after each extensioncycle. A 10-min final extension at 72 1C was performed

before the melting curve analysis (70951C, readingevery 1 1C and holding for 3 s). The Opticon 2 andChromo4 Real-Time Detector Systems (Bio-Rad Labora-tories, Mississauga, Ontario, Canada) were used forPCR. One negative PCR control was included on everyamplification plate for each primer set as well as sevenstandards for normalization purposes. These standards

were serial dilutions of the target amplicon generatedwith conventional PCR, purified, quantified and dilutedsuch that the seven standards were at concentrationsof 3.85 ng, 0.385 ng, 38.5pg, 3.85pg, 0.385pg, 38.5fg and3.85fg. Analysis of any given plate required a minimumof four standards.

Statistical analysisThe cycle threshold (Ct) of the 3.4 kb deletion wascompared to the Ct of total amount of mtDNA (12SrRNA gene) and the amount of nuclear DNA (tumornecrosis factor). A tumor necrosis factor Ct of440 for asample was considered a failed reaction because thisreflected insufficient template input. This restrictionresulted in seven failed patient samples, which wereremoved from further analysis. Template DNA (10 ng)was seeded into each well for the quantitative PCR assay.Because tumor necrosis factor measured template input,the Ct variance across all samples was within 1 cycle;hence further analysis relied solely on 3.4 kb deletion Ct.

The mean and standard errors were calculated for each

tissue classification. Two-by-two tables were constructedfrom the Cts for cancer and benign outcomes, and usedto compute assay performance. Similarly, with referenceto second biopsy outcomes, receiver-operating character-istics (ROC) curves were constructed for deletion Cts andthe area under the curves was computed. All the testswere two-sided and statistical significance was consid-ered at a P-valueo0.05. SPSS version 13 (SPSS, Chicago,IL, USA) was used for all statistical analyses.

Results

The clinical and demographic data for all 101 subjects aresummarized in Table 1, and the pathological charac-teristics of the malignant patients are found in Table 2.The mean (s.d.) age, PSA and duration between

biopsies for the participants were 60.64 (7.99) years,7.09 (4.93) ngml1 and 7.72 (5.27) months, respec-tively. The group with ASAP was significantly youngerthan other patients (P 0.05). On the initial biopsy(tested in this assay), 8 patients had no histopathologicfindings in any of their biopsies, 15 had isolated ASAP,28 had isolated PIN, 7 had both ASAP and PIN, and 43had findings such as hyperplasia, atrophy and prostaticinflammation. The majority of patients with subsequentdiagnosis of cancer appear to have limited cancer (o50%core involvement, Gleason score o7 and o3 coresinvolved, see Table 2). Seven patients were excludedfrom further analysis because their samples failed tomeet assay quality standards. Hence, the final dataoutput based on 94 patients is composed of 20 malignantand 74 benign biopsies.

Because the Cts of our previous assays were generallyhigh,9 we optimized the assay to improve efficiency andspecificity. We redesigned the primers for the deletionand this dropped the annealing temperature from 66 to64.4 1C. Importantly, in this study Sybr Green reagentswere used that have been specifically optimized forefficiency and sensitivity. All of these changes culmi-nated in efficient deletion target amplification. Theefficiencies for all runs in this assay were tight and

mtDNA assay for making prostate re-biopsy decisionK Robinson et al

127

Prostate Cancer and Prostatic Diseases
http://www.nature.com/pcan


3/6

480% compared to the highly variable efficiencies (6080%) from the previous study.

In our previous study benign, malignant and histolo-gically normal-appearing biopsies adjacent to a malig-nant focus were tested. The greatest separation of assayscores was achieved between the malignant and benign

biopsies. However, this study assayed only biopsiesshowing either benign histology, ASAP or HGPIN.Hence to establish a usable assay cutoff for accuratedelineation of benign versus malignant prostates, weused only biopsies without suspicious histological find-ings and those with a subsequent diagnosis of cancer. Weidentified 8 patients (48 core samples) that met thecriteria for benign disease and 14 malignant patients(84 samples) were selected. The mean Cts for benign andmalignant cases were 32.3 and 29.4, respectively, indicat-ing a 7.9-fold difference in the levels of the deletion

between the two groups. By testing the sensitivityand specificity using an ROC curve, the optimal Ctfor accurate prediction of a positive re-biopsy outcomewas 31. Thus, the positive likelihood ratio of a missedconcurrent tumor increases or decreases with decreasingor increasing Ct, respectively (Figure 1). At this Ct cutoff,there was a statistically significant mean Ct difference

between the two groups (P 0.001). A sensitivity of79% and a specificity of 88% were achieved. Assayperformance by ROC showed an AUC of 0.893 (95% CI;0.7541.031) (Figure 2).

The 31-Ct cutoff was used to test the outcomes ofall samples in a blinded fashion. Our initial data ana-lysis revealed that a disproportionately high number

of prostates with initial or subsequent findings ofASAP were called malignant by the assay. These wereremoved and the remaining benign samples were testedseparately. The difference between the mean Cts ofthe benign and malignant samples was significant atP 0.001. In reference to the histopathologic diagnosison second biopsy, the assay had a sensitivity of 84%and a specificity of 54%. In men with initial negative

biopsies, the false-negative rate was low and the negativepredictive value was 91%. The AUC was 0.749 (95% CI;0.6250.873) (Figure 3).

Of the 101 patients in this study, 22 had ASAP oninitial biopsy of which 10 were diagnosed with cancerwithin a year. This rate (45.5%) of cancer on re-biopsy ofpatients with ASAP is consistent with other reports.10

The assay also predicted the presence of concurrentcancer in 8 of the remaining 12 subjects who had nocancer on re-biopsy. Six patients with initial negative

biopsies were found to have ASAP on re-biopsy, andthe assay predicted the presence of tumor on the initial

Table 2 Clinical data on patients with prostate cancer

PSA (ngml1):N

Gleasonscore:

N

Biopsyinvolved(%): N

No. ofcores

involved: N

No. ofpositive cores

by assay

04: 1 6: 13 110: 10 12: 16 12: 64.110: 17 7: 1 10.150: 7 34: 3 34: 4410: 2 8: 1 450: 1 NR: 1 56: 7

NR: 1 NR: 2 NA: 3 (Benign)

Abbreviations: NA, not applicable; NR, not recorded. 31 CYCLE THRESHOLDCUTOFF

Increasing likelihood of

missed prostate cancer

Increasing likelihood of

benign prostate gland

Positive likelihood ratio PCR cycle threshold

35.0

33.5

27.0

29.5

1.0

1.2

1.9

3.3

5.4

Figure 1 A depiction of the scale used in making predictions ofthe likelihood of no tumor or missed concurrent tumor during thebiopsy procedure. PCR, polymerase chain reaction.

Table 1 Clinical and demographic characteristics of patients

PCaa ASAPa HGPINa Benigna P-value

No. of patients 20 10 14 57Age (years) means.d. 63.057.46 56.228.28 61.678.35 60.977.31 0.50PSA (ng ml1) means.d. 6.352.53 7.955.41 6.836.83 7.325.04 0.652Time between biopsies (months)s.d. 7.756.34 5.834.19 7.214.99 9.165.12 0.179

Gleason gradep6 177 2 NA NA NA NAX8 1

% Core involvedp10 9410, p50 8 NA NA NA NA450 1NA 2

Abbreviations: ASAP, atypical small acinar proliferation; HGPIN, high grade prostatic intraepithelial neoplasia; NA, not applicable; PCa, prostate cancer.aFindings on second biopsy procedure.


128



4/6

biopsy in all six patients. Thus, 78% of the cases withASAP on either the initial or subsequent biopsy wereflagged as having cancer by the assay (Figure 4).

The extent of the field alterations in the prostatedefined by the 3.4-kb mitochondrial genome deletion isunknown. However, given that the deletion was able topredict a positive biopsy in 17 out of 20 (85%) individualswith cancer a year prior, we envisioned this as an earlyevent that might involve multiple or large areas of theprostate. To gain an insight into the possible extent ofthis field effect, we looked at the number of cores calledpositive by the assay in each malignant patient. Of the17 positive cases, 11 had more than 3 positive biopsies,which is suggestive of an extensive field effect (Table 2).

The archived samples used for the study werecollected between 1997 and 2008 (spanning 12 years).Because both the time of fixation before embedding andage of the block adversely affect nucleic acid quality, we

Area Under the Curve

Test Result Variable(s): Deletion

0.893 0.071 0.003 0.754 1.031Area Std. Error

Asymptotic

Sig. Lower Bound Upper Bound

A symptotic 95% ConfidenceI nterval

p= 0.001

Sensitivity

1.00

0.80

0.60

0.40

0.20

0.000.00 0.20 0.40 0.60 0.80 1.00

33a b

32

31

30

Diagnosis1-Specificity

Benign Malgnant

M

ean+-1SE

Deletion

Figure 2 Assay performance in discriminating between patients whose initial and subsequent biopsies were benign without anyhistopathologic changes and those with malignant diagnosis on subsequent biopsies. (a) Mean cycle threshold (Ct)s.e. for the benign andmalignant groups (P0.001). (b) Receiver-operating characteristic curve for the deletions Cts calculated with reference to subsequenthistopathologic outcomes.

Area Under the Curve

a b

Test Result Variable(s): Deletion

0.749 0.063 0.001 0.625 0.873

Area Std. ErrorAsymptotic

Sig. Lower Bound Upper Bound

A symptotic 95% ConfidenceInterval

Mean+-1SEDeletion

31.5

31.0

30.5

30.0

29.5

29.0

Diagnosis

Benign Malignant

1- Specificity

Sensitivity

1.00

0.80

0.60

0.40

0.20

0.000.60 0.80 1.000.400.200.00

Figure 3 Assay performance in discriminating between patients who remained benign from those diagnosed with cancer on re-biopsy.(a) Mean cycle threshold (Ct)s.e. for the benign and malignant groups (P 0.001). (b) Receiver-operating characteristic curve for thedeletions Cts calculated with reference to subsequent histopathologic outcomes.

31.5

31.0

30.5

30.0

29.5

29.0

28.5

Group

Mean+-1SEDeletion

Benign Malignant ASAP

Figure 4 Atypical small acinar proliferation (ASAP) in the prostateis classified by the assay as malignant in a majority of cases.


129



5/6

looked at how these affected assay performance. Therewas no significant correlation of Cts and PCR efficiencieswith block ages, indicating the lack of PCR interference

by block ages. This finding is likely due to the efficiencyof the DNA extraction from formalin-fixed samplesusing QIAmp kits, as well as the small-sized amplicon(273 bp) targeted by the quantitative PCR assay.

Discussion

The reliance on PSA screening for making biopsydecisions suggests that many men (especially thosewith mildly elevated PSA results) who undergo trans-rectal biopsy for PCa diagnosis will appear to be negativefor cancer on initial biopsy. A repeat biopsy of thispopulation within 5 years will reveal the presence ofcancer in up to 30% of the patients, indicating thatthe prostate biopsy procedure is associated with a highsampling error. These high-risk men whose initial

biopsies appear negative would benefit from accurate

ancillary tests that establish the risk of PCa in individualpatients.

This clinical study was conducted with two primaryobjectives: first, to establish a clinically usable cutoffCt, and second, to assess the performance of the assayin accurately predicting re-biopsy outcomes using theestablished Ct cutoff. In the optimized assay, a Ct cutoffof 31 gave the best performance, with a sensitivity of84%, an NPV of 91% and an area under the ROC curveof 0.749.

These results are due to the presence of a field effect.A number of investigators have confirmed the conceptof precancerous molecular events in normal-appearingcells using different markers.1113 The obvious benefit of

such field biomarkers is in the selection of men at highrisk of cancer for close monitoring, while reassuring thelow-risk population.

PCa is ultimately found in people diagnosed withASAP at a rate of 42100%. Indeed, several authors haveemphasized that the diagnosis of ASAP may providefalse reassurance, because it is in fact infiltrating carci-nomas in many cases.14 It is not surprising that amolecular test will be positive in most ASAP cases.These men are high risk for PCa. The few remaining areunlikely to have PCa, and the mandate to re-biopsycould be relaxed.

The field concept posits that tumors develop inmolecularly preconditioned fields. Histologically nor-mal-appearing cells adjacent to most cancers containgenetic imprints indicative of malignancy. These mole-cular alterations associated with cancer fields affordan enormous opportunity for cancer management.Several groups have confirmed field cancer effectsin prostatic tissues.12,15,16 Chemoprevention or earlydetection strategies would benefit from the profilingof cancer field biomarkers in high-risk populations. Ahistologically normal surgical margin may warrantmolecular examination to preclude submicroscopicinvolvement.17 An underutilized advantage of the fieldconcept of carcinogenesis lies in improving needle

biopsy sampling errors by profiling the apparentlynormal biopsies. Negative prostate biopsies in high-riskpatients may benefit from field profiling. We show that a

single mitochondrial deletion assay for field effects inprostatic tissue has significant potential as a biomarkerfor PCa.

Conflict of interest

The authors Kerry Robinson, Jennifer Creed, BrianReguly, Cortney Powell, Roy Wittock, Daniel Klein,Andrea Maggrah, Ryan L. Parr and Gabriel D. Dakubowork for and own stocks in Genesis Genomics. LaurenceKlotz is a paid consultant with Genesis Genomics.

Acknowledgements

This work was supported in part by grants from IndustryCanada (FedNor), Sudbury, Canada, and NorthernOntario Heritage Fund Corporations (NHOFC), SaultSte Marie. We thank the patients and their physician forthe samples.

References

1 Dhanasekaran SM, Barrette TR, Ghosh D, Shah R, Varambally S,Kurachi K et al. Delineation of prognostic biomarkers in prostatecancer. Nature 2001; 412: 822826.

2 Satoh A, Matsumoto K, Nakamura S. Is interval from an initial biopsy a significant predictor of prostate cancer at repeatbiopsies? Int J Urol 2006; 13: 224227.

3 Eggener SE, Roehl KA, Catalona WJ. Predictors of subsequentprostate cancer in men with a prostate specific antigen of2.64.0 ng/ml and an initially negative biopsy. J Urol 2005; 174:500504.

4 Singh H, Canto EI, Shariat SF, Kadmon D, Miles BJ, Wheeler TM

et al. Predictors of prostate cancer after initial negative syste-matic 12 core biopsy. J Urol 2004; 171: 18501854.5 Hong YM, Lai FC, Chon CH, McNeal JE, Presti Jr JC. Impact of

prior biopsy scheme on pathologic features of cancers detectedon repeat biopsies. Urol Oncol 2004; 22: 710.

6 Epstein JI, Herawi M. Prostate needle biopsies containingprostatic intraepithelial neoplasia or atypical foci suspiciousfor carcinoma: implications for patient care. J Urol 2006; 175(3 Pt 1): 820834.

7 Brausi M, Castagnetti G, Dotti A, De Luca G, Olmi R, CesinaroAM. Immediate radical prostatectomy in patients with atypicalsmall acinar proliferation. Over treatment? J Urol 2004; 172:906908 discussion 908-909.

8 Fowler Jr FJ, Barry MJ, Walker-Corkery B, Caubet JF, Bates DW,Lee JM et al. The impact of a suspicious prostate biopsy onpatients psychological, socio-behavioral, and medical care out-

comes. J Gen Intern Med 2006; 21: 715721.9 Maki J, Robinson K, Reguly B, Alexander J, Wittock R, Aguirre A

et al. Mitochondrial genome deletion aids in the identification offalse- and true-negative prostate needle core biopsy specimens.Am J Clin Pathol 2008; 129: 5766.

10 Bostwick DG, Meiers I. Atypical small acinar proliferation in theprostate: clinical significance in 2006. Arch Pathol Lab Med 2006;130: 952957.

11 Dhir R, Vietmeier B, Arlotti J, Acquafondata M, Landsittel D,Masterson R et al. Early identification of individuals withprostate cancer in negative biopsies. J Urol 2004; 171: 14191423.

12 Mehrotra J, Varde S, Wang H, Chiu H, Vargo J, Gray K et al.Quantitative, spatial resolution of the epigenetic field effect inprostate cancer. Prostate 2008; 68: 152160.

13 Rizzi F, Belloni L, Crafa P, Lazzaretti M, Remondini D,Ferretti S et al. A novel gene signature for molecular


130



6/6

diagnosis of human prostate cancer by RT-qPCR. PLoS ONE2008; 3: e3617.

14 Jonathan I, Epstein XJY. Findings of atypical glands suspiciousfor cancer. In: Prostate Biopsy Interpretation (3rd edn) LippincottWilliams & Wilkins: Philadelphia, pp 177184.

15 Dakubo GD, Jakupciak JP, Birch-Machin MA, Parr RL. Clinicalimplications and utility of field cancerization. Cancer Cell Int 2007; 7: 2.

16 Haaland CM, Heaphy CM, Butler KS, Fischer EG, Griffith JK,Bisoffi M. Differential gene expression in tumor adjacenthistologically normal prostatic tissue indicates field canceri-zation. Int J Oncol 2009; 35: 537546.

17 Theodorescu D, Frierson Jr HF, Sikes RA. Molecular determi-nation of surgical margins using fossa biopsies at radicalprostatectomy. J Urol 1999; 161: 14421448.


131


accurate predition of repeat prostated biospys outcomes_robinson_prostate cancer prostatic dis_2010

Documents