Variability in Diagnostic Opinion Among Pathologistsfor Single Small Atypical Foci in Prostate Biopsies

Theodorus H. Van der Kwast, MD, PhD,* Andrew Evans, MD, PhD,* Gina Lockwood, MSc,w

Doug Tkachuk, MD,*z David G. Bostwick, MD, MBA,y Jonathan I. Epstein, MD,J

Peter A. Humphrey, MD,z Rodolfo Montironi, MD,# Geert J. L. H. Van Leenders, MD, PhD,

C-G.Pihl, MD, ** Carl-Gustaf Pihl, MD, w w Ingrid Neetens, MD,zz Paula M. Kujala, MD,yy

Marita Laurila, MD,yy Catharine Mazerolles, MD,JJ Lukas Bubendorf, MD, PhD,zz

Antonio Finelli, MD,## Kemp Watson,z and John Srigley, MD***

Abstract: Pathologists are increasingly exposed to prostate

biopsies with small atypical foci, requiring differentiation

between adenocarcinoma, atypical small acinar proliferation

suspicious for malignancy, and a benign diagnosis. We studied

the level of agreement for such atypical foci among experts in

urologic pathology and all-round reference pathologists of the

European Randomized Screening study of Prostate Cancer

(ERSPC). For this purpose, we retrieved 20 prostate biopsies

with small (most <1mm) atypical foci. Hematoxylin and eosin-

stained slides, including 10 immunostained slides were digita-

lized for virtual microscopy. The lesional area was not marked.

Five experts and 7 ERSPC pathologists examined the cases.

Multirater k statistics was applied to determine agreement and

significant differences between experts and ERSPC pathologists.

The k value of experts (0.39; confidence interval, 0.29-0.49) was

significantly higher than that of ERSPC pathologists (0.21;

confidence interval, 0.14-0.27). Full (100%) agreement was

reached by the 5 experts for 7 of 20 biopsies. Experts and

ERSPC pathologists rendered diagnoses ranging from benign

to adenocarcinoma on the same biopsy in 5 and 9 biopsies,

respectively. Most of these lesions comprised between 2 and 5

atypical glands. The experts diagnosed adenocarcinoma (49%)

more often than the ERSPC pathologists (32%) (P<0.001). As

agreement was particularly poor for foci comprising <6 glands,

we would encourage pathologists to obtain intercollegial con-

sultation of a specialized pathologist for these lesions before a

carcinoma diagnosis, whereas clinicians may consider to perform

staging biopsies before engaging on deferred or definite therapy.

Key Words: prostate cancer, prostate biopsy, pathology,

diagnostics, interobserver variation, virtual microscopy

(Am J Surg Pathol 2010;34:169–177)

Small atypical foci occupying less than a few milli-meters in prostate biopsies can include benign

glandular structures, but also small foci of adenocarci-noma. Such small foci are frequently encountered incurrent pathology practice. They comprised about 4% ofbiopsies during population-based prevalence screeningwith prostate-specific antigen (PSA) testing,16 represent-ing between 16% and 34% of the detected cancers duringsubsequent screening rounds.16,19,20 Among these smallatypical foci pathologists try to distinguish adenocarci-noma and benign mimickers of prostatic adenocarci-noma,11 and those, which fall short of the criteria foradenocarcinoma are termed atypical small acinar pro-liferations (ASAP) suspicious for but not diagnostic ofadenocarcinoma.12 The frequency of these ASAP diag-noses in contemporary large prostate biopsy series variesfrom <1% to 23%, averaging around 5%.4,19 This largevariation may be attributed to differences in patientpopulations, in quality of biopsies and their histopatho-logic processing or in interpretation and diagnosticconfidence of pathologists.

Criteria to distinguish these suspicious lesions fromadenocarcinoma are not absolute and they include sizeof the area in question, the number of atypical acini, andthe presence of nuclear hyperchromasia and prominentnucleoli.13 It is now commonly accepted that its distinc-tion ultimately reflects the expertise and confidence ofthe pathologist.

The clinical impact of a diagnosis on these smallatypical foci may be considerable. If a diagnosis ofadenocarcinoma is rendered, the patient and/or clinicianCopyright r 2010 by Lippincott Williams & Wilkins

From the Departments of *Pathology; wBiostatistics; ##Urology,University Health Network; zObjective Pathology Services, Toronto;***Department of Pathology, Credit Valley Hospital, Mississauga,Canada; yBostwick Laboratories, Glen Allen, VA; JDepartment ofPathology, The Johns Hopkins Medical Institutions, Baltimore, MD;zDepartment of Pathology and Immunology, Washington Univer-sity School of Medicine, St Louis, MO; #Section of PathologicalAnatomy, Polytechnic University of the Marche Region, Ancona,Italy; **Department of Pathology, Erasmus MC Rotterdam, TheNetherlands; wwDepartment of Pathology, Sahlgrenska UniversityHospital, Goteborg, Sweden; zzDepartment of Pathology, Mid-delheim Hospital, Antwerp, Belgium; yyDepartment of Pathology,Centre for Laboratory Medicine, Tampere University Hospital,Tampere, Finland; JJService d’Anatomie et Cytologie Pathologiques,CHU de Toulouse-Rangueil, Toulouse, France; and zzInstitute forPathology, University Hospital Basel, Basel, Switzerland.

Correspondence: Theodorus H. Van der Kwast, MD, PhD, Departmentof Pathology, 11th floor, University Health Network, 200 ElizabethStreet, TorontoM5G 2C4, Canada (e-mail: theo.vdkwast@uhn.on.ca).

ORIGINAL ARTICLE

Am J Surg Pathol � Volume 34, Number 2, February 2010 www.ajsp.com | 169

may opt for a definitive treatment such as radicalprostatectomy or radiotherapy, whereas repeat biopsieswill be performed if a suspicious diagnosis is given. As ofyet, no formal study has been performed to study the levelof agreement among experts and more all-round pathol-ogists in their diagnostic interpretation of small atypicalfoci in prostate biopsies. We also felt that the finding ofconsiderable interobserver variation among pathologistsmay have medicolegal implications and may affect theway pathologists and clinicians deal with these lesions.

MATERIALS AND METHODS

Slide Selection and Slide ImagingA total of 16 sets of prostate needle biopsies were

selected from the archives of the Department ofPathology, University Health Network (UHN), Torontousing natural language search to identify single atypicalfoci suspicious for carcinoma and small foci of adeno-carcinoma. An additional 4 cases with a presumedatypical focus, but considered to be benign, were selectedfrom biopsies during routine pathology reporting by oneof the pathologists (T.H.v.d.K.). All biopsies had been cutat least at 3 levels and after examination of all levels ofeach case the level, which provided the best representationof the lesion according to this pathologist was selected forthe study. Further, the case composition was determined apriori to have some balance in the distribution of diagnoses.Thus, 4 biopsies with a benign diagnosis, 8 biopsiesreported as adenocarcinoma, and 8 biopsies diagnosed byUHN pathologists as suspicious for adenocarcinoma, butlacking criteria for a definite diagnosis (ASAP) wereincluded.

The atypical focus was not marked on the slide, toimitate the real life pathology setting as much as possible.Table 1 gives an overview of the originally reporteddiagnoses of each of the 20 cases included in the study.

High resolution virtual slides of the selected bio-psies, including the corresponding immunohistochemicalstains, if available, were obtained using an AperioScanScope Scanner (Aperio, Vista, CA) to allow theirexamination between 0.5� and 40� objective magnifi-cation. The digitalized slides were uploaded to the studywebsite, http://www.objective.pathology.com/, as JPEGfiles, compressed at medium quality for online viewingthrough a digital microscope interface that allowed navi-gation from the desktop computers of the pathologistswho participated in the study.

Participating PathologistsFive experts in urologic pathology, 4 from North

America, 1 from Europe were invited to participate in thisvirtual microscopy study as well as the 8 members of thepathology committee of the European Randomized studyof Screening for Prostate Cancer (ERSPC). The latterare nonspecialist pathologists with a special interest inurologic pathology and with considerable experience inreading prostate biopsies. All participants were informedthat the aim of the study was to assess the interobserver

variation among pathologists for their diagnostic opinionof small atypical foci in prostate biopsies. The 5 invitedinternationally renowned urologic pathologists and 7 of8 members of the pathology committee of the ERSPCagreed to participate in the study. They were onlyinformed that the study comprised 20 separate cases,but the case composition was not revealed to them.Further, anonymity of the responses of the individualparticipants was promised. Subsequently, a unique identi-fication number was assigned to each of the participants,which was only known to the website producer (K.W.). Toperform the study, the participants accessed the websitefrom their desktop computer. The website displayed each ofthe 20 biopsy cases, with corresponding drop down menufor a total of 5 alternatives for diagnosis: (1) benign, nototherwise specified, (2) benign, atypical adenomatoushyperplasia, (3) benign, atrophy, (4) high-grade prostaticintraepithelial neoplasia (PIN), (5) high-grade PIN withatypical glands, suspicious for adenocarcinoma, (6) high-grade PIN associated with adenocarcinoma, (7) atypicalglands, suspicious for adenocarcinoma (ASAP), and (8)microfocus of adenocarcinoma. In addition, the partici-pants were able to provide comments on the individualcases. Clinical data such as patient’s age, serum PSA level,or clinical findings were not provided.

Data Collection and StatisticsCompleted questionnaires could be downloaded,

printed and faxed, or e-mailed to the independent datacollector associated with the study website (K.W.—Objective Pathology). The anonymized data were pro-vided to the study coordinator (T.H.v.d.K.) who collatedthem in an Excel file, after compression of the diagnoses

TABLE 1. List of Biopsies With Original Diagnosis and TheirPathology History

Case UHN Diagnosis No. Acini Immunohistochemistry

1 Adenocarcinoma 11 NA2 Adenocarcinoma 10 NA3 Suspicious 3 NA4 Suspicious 2 NA5 Suspicious 7 NA6 Benign 26 NA7 Benign 6 NA8 Adenocarcinoma 3 p63/AMACR9 Adenocarcinoma 13 NA

10 Suspicious 3+PIN NA11 Adenocarcinoma 3 NA12 Benign 2 NA13 Suspicious 36 p63 and AMACR14 Suspicious 3 p6315 Suspicious 4 HMWK and AMACR16 Suspicious 5 HMWK17 Benign 12 p63 and AMACR18 Adenocarcinoma 10 p6319 Adenocarcinoma 8 p6320 Adenocarcinoma 20 p63 and AMACR

AMACR indicates alpha-methyl CoA racemase; HMWK, high molecularweight keratin; NA, not available; PIN, prostatic intraepithelial neoplasia; UHN,University Health Network.

Van der Kwast et al Am J Surg Pathol � Volume 34, Number 2, February 2010

170 | www.ajsp.com r 2010 Lippincott Williams & Wilkins

in 3 categories: (1) adenocarcinoma, (2) suspicious foradenocarcinoma, and (3) benign, which includes atrophy,atypical adenomatous hyperplasia, and high-grade PIN.Multiple-reader k statistics and 95% confidence interval(CI) were calculated for each of the 2 groups of patho-logists (experts and ERSPC pathologists).

RESULTSThe diagnoses of the expert and all-round (ERSPC)

pathologists are listed in Table 2. The k value for theexpert pathologists (0.39; CI, 0.29-0.49) was significantlyhigher than that for the ERSPC pathologists (0.21; CI,0.14-0.27). The 5 experts achieved 100% agreement for7 of 20 biopsies, with consensus for 4 adenocarcinomas,2 benign diagnoses, and 1 suspicious diagnosis. These7 consensus diagnoses of the experts were identical to thediagnoses originally reported by the UHN (specialist)pathologists. The 7 ERSPC pathologists achieved 100%agreement for 1 biopsy (case 6), also diagnosed as benignby the urologic experts and UHN pathologists. If theagreement was calculated based on the original UHNpathologists’ diagnoses of adenocarcinomas or benign,the k value of the expert pathologists increased to 0.42and 0.55, respectively and those of the ERSPC patholo-gists to 0.26 and 0.27, respectively.

Expert and ERSPC pathologists rendered a highlyvariable diagnosis on the same biopsy varying frombenign to suspicious and to adenocarcinoma in 5 (ie, cases3, 11, 12, 15, and 16; Table 2A) and 10 biopsies(Table 2B), respectively (Figs. 1A, B, Figs. 2A, B, andFigs. 3A, B). For 4 of the 5 cases with expert pathologistdiagnoses ranging from benign to adenocarcinoma, theoriginal UHN diagnosis was suspicious, in the remainingcase (case 11) the UHN diagnosis was adenocarcinoma(Fig. 2A).

The expert pathologists diagnosed an adenocarci-noma significantly (P=0.001) more often (49%) than theERSPC pathologists (32%), at the expense of the numberof benign diagnoses, whereas the number of suspiciousdiagnoses remained about the same. The frequency ofadenocarcinoma diagnoses among the 5 experts (Table2A) varied from 6 to 13 biopsies (median 10 biopsies) andamong the 7 ERSPC pathologists (Table 2B) from 2 to 11biopsies (median 7 biopsies). Further, the number ofsuspicious diagnoses (ASAP) by experts varied from 3to 8. Strikingly, 4 of the 5 experts considered case 18(reported by UHN as adenocarcinoma) an adenocarci-noma, whereas 6 of 7 ERSPC pathologists regarded thelesion as suspicious for adenocarcinoma (Figs. 3C, D).The morphology of this lesion involving about 1.5mm ofthe biopsy demonstrated multiple small-to-medium� sizedglands, with an abnormal architecture, and prominentnucleoli in several nuclei, but with a rather pale cytoplasm,indistinguishable from that of the surrounding benignglands.

The expert pathologists were significantly less inagreement with the UHN diagnoses for the suspiciousfoci (p=0.01). For the ERSPC pathologists, agreement

with the UHN pathology diagnosis was highest for thebenign cases, intermediate for the carcinoma cases, andlowest for the suspicious cases.

As cases 3, 11, 12, 15, and 16 elicited the widestrange of diagnostic opinions among the 5 experts, weanalyzed their histopathologic features in more detail.They all represented very small lesions (Fig. 1A, Figs. 2A,B, Figs. 3A, B), consisting of 2 to 5 small-sized acini,extending <1mm of the biopsy. Case 13 (Figs. 2C, D)was diagnosed as benign by 4 of 5 experts and 5 of 7ERSPC pathologists, and as adenocarcinoma by 1 expertand 1 ERSPC pathologist. In 2 cases, the nuclear featureswere difficult to assess, as the nuclei lacked detail andprominent nucleoli were not identified unequivocally (eg,case 11, Fig. 2A). In Figure 4 we show 2 cases, which hada better agreement of diagnoses among expert patholo-gists (4A, B) than the ERSPC pathologists.

DISCUSSIONIt may sometimes be challenging for the pathologist

to deliver a definite diagnosis of adenocarcinoma inprostate biopsies, particularly if the size of the lesion istoo small to judge the presence of an infiltrative pattern.Under these circumstances, the pathologist needs to relyon other parameters, such as architectural and cyto-nuclear atypia and the lack of basal cells as witnessedby immunostaining for basal cell markers.11 This issuehas become more pertinent in recent years due tostage reduction of prostate cancer as the consequenceof widespread PSA testing and increased numbers ofbiopsies. Also repeat biopsies in the context of activesurveillance treatment might lead to an increased fre-quency of small foci of adenocarcinoma, high-grade PIN,and lesions reported as suspicious for malignancy orASAP. Although ancillary techniques may be of help toconfirm a morphologic diagnosis of adenocarcinomabased on hematoxylin-eosin� stained slides, pathologistsshould not fully rely on them. Several mimickers of smallfoci of prostate cancer, such as atypical adenomatoushyperplasia and some morphologic variants of atrophymay have a similar immunoprofile, that is, lack of basalcells and positivity for racemase.1,8,11 Further, the smallfocus of atypical gland suspicious for cancer may not bepresent any more in the section used for immunohisto-chemistry. Two examples of the latter were represented inour virtual microscopy biopsy series. Although weprovided immunostained slides for 10 of the 20 cases, itis likely that the absence of additional stains in some ofthe remaining cases could have increased the number ofsuspicious diagnoses, whereas reducing the interobserveragreement for some lesions.

The focus of this study was on the very smallproliferations of atypical glands, which cause mostuncertainty in the daily pathology practice. Pathologistsencountering these single small atypical foci may betriggered to consult a colleague or ask a second opinionfrom an expert in the field. As was previously reported,second opinion for cases diagnosed as suspicious or

Am J Surg Pathol � Volume 34, Number 2, February 2010 Single Small Atypical Foci in Prostate Biopsies

r 2010 Lippincott Williams & Wilkins www.ajsp.com | 171

TABLE 2. Distribution of Diagnoses by Urologic Experts (A) and European RandomizedStudy of Screening for Prostate Cancer Pathologists (B)

A

1 1 1 1 1 1

2 1 1 1 1 1

2 1 2 2 3 2

1 1 1 1 2 2

1 1 1 2 2 2

3 3 3 3 3 3

2 3 2 3 3 3

1 1 1 1 1 1

1 1 1 1 1 1

2 2 2 2 2 2

2 2 1 3 2 1

3 2 2 1 1 3

3 3 3 1 3 2

2 1 1 1 2 2

3 2 1 1 1 2

3 3 2 1 2 2

3 3 3 3 3 3

2 1 1 1 1 1

1 1 1 1 1 1

2 2 1 1 2 1

Nr of susp 8 5 5 3 7 8

6 10 12 13 8 8

UGE 1 UGE 2 UGE 3 UGE 4 UGE 5 UHNDx

Case 1

Case 2

Case 3

Case 4

Case 5

Case 6

Case 7

Case 8

Case 9

Case 10

Case 11

Case 12

Case 13

Case 14

Case 15

Case 16

Case 17

Case 18

Case 19

Case 20

B ERSPC 1 ERSPC 2 ERSPC 3 ERSPC 4 ERSPC 5 ERSPC 6 ERSPC 7 UHN Dx

Case 1 1 1 1 1 2 1 1 1

Case 2 2 1 2 2 1 1 1 1

Case 3 2 2 2 3 3 2 1 2

Case 4 1 2 3 1 3 1 1 2

Case 5 1 3 2 2 3 2 1 2

Case 6 3 3 3 3 3 3 3 3

Case 7 3 3 3 3 3 2 3 3

Case 8 1 3 1 1 1 1 1 1

Case 9 1 2 1 1 2 1 1 1

Case 10 2 3 2 2 3 2 2 2

Case 11 2 3 2 2 3 3 1 1

Case 12 3 3 3 1 1 2 2 3

Case 13 3 3 3 2 1 3 3 2

Case 14 1 3 2 3 2 2 2 2

Case 15 2 3 3 1 3 1 2 2

Case 16 3 3 3 2 3 2 3 2

Case 17 3 2 3 3 1 3 2 3

Case 18 2 2 2 2 2 2 1 1

Case 19 1 2 1 2 2 1 1 1

Case 20 1 3 3 1 1 2 1 1

Nr susp 6 6 7 8 5 9 5 8

Nr of ca 8 2 4 7 6 7 11 8

Nr of ca

Squares represent the diagnoses.Black 1 indicates adenocarcinoma; gray 2, suspicious; white 3, benign.

Van der Kwast et al Am J Surg Pathol � Volume 34, Number 2, February 2010

172 | www.ajsp.com r 2010 Lippincott Williams & Wilkins

ASAP have a high likelihood to be changed into either abenign or adenocarcinoma diagnosis upon review by anexpert, reducing the number of uncertain diagnoses.4,14

Our study demonstrates very clearly that the expert patho-logists diagnosed adenocarcinoma significantly more oftenthan their generalist colleagues and 3 of the 5 expertpathologists also diagnosed adenocarcinoma more fre-quently (n=10, 12, and 13) than the UHN (specialist)pathologists (n=8). Although suspicious (ASAP) diag-noses by experts varied markedly, ranging from 3 to 8cases, 4 of the 5 experts rendered less often a suspiciousdiagnosis as compared with the original (n=8) UHNpathologists, supporting the observation that expertreview will more often lead to a definitive diagnosis.4,14

Some caution is, however, warranted, as for 3 cases,originally reported by UHN pathologists as suspicious,the definite diagnosis by the experts varied from benign toadenocarcinoma, reflecting the potential for substantialinterobserver variation among the expert pathologists.This underlines the view that for these small atypical focino gold standard exists, and even a majority opinion of

experts would not necessarily reflect the ‘‘true biology’’ ofthe lesion.

It may happen that second opinion is asked formedicolegal reasons, for instance when no cancer isidentified in a prostatectomy after a needle biopsydiagnosis of adenocarcinoma.2,3 Analysis of literaturesuggests that identification of a single small focus ofadenocarcinoma in a prostate biopsy increases the like-lihood of this phenomenon in a subsequent prostatect-omy specimen.9,15 Potentially, 3 main reasons can explainthis: (1) the lesion is too small to be identified duringroutine work-up of the prostatectomy specimen,10 (2)the diagnostic interpretation on the needle biopsy wasincorrect, and (3) the occurrence of a mix-up of biopsyspecimens.12 The results of our study could imply thatfor some single small atypical foci, the diagnosis issuedby reviewing expert pathologists may vary from benignto suspicious and to adenocarcinoma. An overview ofmalpractice claims in California during the periodbetween 1998 and 200317 showed that ‘‘misdiagnoses’’on prostate biopsies represented a minority of the cases

Case 3 Case 3

BA

Case 5 Case 7

C D

FIGURE 1. Micrographs of cases 3 (A and B), 5 (C), and 7 (D): Micrographs 1 A (20� ) and B (40� ) depict 3 small atypical aciniwith slight cytonuclear atypia, including somewhat conspicuous, but not prominent nucleoli. Case 5 (C) represents a larger lesionof about 7 atypical glands, and the insert demonstrates the presence of a rather prominent nucleolus. Case 7 (D) is characterizedby a few small-to-medium� sized glands with eosinophilic cytoplasm and monotonous small bland nuclei.

Am J Surg Pathol � Volume 34, Number 2, February 2010 Single Small Atypical Foci in Prostate Biopsies

r 2010 Lippincott Williams & Wilkins www.ajsp.com | 173

(9/335 pathology claims). Although in a previous reviewperiod, the false positive diagnoses of prostate cancerdominated; in this latest review, only 3 of the 9 casesrepresented a false positive diagnosis. Our study demon-strates that even experts in urologic pathology can havedifferent diagnostic thresholds for making a diagnosisof a microfocus of adenocarcinoma. Given the observedvariation in expert opinion for some ‘‘atypical foci’’ itfollows that it may be challenging to demonstrate in suchcases a ‘‘misdiagnosis’’ of the biopsy when associated witha pT0 carcinoma in the corresponding radical prosta-tectomy.

In a previous study, comparison of foci reported assuspicious for carcinoma (ASAP) and adenocarcinomarevealed that suspicious lesions had a mean extent of0.4mm, comprising in average 11 acini, whereas small

foci of carcinoma measured in average 0.8mm andconsisting of 17 acini, but with considerable overlap.13

We noticed that the largest variation in diagnoses on abiopsy by expert pathologists was for the subset of lesionscomprising 5 or fewer atypical glands. As these lesionsoften seem to be challenging, it is our opinion that inparticular generalist pathologists should be cautious inissuing a diagnosis of adenocarcinoma on lesions of thissmall size. The majority of radical prostatectomiessubsequent to an initial diagnosis of ‘‘focal’’ or ‘‘minute’’cancer harbor ‘‘insignificant’’ prostate cancers, with aminority of specimens (up to 3%) containing a pT0cancer and in about 5% to 8%, a carcinoma withextraprostatic extension can be found.18 Due to thepotential for variation in diagnostic opinion amongpathologists for single small atypical foci and the wide

Case 12

B

Case 11

A

Case 13 Case 13

DC

FIGURE 2. Micrographs of cases 11 (A, 40� ), 12 (B, 40� ), and 13 (C, 20� and D, 20� ). In case 11, a row of 3 to 4 small-sizedglands demonstrate nuclear hyperchromasia, with some nuclear smudging obscuring the nuclear details. Subsequentimmunostaining (not provided with the study) with revealed absence of basal cell staining (p63) and positivity for AMACRin these glands. Case 12 shows a few very small-sized glands scattered between the antecedent glands difficult to identify at lowpower. They are composed of flattened cells with dense bluish cytoplasm and some nuclear enlargement (B). Theimmunostaining (cocktail of high molecular weight keratin and AMACR) highlights the atypical glands (insert in B). Case 13consists of a larger focus of small-to-medium� sized glands with architectural atypia and cytonuclear atypia (insert in A), but thelesion is not showing obvious infiltration (C). Immunostaining with p63 (D) reveals a few scattered positive cells lining some butnot all of the glands within the lesion. AMACR indicates alpha-methyl CoA racemase.

Van der Kwast et al Am J Surg Pathol � Volume 34, Number 2, February 2010

174 | www.ajsp.com r 2010 Lippincott Williams & Wilkins

range of cancer features found in subsequent correspond-ing prostatectomy specimens, clinicians may considerrepeat (staging) biopsies for these lesions to select potentialcandidates for definite or deferred (active surveillance)therapy.

Our virtual microscopy study has several limita-tions. First, the lesional area was not marked on the slide.Thus, a benign diagnosis rendered by a participant of thisstudy may not necessarily be attributed to the interpreta-tion of the abnormality, but this may also reflect thepossibility that the lesion was overlooked. The lattercould in fact explain several benign diagnoses made byERSPC pathologist 2 who considered biopsies 8 and 20 asbenign, whereas most other pathologists diagnosed thislesion either as adenocarcinoma or suspicious for cancer(ASAP). Second, we must also recognize the possibility ofa study bias, whereby some of the pathologists may not

have devoted the same level of careful analysis that theywould apply to a diagnostic case that has actualimplications for a patient. For example, expert patholo-gist 1 diagnosed cases 12, 15, and 16 as benign, wheremost of the other experts diagnosed these biopsies eitheras suspicious or adenocarcinoma, implying that thesuspicious focus was overlooked. Another limitation ofthe study is that some pathologists may have ‘‘aggres-sively’’ diagnosed carcinoma in this study in view of thelack of implications for patient care, whereas in practicethey might have diagnosed the same biopsies as suspi-cious (ASAP). Other experts may have approached thecurrent study attempting to diagnose lesions in an entirelyanalogous fashion to a clinical case. This differencein approach could have accounted for some of thediscrepancy between suspicious and carcinoma diagnosesamongst the expert pathologists.

Case 16 Case 15

BA

Case 18

C

Case 18

FIGURE 3. Micrographs of cases 15 (A, 20�), 16 (B, 40�), and 18 (C, 10� and D). In case 15 (A), about 4 very small-sizedglands, mostly without a lumen may be noted with considerable cytonuclear atypia. Immunostaining with high molecular weightcytokeratin demonstrates lack of a few glands at the edge of the biopsy (insert). Case 16 (B) consists of a few longitudinally cutglands composed of cells with dense cytoplasm and slightly hyperchromatic nuclei, but nuclear details cannot easily bedistinguished. Immunostaining was performed, but the focus had disappeared. The lesion in case 18 (C and D) consists of 2separate, but nearby foci of small-sized glands with pale cytoplasm and prominent nucleoli (D, 40�). Immunostaining for p63was negative for the atypical glands (insert in C).

Am J Surg Pathol � Volume 34, Number 2, February 2010 Single Small Atypical Foci in Prostate Biopsies

r 2010 Lippincott Williams & Wilkins www.ajsp.com | 175

Finally, the suitability of virtual microscopy todiagnose these subtle prostatic lesions may be questioned.Some users reported that they experienced slowness in theacquisition of the images, particularly during the scanningof the biopsies at low power to identify the single atypicalfocus. This may have influenced the diagnostic process.On the other hand, similar virtual microscopy studiesshowed excellent results for assigning a Gleason score6

and for the diagnosis of small foci of adenocarcinomausing web-based virtual microscopy.5,7 In our study, theslides were scanned at an objective 40� magnification,which allowed a sufficient resolution to identify promi-nent nucleoli if present. Another advantage of the use ofvirtual microscopy is the assurance that all observersexamined exactly the same image.

In conclusion, this study demonstrates that expertpathologists show a better agreement for single smallatypical foci than their all-round (ERSPC) colleagues. Inaddition, most but not all experts diagnosed adenocarci-noma more frequently than the ERSPC pathologists andthe UHN (specialist) pathologists. However, for most ofthe foci comprising less than 6 small-sized acini, eventhe experts rendered highly variable diagnoses. Cliniciansshould be aware of this potential for variation in dia-gnostic opinion. We would encourage that generalistpathologists obtain intercollegial consultation before acarcinoma diagnosis on atypical foci of less than 6atypical glands.

REFERENCES1. Adley BP, Yang XJ. Alpha-methylacyl coenzyme A racemase

immunoreactivity in partial atrophy of the prostate. Am J ClinPathol. 2006;126:849–855.

2. Bostwick DG, Bostwick KC. Vanishing prostate cancer in radicalprostatectomy specimens: incidence and long-term follow-up in 38cases. BJU Int. 2004;94:57–58.

3. Cao D, Hafez M, Berg K, et al. Little or no residual prostate cancerat radical prostatectomy: vanishing cancer or switched specimen?: amicrosatellite analysis of specimen identity. Am J Surg Pathol. 2005;29:467–473.

4. Epstein JI, Herawi M. Prostate needle biopsies containing prostaticintra-epithelial neoplasia or atypical foci suspicious for carcinoma:Implications for patient care. J Urol. 2006;175:820–834.

5. Fine JL, Grzybicki DM, Silowash R, et al. Evaluation of whole slideimage immunohistochemistry interpretation in challenging prostateneedle biopsies. Hum Pathol. 2008;39:564–572.

6. Helin H, Lundin M, Lundin J, et al.Web-based virtual microscopyin teaching and standardizing Gleason grading. Hum Pathol. 2005;36:381–386.

7. Helin HO, Lundin ME, Laakso M, et al. Virtual microscopy inprostate histopathology: simultaneous viewing of biopsies stainedsequentially with hematoxylin and eosin, and alpha-methylacyl-coenzyme A racemase/p63 immunohistochemistry. J Urol. 2006;175:495–499.

8. Herawi M, Parwani AV, Irie J, et al. Small glandular proliferationson needle biopsies: most common benign mimickers of prostaticadenocarcinoma sent in for expert second opinion. Am J SurgPathol. 2005;29:874–880.

9. Herkommer K, Kuefer R, Gschwend JE, et al. Pathological T0prostate cancer without neoadjuvant therapy: clinical presentationand follow-up. Eur Urol. 2004;45:36–41.

10. Humphrey PA. Complete histologic serial sectioning of a prostategland with adenocarcinoma. Am J Surg Pathol. 1993;17:468–472.

11. Humphrey PA. Diagnosis of adenocarcinoma in prostate needlebiopsy tissue. J Clin Pathol. 2007;60:35–42.

12. Iczkowski KA, MacLennan GT, Bostwick DG. Atypical smallacinar proliferation suspicious for malignancy in prostate needlebiopsies: clinical significance in 33 cases. Am J Surg Pathol. 1997;21:1489–1495.

13. Iczkowski KA, Bostwick DG. Criteria for biopsy diagnosis ofminimal volume prostatic adenocarcinoma: analytic comparisonwith nondiagnostic but suspicious atypical small acinar prolifera-tion. Arch Pathol Lab Med. 2000;124:98–107.

14. Novis DA, Zarbo RJ, Valenstein PA. Diagnostic uncertaintyexpressed in prostate needle biopsies. A College of AmericanPathologists Q-probes Study of 15,753 prostate needle biopsies in332 institutions. Arch Pathol Lab Med. 1999;123:687–692.

15. Postma R, de Vries SH, Roobol MJ, et al.Incidence and follow-upof patients with focal prostate carcinoma in 2 screening rounds afteran interval of 4 years. Cancer. 2005;103:708–716.

Case 20

BA

Case 19

FIGURE 4. Micrographs of cases 19 (A, 20�) and 20 (B, 40�). Case 19 displays a few medium-to-large-sized glands, lined by cellswith dense cytoplasm and prominent nucleoli. The immunostaining with p63 fails to stain these glands (insert). The lesional areain case 20 (D) consists of a mixture of small-sized glands, 1 set with the features of atrophic glands (glands with small,hyperchromatic basal-like nuclei) and the other with somewhat larger, more open nuclei, and more pale cytoplasm. The latterglands lack basal cells as is apparent from the lack of p63 staining (insert).

Van der Kwast et al Am J Surg Pathol � Volume 34, Number 2, February 2010

176 | www.ajsp.com r 2010 Lippincott Williams & Wilkins

16. Postma R, Schroder FH, van Leenders GJ, et al. Cancer detectionand cancer characteristics in the European Randomized Study ofScreening for Prostate Cancer (ERSPC)–Section Rotterdam. Acomparison of two rounds of screening. Eur Urol. 2007;52:89–97.

17. Troxel DB. Medicolegal aspects of error in pathology. Arch PatholLab Med. 2006;30:617–619.

18. Van der Kwast TH, Wolters T, Evans A, et al. Single prostaticcancer foci on prostate biopsy. Eur Urol Supp. 2008;7:549–556.

19. Vis AN, Van der Kwast TH. Prostatic intra-epithelial neoplasia andputative precursor lesions of prostate cancer: a clinical perspective.BJU Int. 2001;88:147–157.

20. Zackrisson B, Aus G, Bergdahl S, et al. The risk of findingfocal cancer (less than 3mm) remains high on re-biopsy ofpatients with persistently increased prostate specific antigenbut the clinical significance is questionable. J Urol. 2004;171:1500–1503.

Am J Surg Pathol � Volume 34, Number 2, February 2010 Single Small Atypical Foci in Prostate Biopsies

r 2010 Lippincott Williams & Wilkins www.ajsp.com | 177