Pierre Nevoux, Department of Urology, Claude Huriez Hospital, CHRU 59037 Lille, France. e-mail: email@example.com
What's known on the subject? and What does the study add?
Much of our understanding of the pathological basis of prostate cancer comes from our analysis of radical prostatectomy specimens. Prostate cancer diagnosed by transrectal ultrasonography-guided biopsy is more likely to be posterior and basal in orientation rather than anterior or apical. Quantitative tissue analyses have not been undertaken both with details and in an unselected population, e.g. prostate specimens from autopsy cystoprostatectomy series from bladder cancer.
Quantitative tissue analysis of incidentally detected prostate cancer such as largest cancer surface area, volume, site of origin, multifocality and laterality could be of paramount importance when trying to understand the findings of screen-detected programmes and focal therapy. Cancers were found in 30% of prostates. In the 96 prostates, 215 cancer foci were identified (mean 2.24). Prostate cancer was multifocal in 60% and bilateral in 80% of cases. The site of origin was in the peripheral and transition zone (TZ) in 75% and 25%, respectively. Overall, 90% of cancer foci were clinically insignificant with volume of <0.5 mL and no grades 4–5. In all, 75% of the cancer foci were in the peripheral zone, the remainder were within the TZ. One third of cancer foci were anteriorly located beyond the area sampled by posterior biopsies. One fifth of cancer foci were ≤6 mm of the apex.
• To describe multifocality, volume and location of prostate cancers incidentally found in cystoprostatectomy specimens. Quantitative tissue analysis of prostate cancer in a population free of the evaluation bias associated with prostate-specific antigen level and biopsy is important as some men are likely to be offered tissue-preserving therapeutic strategies in the future.
PATIENTS AND METHODS
• Cystoprostatectomy specimens for bladder cancer from 345 consecutive patients without clinically manifest prostate cancer were included.
• Cancers were found in 104/345 (30%) of prostates. Cases with largest cancer >2 mL (eight patients) were excluded from morphometric study. Quantitative tissue analysis of 3-mm step-sectioned glands included largest cancer surface area, volume, site of origin, multifocality and laterality.
• In the 96 prostates, 215 cancer foci were identified (mean 2.24). Prostate cancer was multifocal in 58% and bilateral in 79% of cases.
• Of the 215 cancers, 90% were <0.5 mL and 79% <0.2 mL. Overall, 88% of cancer foci were clinically insignificant with a volume of <0.5 mL and no grades 4–5.
• In all, 75% of the cancer foci were in the peripheral zone, the remainder were within the transition zone.
• One third of cancer foci were anteriorly located beyond the area sampled by posterior biopsies. One fifth of cancer foci were ≤6 mm of the apex.
• Limitations include the fact that cystoprostatectomy cancer foci are at an earlier stage than screened-detected cancers.
• This detailed morphometric analysis of prostate cancer foci in a population that is free from the selection bias associated with screening can help inform our diagnostic and treatment strategies.
Much of our understanding of the pathological basis of prostate cancer comes from our analysis of radical prostatectomy (RP) specimens. This has served us well, informing our understanding of both the disease burden at the time of surgery and also in relation to future progression of treated disease modeled on preoperative baseline variables . Much of the last 20 years has been spent trying to understand the relationship between these baseline variables (age, PSA level, clinical stage, cancer grade and burden) vs the ‘true’ grade and burden of disease within the prostate. However, the validity of this approach is starting to be questioned as the products of it are subject to sequential, cumulative and ill-understood selection factors. A high PSA level is over-represented in older men and in men with large prostates. Because of confounding, as well as other ill-understood factors, its use has not always resulted in the observations that we would expect. PSA screen-detected prostate cancer has, in one recent analysis, been shown to exhibit more aggressive components than prostate cancer diagnosed incidentally and verified by RP . To gain a more complete understanding of the number, location, and burden of prostate cancer lesions we need to look at prostates that are free of the three selection biases (biomarker expression, random and systematic error associated with biopsy, systematic and preference sensitive application of RP) to better understand the true morphometric epidemiology of prostate cancer. Prostate cancer diagnosed by TRUS-guided biopsy is more likely to be posterior and basal in orientation rather than anterior or apical. Detailed morphometric analysis has not been undertaken in an unselected population, and could be of paramount importance when trying to understand the findings of screen-detected programmes and treatment options.
In this article we describe in detail the qualitative and quantitative features of prostate cancer foci in an unselected population.
PATIENTS AND METHODS
The present study was approved by the ethical committee from Stanford Medical Center. We evaluated 346 consecutive patients without clinically manifest prostate cancer at DRE or PSA level (after 1986) and who underwent cystoprostatectomy for invasive bladder cancer from 1983 to 1997 (Department of Urology, Stanford University School of Medicine, CA, USA). Preoperative data collected included age, PSA level and prostate volume for all patients with cancer. One case was excluded due to a lack of histological slides for review. Prostate cancer was found in 104/345 (30%). Eight patients with large volumes of cancer >2 mL were excluded, as it was impossible to determine multifocality and site of origin due to most of the gland being involved with cancer. In these eight cases, the mean (median; range) volumes of the eight largest separate cancers was 5.665.38; 2.57–12.13) mL. Overall 96 cases with prostate cancer were analysed.
Prostate specimens were inked, fixed and sectioned according to the Stanford protocol , after being dissected from the bladder. Cancer morphometric analysis included largest surface area, volume, grade and spatial distribution. A reconstructed histological map of each prostate was created. Contours of histological zones as well as the outlines of each cancer were drawn on the slides under the microscope and then transferred onto tracing paper. The number of separate cancers for each specimen was determined as follows: two foci were considered as separate cancers if separated by ≥3 mm. Cancer surfaces were measured with computer software (ImageJv1.38e freeware). Cancer volumes were calculated by multiplying the resulting tumour areas by section thickness (3 mm) and a predetermined correction factor (1.5) to correct for tissue shrinkage during fixation . For surfaces (cm2) and distances (mm), the correction factors were derived from the 1.5 correction factor for volume and were 1.22 and 1.10, respectively. Definition of clinical significance of a cancer foci is presence of grade 4 or 5 or tumour volume >0.2 mL or 0.5 mL [5–7].
In multifocal cases, the distance between the centres of the largest and the second largest cancer was measured (Fig. 1). The distance we calculated corresponded to the hypotenuse of the right-angled triangle of which the vertex was the projection of the centre of one cancer on the reference plane of the other cancer. Each center of cancer was located in its reference plane. The sum of levels between the two reference planes multiplied by 3 mm (slice thickness) yielded the cranio-caudal distance. Transverse and cranio-caudal distances were measured manually and the hypotenuse was calculated.
In the 96 cases assessed, the mean (range) age was 5930–87) years and the mean (interquartile) range PSA level and prostate volume was 2.81.1–3.8) ng/mL and 39.730–47) mL, respectively. There was leason grade 4 or 5 in 12 (12%) cases and was confined to the index lesion in all prostates (index lesion volume was >0.5 mL in eight prostates and <0.5 mL in four). In these 12 cases, there was extraprostatic extension in two cases with index cancer volumes of 1.3 mL and 1.96 mL. Surgical margins were positive in six (6%) cases with a mean (range) index cancer volume of 1.240.45–1.96) mL.
Within the 96 prostates, 215 prostate cancers foci were identified. The location of the 215 separate cancers with reference to the fixed points of anterior and posterior limits of the prostate, the anatomical zones of the prostate and finally to the cranio-caudal axis are shown in Table 1 and Fig. 2. The distribution of the largest cancer surfaces and volumes of cancers are shown in Table 2. The volumes were <0.5 mL, <0.2 mL and <0.1 mL in 90%, 79% and 71%, respectively. The sum of 96 largest cancers volumes made up >77% (26.52/34.33 mL) of the sum of all 215 cancer volumes. The contours of the largest cancer surfaces according to volume ranges are shown in Fig. 3. In all, 75% of the cancer foci were peripheral in location, the remainder were located within the transition zone (TZ) and two thirds (66%) were posterior. One fifth (20%) of cancer foci were at ≤6 mm from the apical limit. Half of these apical cancers (22/44) were anterior.
Table 1. Location distribution of the 215 separate cancers in a series of 96 cystoprostatectomy specimens according to the anterior and posterior parts of the gland, the histological zones PZ and TZ and to the cranio-caudal axis. Anterior and posterior parts of the gland were defined according to an anterior-posterior separation line, 17 mm distant from the rectal surface, which corresponds to the length of an 18-G needle biopsy notch
Cancer frequency in posterior part vs anterior part of the gland, n (%)
Cancer frequency in PZ vs TZ and AFMS, n (%)
TZ and AFMS
Base n= 59
Mid n= 96
Apex n= 60
Whole gland n= 215
Table 2. Characteristics of largest volumes of the 215 cancers of ≤2 mL found in the 96 cystoprostatectomy specimens
Cancer volume, mL
Largest cancer volume
Total cancer volume
Sum of all smaller cancer volume
The mean number of cancer foci per prostate was 2.24. Multifocality and laterality are shown in Fig. 4. In multifocal cases, distances between the largest foci and following two largest foci according to their centres and nearest limits are shown in Table 3.
Table 3. Distances between the largest lesions and following two largest lesions according to their centres and nearest limits and their laterality (lesion 1 is the largest lesion, lesion 2 is the second largest lesion and lesion 3 is the third largest lesion in a same specimen)
Mean distance (range), mm:
All multifocal lesions
…between the centres of the largest surfaces of lesions 1 and 2
…between nearest contour borders of the largest surfaces of lesions 1 and 2
…between the centres of the largest surfaces of lesions 1 and 3
…between nearest contour borders of the largest surfaces of lesions 1 and 3
The present results show that prostate cancer was present in almost a third of these men. Prostate cancer was multifocal in 58% and most of them were bilateral (79%). However, all high-grade disease was confined to the index lesion and most of the contralateral cancer foci (to the index lesion) had characteristics that we would normally associate with indolent or clinically insignificant cancer. We think this to be the most detailed morphometric analysis to have been undertaken in an unselected population .
The first, relates to the population in question. Is the occurrence of radical cystoprostatectomy truly independent of prostate cancer risk? There is no way of being sure of any link between urothelial cancer and prostate cancer, so for the purposes of the present study we have assumed that the two are independent. The present 30% overall prevalence rate was nearly identical to the 32% prevalence rate of prostate cancer found in men who underwent autopsy in the sixth decade of life. That study and our own shared the 3-mm step-section sampling frame and, as a result, have comparable reference tests .
The second relates to the processing of the specimen. All steps were taken to account for gland distortion and gland shrinkage and appropriate adjustments were made. However, radical cystoprostatectomy specimens are still subject to sampling and at 3 mm some small foci were overlooked or were deemed to be one when in reality they were two. If a bias has resulted it will be in the direction of us declaring fewer cancers than are, in reality, present.
The third relates to our understanding of the pathogenesis of clinically important cancers from the large pool of clinically unimportant cancers that we know to exist from the third decade of life. In the Soos et al.  autopsy series, all the cancers in men aged 40 years were deemed to be insignificant and <0.5 mL. In the men aged >80 years, only half the cancer foci could be classified as insignificant. The present series was in between these two extremes, with a mean age of 60 years. Just over 1 in 10 of the foci that we detected was attributed clinical significance.
COMPARISON WITH OTHER STUDIES
Clearly, the present population has, on average, a lower age, a lower volume of largest cancer and a lower number of separate foci than we would see in a RP series (Table 4) [4,10–13]. From a quantitative perspective the features that we describe in the men with cancer are similar to the secondary prostate cancer foci found in certain RP series (Table 4). This is something we would expect as men who test negative to TRUS-guided biopsy are, on average, likely to have lower volume disease than those who test positive, given the random component of the test. As we have stated above, the present series differs from a RP population as it is free of the selection bias conferred by the sequential requirement of positive tests results. It would appear that PSA level coupled with TRUS-guided biopsy generates cases that have index lesions of greater volume and may exhibit more multifocality than the series derived from radical cystectomy . Having said this the mean age of the men with prostate cancer in the present series was closer to the mean age of published RP series  than it is to the mean age seen in cystoprostatectomy series . The site of origin and spatial distribution in the present series is different from the 59% and 41% reported in a RP series in which the decision to operate was predicated on the presence of posterior disease, based as it was on TRUS-guided posterior routine sampling (Table 4).
Table 4. Comparison between incidental (from our series of 96 cystoprostatectomies) and clinical prostate cancer from RP series. All series were assessed according to Stanford Protocol (transverse sections at 3 mm)
Prostate cancer features
Mean or median age, years
Mean PSA level, ng/mL
Mean prostate volume, mL
Number of separate cancer
Insignificant prostate cancer (<0.5 mL, no Gleason pattern 4/5), %
Median total largest tumour volume, mL
Median total smallest tumour volume, mL
% location PZ/TZ
% extraprostatic extension/surgical margins
Absence of Gleason pattern 4/5 not included; –, unknown.
Looking now at other cystoprostatectomy series, it would seem that our own morphometrics concur with those of others [15,16]. In the series reported by Mazzucchelli et al.  and Barbisan et al.  most of the cancer foci were deemed insignificant (87% and 81%, respectively). The presence of apical disease varies widely according to the definition used. In the present study, apical cancer foci (≤6 mm of the apical limit) were seen in one fifth of prostates. In two other cystoprostatectomy series, the prevalence of apical cancers varied between 2% in Montironi et al. series  and 60% in the Revelo et al.  (≤10 mm of the apical limit). In the present study, half of the apical cancers were anteriorly located.
There are four important clinical implications to the present findings.
The first relates to biopsy strategies. RP specimens tell us about the cancers that PSA-TRUS biopsy found. They do not tell us about the ones that were over-looked as these patients do not go to surgery. The present data, free as it is of selection bias, can assist in informing on the optimal biopsy strategy that should allow it to perform better than chance. In Fig. 3, posterior systematic biopsies would have diagnosed all of posterior peripheral zone (PZ) cancers of >0.2 mL, as they fill most of one of the 12 sectors sampled by these posterior biopsies. To detect the rest of anteriorly located cancers, an anterior systematic biopsies scheme should have sampled anterior horns of the PZ at mid-gland, anterior third of the TZ and anterior fibromuscular stroma (AFMS) at mid gland and tissue anterior to the urethra at apex.
The second concerns the diagnostic strategies that we might use in the future . The most promising is imaging the prostate before biopsy, either as a triage test to reduce the number of unnecessary biopsies or to target biopsies, in those that need them, to areas of interest. Modern, multi-sequence MRI performs well at volumes >0.5 mL and will systematically under-detect lesions that are ≤0.2 mL . In present series these were all Gleason 3+3 lesions and almost certainly indolent in nature.
The third relates to the emerging technique of focal therapy for prostate cancer . Multifocality presents a challenge to the number of patients that may prove eligible for such an approach. In the present series the multifocality was limited to one side of the prostate. This does raise the possibility of hemi-ablation in a significant group of men. Our average distances between ipsilateral foci should help in tissue preservation strategies.
Finally, the analysis of prostate cancer that has not been subject to any form of sampling artefact gives us insight into the pathogenesis of clinically important prostate cancer. It is a reasonable assertion that if clinical detection tends to yield more burden of disease than we would expect by chance clinically and incidentally detected cancers and have the same heterogeneous biological potential regardless of method of detection. Clinically detected cancers are selected from the wider pool of cancer. They tend to be larger, as the coupling of PSA level and TRUS-guided biopsy is more likely to detect a larger cancer than a smaller cancer assuming everything else to be equal . Larger cancer foci are, in turn, positively associated with higher Gleason grade . What is not known is the efficiency by which our current diagnostic strategies can select men with significant prostate cancer from the pool of men that truly have this. Detection strategies at the time were not as sensitive (PSA level and six TRUS biopsies) as current strategies.
In conclusion, this detailed morphometric analysis of prostate cancer foci in a population that is free from the selection bias associated with screening has clinical implications for diagnostic and treatment strategies.
We would like to acknowledge the important contribution of John McNeal for the initial pathology report, Rosalie Nolley for the morphometry and database work, and Michel Gilles for the figures work.
CONFLICT OF INTEREST
Mark Emberton receives research support from the UK's National Clinical Research Institute University College London Hospitals/University College London Comprehensive Biomedical Research Centre, London, UK.