Fax: (011) 31 104089847
Potentially advanced malignancies detected by screening for prostate carcinoma after an interval of 4 years
Article first published online: 16 JAN 2004
Copyright © 2004 American Cancer Society
Volume 100, Issue 5, pages 968–975, 1 March 2004
How to Cite
Postma, R., Roobol, M., Schröder, F. H. and van der Kwast, T. H. (2004), Potentially advanced malignancies detected by screening for prostate carcinoma after an interval of 4 years. Cancer, 100: 968–975. doi: 10.1002/cncr.20048
- Issue published online: 18 FEB 2004
- Article first published online: 16 JAN 2004
- Manuscript Accepted: 3 DEC 2003
- Manuscript Revised: 24 NOV 2003
- Manuscript Received: 3 SEP 2003
- prostate carcinoma;
- PSA testing;
- needle biopsy;
- quantitative pathology
At the Rotterdam branch of the European Randomized Study of Screening for Prostate Cancer, a cohort of 19,970 men ages 55–75 years is screened at an interval of 4 years. Screening includes systematic sextant needle biopsy for men with elevated prostate-specific antigen (PSA) levels and/or positive findings on digital rectal examination or transrectal ultrasound. Detection during the second screening round of a large number of high-grade (Gleason Grade 4 or 5) malignancies and/or a large number of malignancies in general could be considered the result of a failure to identify these malignancies at an early stage, during prevalence screening.
Men diagnosed during the second screening round with potentially advanced carcinoma (PAC), characterized by a biopsy Gleason score of 7 (4 + 3, or 3 + 4 with > 30% malignant involvement) or a biopsy Gleason score of 8–10, were identified. Clinical data, including PSA values on prevalence screening, biopsy history, clinical stage, and follow-up data, were retrieved for these patients. Tumor features were further analyzed in radical prostatectomy specimens.
During the second screening round, 503 malignancies, including 30 (6.0%) with features of PAC on diagnostic biopsy, were detected in 11,210 patients. Curative treatment was offered to 26 patients. Prostatectomy demonstrated the presence of organ-confined disease in 11 of 12 specimens, and tumor volume ranged from 0.11–7.93 cm3 (median, 1.05 cm3). PSA failure was noted in 6 of 22 patients who were offered curative therapy.
PAC is a rare finding in the second round of screening after a 4-year interval, and a substantial proportion of PAC cases detected in the second screening round represent organ-confined disease. The findings of the current study suggest that the screening protocol used is sufficiently effective for detecting > 95% of malignancies before they develop features that would make them incurable. Cancer 2004;100:968–75. © 2004 American Cancer Society.
In the Western world, the observed incidence of prostate carcinoma is increasing, due to the general availability of serum tests for prostate-specific antigen (PSA) and the aging of the population. Early detection of prostate carcinoma using PSA testing may result in decreased prostate carcinoma mortality, but definite evidence of this effect has not yet been found. The European Randomized Study of Screening for Prostate Cancer (ERSPC) is an international, multicenter, population-based trial designed to investigate the impact of PSA testing on mortality.2 Participants are randomized into screening and control arms, and a final outcome with respect to endpoints such as prostate carcinoma mortality is not expected to emerge before 2008.
Analysis of intermediate parameters to assess the efficacy of the screening protocol currently used by the Rotterdam section of the ERSPC appears to be warranted. Previously, it was claimed that annual PSA testing in combination with digital rectal examination (DRE) would lead to the detection of all malignancies at a curable stage.3 In fact, the American Cancer Society recommends yearly PSA testing.4 Therefore, the 4-year interval between screening rounds in the Rotterdam section of the ERSPC may be considered excessively long. In addition, according to the ERSPC protocol, diagnosis of prostate carcinoma is established on systematic sextant biopsy. Systematic sextant biopsy currently is the subject of controversy, because it has been reported that approximately 30% of malignancies go undetected when this procedure is used.5 The earlier finding by the Rotterdam section of the ERSPC that the incidence of prostate carcinoma did not decrease significantly in the second round of screening (incidence, 3.9%) relative to the first round (incidence, 4.3%) may support the view that systematic sextant biopsy has insufficient sensitivity.6 More extensive biopsy procedures have been recommended, especially for large prostates and small tumors.7 Thus, it is possible that in a program involving a 4-year interval between screening rounds and systematic sextant biopsy, cases of prostate carcinoma may go undetected; these undetected cases could manifest themselves as clinically advanced prostate carcinoma during the 4-year interval, or they could be detected during the second round of screening.
To generate a simple parameter for describing the efficacy of screening for prostate carcinoma, several attempts were made to categorize the outcome of prostate needle biopsy in various subsets of patients.6–9 It was reported that among biopsy samples with Gleason score 7, those with a dominant poorly differentiated component (4 + 3) were associated with more adverse pathologic findings compared with samples with a smaller poorly differentiated component (3 + 4).8 Increased prostate carcinoma mortality among patients with Gleason score 7 and even greater mortality among those with Gleason score 8–10 were reported in comparison with patients with Gleason score 6.9 Other studies demonstrated a relation between extent of malignancy in the biopsy sample and pathologic stage in the prostatectomy specimen.10, 11 Thus, an arbitrary categorization model defining potentially advanced prostate carcinoma (PAC) on the basis of Gleason score and the amount of tumor present on sextant biopsy can be constructed.6 Using such an arbitrary model, we considered detection during the second screening round of a significant number of cases of PAC (defined by a predominance of Gleason pattern 4 and/or 5 or by > 30% malignant involvement in samples with Gleason score 7 [3 + 4]) on needle biopsy to be a possible indicator of the inadequacy of the current screening protocol.
In the current study, we analyzed the frequency of PAC as observed on sextant biopsy during the second round of screening for prostate carcinoma, and we compared these findings with the findings in the corresponding prostatectomy specimens and with clinical stage and follow-up data. Furthermore, the screening history of patients with PAC was analyzed, with the goal of determining why early detection of these cases of PAC was unsuccessful.
MATERIALS AND METHODS
In the Rotterdam section of the ERSPC, 42,376 participants ages 55–75 years were randomized into either a screening arm (n = 21,210) or a control arm (n = 21,166). In the first round, which took place between October 1991 and December 1998, 19,970 patients actually underwent screening, and 4243 of these patients subsequently underwent prostate biopsy. These figures include participants from our pilot study, in which the same screening protocol was used for patients with PAC.
Screening for prostate carcinoma involved PSA determination, DRE, and transrectal ultrasonography (TRUS). Systematic sextant needle biopsy of the prostate was recommended for participants who had either elevated PSA levels (≥ 4.0 ng/mL), abnormal findings on DRE, or abnormal findings on TRUS.
The protocol was simplified in May 1997, when sextant biopsy was recommended for patients with PSA levels ≥3.0 ng/mL. Abnormalities on DRE and/or on TRUS were no longer used as indicators for biopsy. The second screening round, in which this updated screening protocol was used, began in October 1995 and is ongoing at present. By March 2003, 11,210 of the expected 13,390 participants had been screened, and 2607 subsequently underwent prostate biopsy. Among patients with negative biopsy findings in the first screening round (n = 3151), 60% also underwent screening in the second round. Those who did not undergo screening in the second round declined biopsy (12.6%), were too old to undergo screening (16.0%), or died (4.0%); the remaining 7.4% of patients had unknown reasons for not undergoing screening. Among patients who were offered sextant biopsy in the second screening round, 12.6% did not undergo biopsy; 10.9% declined the offer, and 1.7% were not allowed to undergo biopsy because they were receiving anticoagulation therapy. The expected number of participants in the second round was equal to the total number of participants in the first round minus the number of 1) participants with prostate carcinoma detected in the first round; 2) patients age > 75 years; and 3) patients who died during the 4-year interval between screening rounds.
PSA levels were measured using the Hybritech Tandem E assay (Beckman-Coulter, San Diego, CA), and blood samples were stored to allow repeat measurements. All participants signed informed consent forms before entering the screening study.
During the second round of screening, two side studies were performed. The impact of PSA doubling on the detection of clinically relevant prostate carcinoma cases was assessed in 5658 patients; men in whom PSA levels increased twice or more between the first and second screening rounds were offered sextant biopsy. In addition, we investigated the incidence of prostate carcinoma in 885 men with PSA levels of 2–4 ng/mL; men in this side study also were offered sextant biopsy. Participants in these side studies were included in the main analysis as well.
Systematic sextant biopsies were obtained during longitudinal and cross-sectional ultrasonographic scanning of the prostate. A seventh biopsy was obtained if a hypoechogenic lesion was visible on TRUS. One pathologist (T.H.v.d.K.) reviewed all biopsy samples of malignant disease to avoid interobserver variation in Gleason grading. During the current review, the number of biopsies and the sizes of biopsy samples were recorded along with proportion of tumor involvement and proportion of each Gleason grade.
Slides of radical prostatectomy samples from participants with prostate carcinoma were retrieved from the pathology archives of Erasmus Medical Center (Rotterdam, The Netherlands) and other hospitals in the Rotterdam area. A single protocol for total embedding of prostate samples was used in all pathology laboratories to ensure accurate measurement of tumor volume, tumor grade, and tumor stage. After review, tumor stage and Gleason score were determined. Tumor volume was measured using morphometry, as is described elsewhere.12 For staging of radical prostatectomy specimens, the TNM classification for prostate carcinoma was used.13
Using the database of the Rotterdam section of the ERSPC, needle biopsy samples from second-round participants with Gleason score ≥ 4 + 3 and from those with Gleason score 3 + 4 and > 30% tumor involvement on sextant biopsy were identified. Data on clinical stage, treatment, and follow-up were retrieved from the ERSPC database.
Statistical analysis was performed using the SPSS software package (SPSS Inc., Chicago, IL). P < 0.05 was indicative of statistical significance. The chi-square test was used to analyze the number of advanced malignancies, and the paired-sample t test was used to compare PSA levels in the first and second screening rounds for men with PAC.
Biopsy Characteristics of PAC in the Second Screening Round
By March 2003, a total of 503 malignancies had been detected in the second screening round, including 30 (6.0%) that qualified as PAC (Table 1). During prevalence screening, 1092 cases of prostate carcinoma, including 214 cases of PAC (19.6%), were diagnosed in 19,970 participants. The decrease in the number of PAC cases detected in the second round was highly significant (P < 0.001). Of the 30 patients in whom PAC was detected, 11 were participants in the study of PSA doubling time, and 5 were participants in the side study of patients with PSA levels of 2–4 ng/mL. Two patients with PAC who had participated in the study of PSA doubling time had PSA levels < 3.0 ng/mL in the second screening round. These participants would not have been offered sextant biopsy in the screening round of the main study. Among patients with PAC who had participated in the study of individuals with PSA levels of 2–4 ng/mL, none had PSA levels < 3.0 ng/mL.
|Round 1||Round 2|
|PAC as a percentage of all malignancies||19.6||6.0|
In the second screening round, 373 participants had a Gleason score of 6 (3 + 3) on sextant biopsy. Six of these 373 participants (1.6%) had > 30% tumor involvement. In all 6 participants, the tumor was clinically confined to the prostate, and the median PSA value among these 6 patients was less than the corresponding value among patients with PAC (3.0 ng/mL vs. 4.7 ng/mL).
The distribution of Gleason scores among patients with PAC in the second screening round is shown in Table 2. There were 72 malignancies with a Gleason score of 3 + 4 in the second screening round; 8 of these 72 (11%) accounted for > 30% of the biopsy sample. Gleason scores of 7 (4 + 3) and 8 were noted in 4 and 15 patients with prostate carcinoma detected on biopsy, respectively. Five of 8 participants who had PAC with a Gleason score of 3 + 4 underwent an additional, seventh biopsy because of a hypoechogenic lesion detected on TRUS; all 5 biopsies were positive for malignant disease. If these hypoechogenic lesions had not been examined selectively on biopsy, four participants would not have been qualified for PAC.
|Gleason score on sextant biopsy||Cases of PAC in second screening round|
|3 + 4||8|
|4 + 3||4|
|4 + 4||12|
|3 + 5||2|
|4 + 5||1|
|5 + 3||1|
|5 + 4||1|
|5 + 5||1|
For convenience, the extent of malignant involvement in needle biopsy samples was classified according to the following arbitrary system: A, 1 biopsy core positive for malignancy; B, 2–3 biopsy cores positive for malignancy; and C, ≥ 4 biopsy cores positive for malignancy. The distribution of results according to this categorization system is shown in Table 3. For all but 1 participant with category A involvement, sextant biopsy samples consisted of < 10% tumor. When the 8 participants with PAC who had Gleason score 7 (3 + 4) and > 30% tumor involvement were excluded, 8 participants had category B involvement and 8 participants had category C involvement.
|Classification||No. of cases|
Development of PSA Values between the First and Second Screening Rounds in Patients with PAC
The PSA values of participants diagnosed with PAC in the second screening round increased significantly between the first and second rounds (P = 0.008; Table 4), with a median PSA doubling time of 5 years. One patient had a greater-than-10-fold increase in PSA levels. In Figure 1, for the 30 patients diagnosed with PAC in the second screening round, individual log PSA values from the first and second screening rounds can be compared. The median PSA velocity was 0.5 ng/mL per year (range, 0.0–12.8 ng/mL per year). For 9 men (30%), PSA velocity was greater than 0.75 ng/mL per year, suggesting rapidly progressive development of malignant disease.
|Screening round||Median PSA (range) for men with PAC on second-round screening|
History of Patients with PAC Detected on Biopsy in the Second Screening Round
In the second screening round, 792 men who had negative biopsy findings during prevalence screening underwent sextant biopsy. Forty-nine of these 792 men (6.2%) were diagnosed with prostate carcinoma in the second screening round, including 5 men with PAC. The initial biopsy findings in the five men with PAC detected in the second screening round were negative for prostate carcinoma and/or high-grade prostatic intraepithelial neoplasia (HG-PIN). Malignant disease was not found on review of these patients' biopsy cores from prevalence screening; in one instance, however, isolated HG-PIN was found in a single gland. Overall, a 10-fold increase in PSA level was detected in 2 participants in the second screening round; after repeating PSA measurements on the serum samples obtained in the first round, it was found that the serum sample from one of these patients was misidentified. This patient actually had a first-round serum PSA level of 6.6 ng/mL, rather than 0.8 ng/mL; thus, he was not offered biopsy in the first screening round.
One participant with category C PAC detected on sextant biopsy had abnormal DRE findings and a PSA level of 3.6 ng/mL in the first screening round. This participant was not present for biopsy in the first round.
Therapy and Follow-Up for Patients with PAC Detected during the Second Screening Round
Data on clinical TNM status and treatment for patients with PAC is summarized in Table 5. In 21 patients (70%), the malignancy was confined to the prostate. Fifteen patients received radiotherapy, with one of these patients receiving adjuvant hormone therapy. Twelve patients underwent radical prostatectomy. Patients who received radiotherapy had a higher clinical T status compared with patients who underwent radical prostatectomy. Two patients received hormone therapy; these patients had elevated PSA levels (42.0 and 59.0 ng/mL, respectively) and malignancy in every biopsy core (category C involvement), and one developed lymph node metastases. At a median follow-up of 29 months (range, 7–73 months), all patients with PAC were alive. PSA levels after therapy were known for 20 patients. PSA progression (i.e., an increase of > 0.2 ng/mL after 1 measurement) was observed in 7 patients. The patient who was managed with watchful waiting had an increase in PSA level from 5.4 ng/mL to 6.3 ng/mL over a follow-up period of 39 months.
|Clinical T status||Treatment||No. of cases|
|Radiotherapy||Prostatectomy||Endocrine therapy||Watchful waiting||Radiotherapy and endocrine therapy|
Radical Prostatectomy Findings in Patients with PAC
The pathologic features of prostatectomy specimens from 12 patients are summarized in Table 6. Downgrading from Gleason score 8 on biopsy to Gleason score 7 on radical prostatectomy was observed in three patients, and upgrading from Gleason score 7 (3 + 4) on biopsy to Gleason score 9 on prostatectomy was observed in two patients. Tumor volume ranged from 0.11–7.93 cm3, with a median value of 1.05 cm3. In the 7 prostatectomy specimens with matching Gleason score 7 (4 + 3) or Gleason score 8 (4 + 4) on sextant biopsy, most tumors were small (< 0.6 cm3); however, 2 tumors had larger volumes (1.46 and 2.07 cm3, respectively). The latter tumor consisted of a transition-zone carcinoma with Gleason score 4 (2 + 2; volume, 1.21 cm3) and a peripheral-zone carcinoma with Gleason score 8 (volume, 0.86 cm3). Among the 4 patients with Gleason score 7 (3 + 4) and > 30% tumor involvement on sextant biopsy, tumor volume ranged from 1.55 cm3 to 7.93 cm3.
|Prostate sample no.||Gleason score at biopsy||Category||Gleason score at RP||PSA velocity (ng/mL per yr)||Tumor volume (cm3)||Clinical T status||Pathologic T status|
|1||4 + 4||B||4 + 5||0.39||0.55||T2a||T2c|
|2||3 + 4||C||4 + 5||0.38||7.93||T1c||T3c|
|3||4 + 4||A||3 + 4||0.43||0.11||T2a||T2a|
|4||4 + 4||A||3 + 4||0.34||0.58||T1c||T2a|
|5||4 + 4||A||4 + 3||0.28||0.24||T3a||T2c|
|6||3 + 4||C||3 + 4||0.14||1.55||T1c||T2c|
|7||4 + 4||B||4 + 4||0.36||1.46||T1c||T2c|
|8||3 + 4||B||3 + 4||0.46||5.15||T1c||T2c|
|9||3 + 4||C||4 + 5||0.98||6.91||T2a||T2c|
|10||3 + 5||B||4 + 5||0.21||0.29||T2b||T2c|
|11||4 + 3||A||3 + 4||0.46||0.63||T2a||T2a|
|12||4 + 4||A||3 + 5||1.60||2.08||T2c||T2c|
Assessment of clinical T status in prostatectomy specimens led to underestimation of disease stage in eight patients and overestimation in only one patient. Second-round PSA values in men who underwent radical prostatectomy ranged from 2.4 ng/mL to 11.0 ng/mL; a significant correlation between PSA level and tumor volume was not found (P = 0.53). The median PSA velocity of patients with PAC treated with radical prostatectomy was 0.5 ng/mL per year; PSA velocity tended to be lower in these patients than in patients with PAC that were not treated with radical prostatectomy (median PSA velocity, 2.2 ng/mL per year; P = 0.06).
The current study found that PAC, defined by either Gleason score ≥ 4 + 3 or Gleason score 3 + 4 and > 30% tumor involvement on sextant biopsy, occurred infrequently in the second screening round in the Rotterdam section of the ERSPC. It was found that after the onset of the PSA era, the incidence of well differentiated malignancies (Gleason score 2–4) decreased14 and the incidence of moderately differentiated malignancies (Gleason score 5–7) increased, as more prostate malignancies were found on biopsy and not coincidently on transurethral resection of the prostate.14, 15 When early detection programs were introduced, a relative decrease in poorly differentiated malignancies (Gleason score 8–10) was noted in the comparison of prostatectomy specimens obtained before the PSA era with specimens from participants in the first screening round.16 Recently, the Rotterdam section of the ERSPC reported a significant reduction in the extent of tumor involvement in biopsy samples and in addition to downgrading in the second screening round.6 The observation of a significantly lower frequency of PAC in the second screening round (6.0%) compared with the first screening round (19.6%) in a considerably larger series confirmed and extended our previous observation. The long calculated lead time for prostate carcinoma (10.3 years in patients ages 55–74 years17) is consistent with the observed decrease in detection of PAC in the second screening round.
The criteria used to define PAC were based on data from the literature indicating that the amount of malignant disease in biopsy samples and the extent of high–Gleason grade disease (Gleason Grade 4–5) are important predictors of outcome.8, 11 Others have questioned the usefulness of biopsy pathology findings in predicting the outcome of organ-confined prostate carcinoma on an individual basis.10 According to a previous study conducted by the ERSPC, biopsy parameters may well be suitable for estimating the proportion of advanced malignancies in the context of a large, population-based study. Using a previously developed categorization model that was based on a limited number of prostate biopsy samples and matched prostatectomy specimens from a prevalence screening round, we demonstrated that approximately 60% of biopsy-defined ‘advanced’ malignancies exhibited extraprostatic extension.6 Thus, it is noteworthy that only 1 of 12 prostatectomy specimens from men with PAC diagnosed in the second screening round contained pT3C disease. This finding suggests that use in the second screening round of the same definition of PAC that was used in prevalence screening may have led to an overestimation of the number of advanced malignancies observed in the second round. Three of four patients with small-focus, Gleason score 8 disease in one biopsy sample (category A involvement) had a relatively low tumor volume on prostatectomy, in which downgrading to Gleason score 7 occurred. In contrast, two of three cases of extensive involvement on needle biopsy (category C involvement) of Gleason score 7 disease were associated with large-volume, Gleason score 9 prostate carcinoma on prostatectomy. Nonetheless, we cannot exclude the possibility that additional malignancies with adverse features may have been present among those biopsy-detected malignancies, which did not meet our arbitrary definition of PAC; in fact, based on previous data, it is likely that additional malignancies with adverse features were present.6 In the second round, when we included malignancies with Gleason score 6 (3 + 3) and ≥ 30% biopsy involvement, only 6 additional cases were identified. Radical prostatectomy was performed in two of these cases and revealed organ-confined disease with Gleason score 6. In summary, the prostatectomy data suggest that a substantial subset of malignancies that were detected in the second round and defined as PAC based on needle biopsy findings were confined to the prostate, although a relatively large proportion (6 of 12 cases) represented poorly differentiated disease (Gleason score 8–9).
Failure to detect prostate carcinoma during prevalence screening was attributable to an undetected abnormality on biopsy in one case and to an erroneous serum PSA determination in one case. In one other case, failure to detect PAC during prevalence screening was patient related, as the patient in question chose not to undergo biopsy. The other 27 cases of PAC that were detected in the second screening round had not been detected by regular prevalence screening. The low tumor volume (< 0.7 cm3) noted in 6 of the 12 PAC lesions treated with radical prostatectomy may have prevented these lesions from being detected 4 years earlier, in the first screening round; however, a selection bias toward malignancies with more favorable features, including small tumor volume, may have been present among men with second-round PAC who underwent prostatectomy compared with those who received radiotherapy.
The efficacy of the current screening protocol also could be measured in terms of the number of clinically manifested carcinomas that occurred during the 4-year interval between screening rounds. A recent analysis of data from the Rotterdam section of the ERSPC revealed that 25 interval carcinomas occurred between 1993 and 1996.19 More noteworthy was the finding that there were no advanced malignancies (Gleason score 4 + 3) among these interval carcinomas.
If rapidly progressive prostate malignancies with short doubling times had accounted for a substantial proportion of PAC cases, this finding would have been indicative of another potential limitation of the current screening protocol. We found that the median PSA velocity among patients with PAC was 0.5 ng/mL per year. In the literature, a PSA velocity of 0.75 ng/mL per year was considered to be predictive of the presence of clinical prostate carcinoma when PSA values were between 4.0 and 10.0 ng/mL.18 Only 30% of patients with PAC detected in the second screening round had a PSA velocity > 0.75 ng/mL. This finding indicates that PSA velocity is not a good predictor of the presence of PAC in this subset of patients. Nonetheless, we cannot exclude the possibility that a number of potentially highly aggressive prostate malignancies exhibiting low PSA levels are hidden in the screened population. It is unclear whether a protocol involving more frequent PSA testing would be beneficial in terms of early detection. The incidence of PAC reported in the current study—approximately 2.0 cases per 1000 participants in the second screening round—appears to be reassuring with regard to the efficacy of our screening protocol.
In conclusion, a 4-year interval between screening rounds in a screening protocol involving PSA testing and sextant biopsy for definite diagnosis of prostate carcinoma appears to be warranted, because 1) the mean lead time for prostate carcinoma is estimated to be 6–12 years, depending on patient age17; 2) the frequency of clinically manifested interval malignancies that have been reported to date is negligible19; and 3) a decreased incidence of potentially advanced malignancies was noted in the second screening round. Furthermore, the performance of the current screening procedure appears to be favorable, with only 2 of 30 cases of PAC detected in the second screening round being attributable to the failure of diagnostic tests during prevalence screening.
- 4American Cancer Society. Available from URL: http://www.cancer.org/docroot/CRI/content/CRI_2_2_3X_How_is_prostate_cancer_found_36.asp? Site area=CRI [accessed July 1, 2003].
- 7A model for the number of cores per prostate biopsy based on patient age and prostate gland volume. J Urol. 1998; 59: 920–924., , , .