SEARCH

SEARCH BY CITATION

Keywords:

  • prostate cancer;
  • active surveillance;
  • TRUS-guided biopsies

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONFLICT OF INTEREST
  8. REFERENCES

Study Type – Diagnostic (exploratory cohort)

Level of Evidence 2b

What’s known on the subject? and What does the study add?

The accuracy of TRUS-guided biopsies in identifying the location of an index cancer at surgery is low. When two biopsies show cancer in the same location, the concordance between the biopsy and surgical location of the cancer increases.

OBJECTIVE

• To evaluate the ability of repeat prostate biopsies to determine the location of the index cancer for men on prostate cancer surveillance.

PATIENTS AND METHODS

• Forty-five men on active surveillance had a record of the locations of their positive diagnostic biopsy, repeat surveillance biopsy and index cancer (i.e. largest cancer) from prostatectomy specimens.

• Logistic regression analysis was used to evaluate the association between two consecutive needle biopsies showing cancer in an identical location and the outcome of finding the index cancer at the same location as the initial diagnostic biopsy.

RESULTS

• Eighteen of 45 (40%) men ultimately had an index cancer at the same location as their diagnostic biopsy.

• Thirteen men had two consecutive biopsies that showed cancer at the same location each time; nine of these men ultimately had an index cancer at that same location.

• In multivariable logistic regression analysis of men with at least two biopsies, having two initial consecutive biopsies with the same location increased the odds (odds ratio 5.9; 95% CI 1.1–31, P= 0.037) of having an index cancer at the same location as the initial biopsy in a cohort of men on active surveillance.

CONCLUSIONS

• A substantial proportion of men in an active surveillance cohort who undergo prostatectomy ultimately have evidence of an index cancer at the same location as their initial biopsy.

• This is more likely to be the case when a repeat biopsy shows evidence of cancer at the same location.


Abbreviations
TRUS

transrectal ultrasound

RP

radical prostatectomy.

INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONFLICT OF INTEREST
  8. REFERENCES

Prostate cancer management depends on an accurate assessment of disease extent to predict patient outcome and direct patients to appropriate treatment modalities. Such an assessment is especially important when considering focal therapy or other treatments that attempt to spare non-cancerous tissue. Transrectal ultrasound (TRUS)-guided needle biopsy is currently the standard of care for the diagnosis of prostate cancer. However, TRUS-guided biopsy can be inaccurate and does not always reflect the final pathology on radical prostatectomy (RP) specimens [1,2]. Many studies have been undertaken in an attempt to increase its accuracy, including increasing the number of needle cores taken at each biopsy or using clinical characteristics at the time of diagnosis. These methods rely on clinicopathological variables at one point in time.

In our institutional active surveillance programme, very low-risk disease is monitored with annual surveillance biopsies. This offers us the unique opportunity to study the accuracy of a series of TRUS-guided biopsies taken over several time-points. Our objective was to determine if a repeat biopsy after the initial diagnostic biopsy can help predict the location of an index cancer in RP specimens of men on prostate cancer surveillance.

PATIENTS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONFLICT OF INTEREST
  8. REFERENCES

Since December 1994, older men with very low-risk prostate cancer presenting to our institution for consultation have been advised that active surveillance is a reasonable management option [3]. This prospective, longitudinal programme has been approved by the Johns Hopkins institutional review board, and all men provided informed consent before entering the programme. Although the programme is primarily recommended to men over the age of 65 years, some younger men enrol because of personal preference or comorbidities. The inclusion criteria for the programme are based on the study by Epstein et al. [4] and include clinical stage T1c prostate cancer with a PSA density <0.15 ng/mL/cm3, in addition to at least a 12-core biopsy (re-reviewed at our institution) involving two or fewer cores showing Gleason score 6 or less prostate cancer, with no Gleason pattern of 4 or 5 and ≤50% cancerous involvement of any individual core. A few men who did not meet these strict criteria chose active surveillance because of comorbidity or personal preference and are included in our prospective database.

Men were followed with semi-annual serum PSA measurements and digital rectal examination, and by annual extended 12–14 core surveillance biopsy. A recommendation for curative intervention was triggered if the biopsy inclusion criteria were no longer met at a surveillance biopsy: Gleason score 7 or greater (any Gleason pattern 4 or 5), or more than two cores positive for cancer, or greater than 50% of a single core with cancer.

From January 1995 to February 2010, 786 men enrolled in active surveillance. A total of 251 men underwent treatment, including 116 who had a radical prostatectomy (RP). Seventy of the 116 men who underwent RP for treatment of their prostate cancer had data available regarding both the laterality of their positive needle core on initial diagnostic biopsy and the laterality of the index cancer (i.e. largest lesion) on RP specimen. Five of these men had bilateral disease at the time of biopsy. The remaining 65 – including 63 who had at least one additional surveillance biopsy after the diagnostic biopsy – constituted our study population for assessing the accuracy of TRUS-guided biopsy in predicting unilateral disease. There were 46 men who had data regarding the sextant location of their positive needle core on initial diagnostic biopsy and of the index cancer on RP specimen. These 46 men – including 45 who had at least one additional surveillance biopsy after the diagnostic biopsy – constituted our study population for assessing the accuracy of TRUS-guided biopsy in predicting the location of the index cancer.

We used the concordance between location of prostate cancer on diagnostic biopsy and location of index lesion on prostatectomy specimen to assess the accuracy of TRUS-guided biopsies in men on active surveillance who undergo prostatectomy. When determining the concordance of laterality between initial needle biopsy and index cancer at surgery, a biopsy was considered to be concordant if a unilateral, positive biopsy resulted in a unilateral index cancer that involved the same side. In terms of the area of the prostate involved, an initial needle biopsy and index cancer at surgery were considered concordant if the index cancer involved both the same area and side as the biopsy, including index cancers that were bilateral.

We also considered the possibility that multiple biopsies showing cancer in the same location could indicate a high-grade or large-volume tumour. Therefore, we performed additional analyses using evidence of Gleason 7 or greater disease in the index lesion or extraprostatic extension on the prostatectomy specimen as outcomes of interest.

Pearson’s chi-squared test and logistic regression analysis were used to test the null hypotheses of no association between concordance of first and second biopsies and our outcomes of interest on the RP specimen. In analysing concordance for laterality, we combined the group of men with no evidence of cancer on their second biopsy with the group of men with concordant biopsies, because a similar proportion went on to have an index lesion on prostatectomy that was concordant with their first biopsy. On multivariable analyses, we controlled for clinical (age, PSA, PSA density) and pathological characteristics (number of positive cores and maximum percentage of core involvement). The data for pathological characteristics were taken from the first biopsy when analysing concordance for laterality because of the collinearity between those pathological variables and having no evidence of cancer at the time of the second biopsy; pathological data from the second biopsy were used when analysing concordance for area. Statistical analysis was performed using stata version 10.0 software. Significance was considered at P < 0.05.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONFLICT OF INTEREST
  8. REFERENCES

Of the 65 men who had unilateral disease on initial needle biopsy, 34 (52%) had a unilateral index cancer that involved the same side, 15 (23%) had a unilateral index cancer that involved the opposite side, and the remaining 16 (25%) had bilateral disease.

Table 1 shows the demographic and clinical characteristics of men who had at least one surveillance biopsy after their diagnostic biopsy. When men with a unilateral diagnostic needle biopsy had a repeat biopsy that was positive on the same side as their diagnostic biopsy, the chance that they would have a concordant index cancer at the time of RP went up to 57%. Of the 12 men who had a repeat biopsy that was negative for cancer, eight went on to have a unilateral index cancer on the same side as their diagnostic biopsy (Table 2).

Table 1.  Demographic and clinical characteristics of men on active surveillance who underwent radical prostatectomy; median (interquartile range)
Clinical and pathological characteristicsMen with available data for laterality on ≥2 biopsies (n= 63)Men with available data for location on ≥2 biopsies (n= 45)Men with concordant biopsies (n= 13)Men without concordant biopsies (n= 32)P value*
  • *

    Results of Student’s t test comparing men with and without concordant biopsies on diagnostic biopsy and surveillance biopsy.

Age, years63.0 (59.0–66.4)62.9 (59.0–66.9)63.6 (60.0–65.2)62.8 (58.6–67.2)0.853
PSA at diagnosis, ng/mL3.9 (1.9–6.1)4.0 (2.0–6.0)4.3 (3.6–6.1)3.7 (1.9–6.0)0.457
PSA density at diagnosis, ng/mL/cm30.09 (0.06–0.14)0.09 (0.06–0.14)0.13 (0.09–0.17)0.07 (0.05–0.14)0.121
Maximum % core involvement with cancer at diagnostic biopsy1.5 (1–10)5 (1–10)5 (1–10)5 (1–10)0.889
Maximum % core involvement with cancer at surveillance biopsy1 (1–30)1 (1–20)10 (1–40)1 (0–20)0.231
Number of cores positive at surveillance biopsy1 (1–3)1 (1–2)2 (1–3)1 (0–2)0.001
Table 2.  Relative laterality of prostate biopsy and index cancer at prostatectomy*
Relative laterality of diagnostic and surveillance biopsyRelative laterality of diagnostic biopsy and index cancerTotal
Concordant, n (%)Contralateral side, n (%)Bilateral index cancer, n (%)
  • *

    Sixty-three men who had at least one surveillance biopsy after their initial diagnostic biopsy and then underwent prostatectomy (chi-squared test for association, P= 0.202).

  • †Biopsies are considered discordant if the index lesion involved the contralateral side of the prostate relative to the diagnostic biopsy, but did not involve both sides.

  • ‡An index cancer was considered bilateral if it involved both sides of the prostate.

Concordant13 (57)4 (17)6 (26)23
Discordant3 (27)6 (55)2 (18) 11
Both sides involved on surveillance biopsy8 (47)3 (18)6 (35)17
No cancer on surveillance biopsy8 (66)2 (17)2 (17)12

On univariate logistic regression analysis, concordance between the diagnostic and surveillance biopsies or having no cancer on the surveillance biopsy was associated with a 2.3-fold increase in the odds of having a unilateral index cancer at the time of RP that was on the ipsilateral side as the diagnostic biopsy, although this result was not statistically significant (95% CI 0.8–6.4, P= 0.105, Table 3). When controlling for other clinical and pathological characteristics on multivariable logistic regression analysis, there was a similarly sized association (OR 3.0, 95% CI 0.9–9.8, P= 0.070). Interestingly, increasing age at diagnosis and maximum per cent core involvement with cancer at diagnostic biopsy were the only clinical variables that were found to be significantly associated with concordance between the laterality of the diagnostic biopsy and the index cancer on RP specimen. The association with age was significant even when controlling for other clinical and pathological variables.

Table 3.  Results of logistic regression analysis for concordance of laterality between diagnostic biopsy and index cancer at prostatectomy*
VariableUnivariate odds ratio (95% CI)P valueAdjusted odds ratio (95% CI)P value
  • *

    Sixty-three subjects with unilateral disease at the time of diagnostic biopsy and at least one follow-up biopsy.

  • †Thirty-two subjects with and 31 subjects without concordance between their diagnostic biopsy and index cancer on prostatectomy specimen had data available for all variables, excluding maximum per cent core involvement.

  • ‡Thirty-one subjects with and 29 subjects without concordance between their diagnostic biopsy and index cancer on prostatectomy specimen had data available for all variables, including maximum percent core involvement.

Age at diagnosis (years)1.1 (1.0–1.3)0.0281.2 (1.0–1.3)0.046
PSA at diagnosis (ng/mL)1.0 (0.9–1.1)0.9790.9 (0.7–1.3)0.712
PSA density at diagnosis (ng/mL/cm3)1.0 (0.9–1.1)0.6911.1 (0.9–1.3)0.497
Maximum % core involvement with cancer at diagnostic biopsy1.1 (1.0–1.3)0.0341.1 (1.0–1.2)0.069
Number of cores positive at diagnostic biopsy1.2 (0.6–2.5)0.6300.3 (0.1–1.9)0.178
Concordance of diagnostic and surveillance biopsy or no cancer on surveillance biopsy2.3 (0.8–6.4)0.1053.0 (0.9–9.8)0.070

The 46 men who had data for the area (apex, mid, base, lateral peripheral zone) of their cancer on both diagnostic biopsy and at the time of surgery accounted for a total of 55 positive needle cores, which resulted in a concordance rate with the index cancer of 40% (Table 4). When an initial surveillance biopsy showed evidence of cancer in a specific location that was concordant with the diagnostic biopsy, there was a 69% chance that the index cancer would also be concordant. On the other hand, if the diagnostic and surveillance biopsies did not show cancer in the same area, there was only a 28% chance that the index cancer was concordant with the initial diagnostic biopsy (chi-squared test, P= 0.011) (Table 5). There was no statistically significant association between concordance of diagnostic and surveillance biopsies and extraprostatic extension or Gleason score >7 disease on prostatectomy specimen (chi-squared test, P > 0.05) (Tables 6,7).

Table 4.  Concordance of location of positive prostate biopsy and index cancer at prostatectomy*
Needle biopsy locationRadical prostatectomy (index cancer)
Same location/total% concordanceSame location and side/total% concordance
  • *

    Forty-six men who underwent prostatectomy, accounting for a total of 55 positive biopsies.

  • †Proportion of patients in whom positive needle core on diagnostic biopsy and index cancer on radical prostatectomy involved the same location of the prostate, regardless of side.

  • ‡Proportion of patients in whom positive needle core on diagnostic biopsy and index cancer on radical prostatectomy involved the same side and the same location of the prostate.

  • §

    §Lateral biopsies were not further stratified into apex, mid, or base.

Apex  11/157310/1567
Mid9/15604/1527
Lateral§4/8503/838
Base7/17415/1729
Total31/555622/5540
Table 5.  Concordance of diagnostic biopsy and index cancer at prostatectomy for men with and without concordance between diagnostic biopsy and a repeat surveillance biopsy*
Concordance of diagnostic and surveillance biopsyConcordance of diagnostic biopsy and surgeryTotal
Yes, n (%)No, n (%)
  • *

    Chi-squared test of association, P= 0.011.

Yes9 (69) 4 (31)13
No9 (28)23 (72)32
Table 6.  Extraprostatic extension at prostatectomy for men with and without concordance between diagnostic biopsy and a repeat surveillance biopsy*
Concordance of diagnostic and surveillance biopsyExtraprostatic extensionTotal
Yes, n (%)No, n (%)
  • *

    Chi-squared test of association, P= 0.497.

Yes5 (38) 8 (62)13
No9 (28)23 (72)32
Table 7.  Gleason score at prostatectomy for men with and without concordance between diagnostic biopsy and a repeat surveillance biopsy*
Concordance of diagnostic and surveillance biopsyGleason scoreTotal
6 or less, n (%)7 or more, n (%)
  • *

    Chi-squared test of association, P= 0.419.

Yes 7 (54) 6 (46)13
No13 (41)19 (59)32

Logistic regression analysis corroborated the association between concordance between diagnostic and surveillance biopsies and concordance between diagnostic biopsy and index cancer at prostatectomy (Table 8). Having concordant biopsies was associated with a 5.9-fold greater odds (95% CI 1.1–30.9, P= 0.037) that the diagnostic biopsy would be concordant with the index cancer, when controlling for variables at the time of diagnosis (age, PSA, PSA density) and at the time of second biopsy (maximum per cent core involvement with cancer, and number of positive cores).

Table 8.  Results of logistic regression analysis for concordance of location between diagnostic biopsy and index cancer at prostatectomy*
VariableUnivariate odds ratio (95% CI)P valueAdjusted odds ratio (95% CI)P value
  • *

    Forty-five subjects with at least one surveillance biopsy after the diagnostic biopsy (18 with concordance between their diagnostic biopsy and index cancer on prostatectomy specimen and 27 without).

Age at diagnosis1.0 (0.9–1.2)0.5351.1 (0.9–1.2)0.466
PSA at diagnosis1.0 (0.9–1.2)0.6050.9 (0.7–1.3)0.674
PSA density at diagnosis1.0 (0.9–1.1)0.4291.0 (0.9–1.2)0.709
Maximum % core involvement with cancer at surveillance biopsy1.0 (1.0–1.0)0.8871.0 (0.9–1.0)0.680
Number of cores positive at surveillance biopsy1.4 (0.8–2.3)0.2561.0 (0.5–2.1)0.962
Concordance of diagnostic and surveillance biopsy5.8 (1.4–23.5)0.0155.9 (1.1–30.9)0.037

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONFLICT OF INTEREST
  8. REFERENCES

In our cohort of men with very low-risk prostate cancer that was managed with active surveillance, a surveillance biopsy that was concordant with the diagnostic biopsy was associated with a greater likelihood that a patient would have a unilateral index lesion at RP and an index cancer in the same location as the diagnostic biopsy, when compared with those with discordance between the diagnostic and surveillance biopsies.

Prostate cancer cannot reliably be visualized on TRUS. As a result, TRUS-guided needle biopsy tends to be a random sampling of the peripheral zone of the prostate. Whereas some pathological characteristics, such as per cent core involvement with cancer at a positive core, can be used to estimate tumour volume at that particular location, there is still inherent variability in the process. Gleason score can also be a predictor of the potential aggressiveness of a tumour; if there are two tumour foci that are both Gleason 6, however, as is sometimes the case for patients on active surveillance protocols, one cannot be sure which one, if any, has metastatic potential. This amount of uncertainty can be tolerated if a patient plans to undergo whole gland treatment of his prostate. If a patient plans to undergo focal therapy and preserve non-cancerous tissue, however, there is need for greater certainty about the exact location of cells with metastatic potential. This study suggests one method for increasing our ability to identify the location of the index cancer.

In the setting of widespread PSA screening, as well as over-detection and over-treatment of localized prostate cancer, focal therapy has become popular as a means of ablating only areas of the prostate that contain disease, while avoiding the adverse effects of radical treatment [5]. The rationale underlying focal therapy rests on the notion that a significant proportion of prostate cancer is unifocal [5], or in cases of multifocal disease, that only a single focus has the propensity to metastasize or result in lethal disease [6].

Hemiablation is one method of focal therapy for unilateral prostate cancer that has recently gained interest [7], especially with the use of cryotherapy. In previous series of men with unilateral prostate cancer on diagnostic biopsy, somewhere between 26 and 40% of patients are estimated to have unilateral disease at the time of RP [1,2,8–10]. This number is generally thought to be higher in patient populations that only include men with low-risk disease. In addition, these previous studies considered disease in addition to the index cancer in determining bilateral disease. When we considered the laterality of the index cancer only, we found that a unilateral diagnostic biopsy indicated a unilateral index cancer on the same side in over 50% of cases.

Attempts at finding clinical or pathological characteristics that are predictive of unilateral disease have been mostly unsuccessful. Number of positive cores, per cent core involvement, prostate volume and PSA have all been shown not to improve the prediction of which patients would have unilateral disease [9,11]. Increasing prostate volume was significantly associated with ipsilateral disease in patients with unilateral disease on biopsy, but not when patients with bilateral disease were included in the analysis [12]. In studies that have included both patients with unilateral and bilateral diagnostic biopsies, negative family history and unilateral diagnostic biopsies were significant predictors of unilateral disease [13]. When comparing the accuracy of sextant and extended biopsy schemes, the accuracy of the biopsy in correctly predicting the laterality of disease increased from 49% on sextant biopsy to 59% on extended biopsy [14]. In our study, patients who had unilateral disease on their diagnostic biopsy were more likely to have a unilateral index cancer on the same side if they had a concordant repeat biopsy or a repeat biopsy that did not show cancer, although this result did not reach statistical significance. Increasing age was the only significant predictor of concordance between the laterality of the diagnostic biopsy and the index cancer.

In addition to evaluating tumours based on laterality, one previous study has further sub-stratified based on specific location within the prostate. The RP specimens of the 281 patients in that study were divided into right vs left and apex vs mid-base. Of the clinicopathological variables that were analysed, only the absence of multifocality on prostatectomy was associated with agreement between biopsy location and location of disease on prostatectomy [10]. Our study further stratifies the location of the mid-base region of the prostate. We found that a second biopsy showing evidence of cancer in the same location as found in the first biopsy was a significant predictor of concordance between the location of the diagnostic biopsy and the index cancer at RP. However, similar to previous studies, we found that the accuracy of TRUS-guided biopsies in finding the correct location of the index cancer on RP was low.

Given the possibility that consecutive concordant biopsies may be indicative of a more aggressive tumour, we also evaluated whether or not extraprostatic extension or increased Gleason scores were more common in patients who had concordant biopsies. Although we did not find a significant association, this part of the analysis may not have been sufficiently powered.

This study has several limitations. First, these results are based on a select group of men with very low-risk disease at the time of biopsy, thereby limiting the generalizability of this study. Second, we looked specifically at the laterality and location of the index cancer at the time of prostatectomy, without regard for secondary or tertiary cancers. We did this with the idea that the index cancer is thought to be the most clinically significant focus of tumour in the prostate [6,15].

In conclusion, men with very low-risk prostate cancer on diagnostic biopsy who wish to know the true extent of their disease may benefit from the information obtained in a repeat surveillance biopsy. In our study of men who underwent prostatectomy after enrolling in an active surveillance protocol, those with concordant biopsies at the time of diagnosis and at the first surveillance biopsy were more likely to have an index cancer that was also concordant with the diagnostic biopsy. Given the data showing the limited accuracy of TRUS-guided biopsies, improved imaging modalities may be needed before patients can be guaranteed that their unifocal disease is being definitively treated. However, if patients enrolled in active surveillance programmes are still contemplating focal therapy for treatment of their cancer, they may benefit from the additional information that a repeat biopsy provides.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONFLICT OF INTEREST
  8. REFERENCES