Accuracy of post-radiotherapy biopsy before salvage radical prostatectomy


  • Supported by the Sidney Kimmel Center for Prostate and Urologic Cancers.

Correspondence: James A. Eastham, Urology Service, Department of Surgery, Memorial Sloan-Kettering Cancer Center, 353 East 68th Street, New York, NY 10021, USA.




  • To determine whether post-radiotherapy (RT) biopsy (PRB) adequately predicts the presence, location, and histological features of cancer in the salvage radical prostatectomy (SRP) specimen. Before salvage treatment, a PRB is required to confirm the presence of locally recurrent or persistent cancer and to determine the extent and location of the prostate cancer.

Patients and Methods

  • SRP was performed between 1998 and 2011 on 198 patients.
  • All patients underwent a PRB. PRB and SRP specimens were evaluated by a genitourinary pathologist. Patients had external-beam RT alone (EBRT; 71%) or brachytherapy with or without EBRT (29%).


  • Of the men undergoing SRP, 26 (14%) were clinical stage ≥T3, with 13% of PRBs with Gleason score ≥8.
  • Cancer was unilateral in 120 (61%) biopsies, with contralateral or bilateral prostate cancer at SRP in 49%. In the SRP specimen, cancer was multifocal in 57%.
  • Cancer was upgraded at SRP in 58% of men, with 20% having an increase in primary Gleason grade.
  • The accuracy of PRB varied by region from 62% to 76%, with undetected cancers ranging from 12% to 26% and most likely to occur at the mid-gland.


  • Radiation-recurrent prostate cancers were often multifocal, and biopsy missed up to 20% of tumours.
  • More than half of the cancers were upgraded at SRP, and many that were unilateral on PRB were bilateral at SRP.

biochemical recurrence


(external beam) radiotherapy


(salvage) radical prostatectomy


More than 240 000 men in the USA will be diagnosed with prostate cancer in 2012 [1], and half of these men will choose radiotherapy (RT) for their treatment [2]. Depending on risk status, 20–40% of men treated with RT for clinically localised prostate cancer will have biochemical recurrence (BCR) ≤ 10 years after treatment [3]. Without intervention after BCR, the rate of clinically detectable recurrence at 5 years is 75%, with half of these patients having distant metastasis [4]. Yet, >70% of recurrences that occur early after RT are due to persistent or radiation-resistant local disease [5]. Thus, timely salvage intervention after RT may be offered to patients with >10 years life expectancy, localised recurrence, and prostate cancer detected at post-RT biopsy (PRB).

The ability of the PRB to accurately characterise the extent and location of residual cancer is critical to treatment planning. If the PRB identifies a small focus of prostate cancer on one side of the prostate, some men will consider focal ablative techniques, whereas young patients with multifocal or high-grade disease may be best treated with whole-gland ablation or salvage radical prostatectomy (SRP) [6]. To date, no study has determined the accuracy of the PRB to predict histological findings in the SRP specimen.

Patients and Methods

A prospectively accumulated database of men who developed BCR after RT for presumed clinically localisedprostate cancer and sought surgical therapy at Memorial Sloan-Kettering Cancer Center was collected with Institutional Review Board approval. All men had a PRB with review of the biopsy and SRP specimen by a genitourinary pathologist. Our cohort (220 patients) consisted of patients who had had SRP after receiving RT. Patients who underwent SRP before 1997, who did not have PRB and consistent pathological evaluation of SRP specimens, were excluded (22 patients), leaving a final cohort of 198 patients. Most patients underwent biopsy at our institution and all patients had a prostate biopsy with ≥12 cores with assistance from a radiological technologist for consistency.

To determine the accuracy of the PRB, we compared the anatomical location and grade of the PRB with the corresponding regions of the prostate at SRP. These anatomical regions (apex, mid, and base) of both the right and left sides were compared with the corresponding regions of the SRP specimen. In all, 121 patients had detailed anatomical biopsies of the apex, mid, and base of the gland and were the primary cohort for the study. For the remaining 77, only laterality was recorded. In many cases, SRP specimens were sectioned en bloc (Fig. 1). Accuracy was defined as a positive or negative PRB result in a given region of the prostate confirmed in the SRP specimen. The primary and secondary Gleason grade of the PRB was compared with final primary and secondary Gleason grade of the SRP specimen. In addition, tumour multifocality was recorded in the SRP specimen.

Figure 1.

Tumour map of SRP specimen. Prostates are sectioned en bloc from apex (A) to base (G). Black lines indicate high-grade prostate cancer, Gleason score ≥ 7; blue lines indicate extraprostatic extension; green lines indicate Gleason score of 6. BN, bladder neck margin; L, left; LA, left apical margin; LN, lymph node; LSV, left seminal vesicle; R, right; RA, right apical margin; RSV, right seminal vesicle. Each bar of scale represents 1 mm, total of 25 mm.


The characteristics of the 198 men who underwent PRB and SRP are listed in Table 1. The median (interquartile range) PSA level was 3.7 (1.7, 6.5) ng/mL. The most common form of RT was external-beam RT (EBRT; 132/187 patients; 71%), with 55 patients (55/187; 29%) undergoing brachytherapy or both EBRT and brachytherapy. The most common clinical stage was T2 (89/186; 48%), whereas it was ≥T3a in 26 (26/186; 14%). Almost half of PRBs identified Gleason score 7 cancer (92/192; 48%), with 24 (24/192; 13%) Gleason score ≥8. Most SRPs were performed open (175/198; 88%). Surgical margins were positive in 14% (27/198), despite advanced pathological stage in 56% (109/195) of SRP specimens (≥pT3), including 26% (51/195) pT3b and 6% (12/195) pT4. Gleason score ≥8 was present in 35% (58/164) of SRP specimens.

Table 1. Patient characteristics (N = 198)
CharacteristicN (%)
Clinical stage (n = 186): 
T171 (38)
T289 (48)
T3a15 (8)
≥T3b11 (6)
Clinical Gleason score (n = 192): 
≤676 (40)
792 (48)
≥824 (13)
Pathological stage (n = 195): 
pT02 (1)
pT284 (43)
pT3a45 (23)
pT3b51 (26)
pT3c1 (1)
pT412 (6)
Pathological Gleason score (n = 164): 
Unable to grade34 (17)
≤65 (3)
7101 (62)
≥858 (35)
Type of RT (n = 187): 
Brachytherapy55 (29)
EBRT132 (71)
Type of SRP: 
Open175 (88)
Laparoscopic15 (8)
Robot-assisted8 (4)
Positive surgical margins27 (14)
Seminal vesicle invasion61 (31)
Multifocal prostate cancer (n = 193)110 (57)

Table 2 describes the anatomical location of the positive PRBs. Overall, positive PRBs ranged from 38% at the right base to 46% of the right apex.

Table 2. Positive PRB results
Biopsy locationAll Patients (N = 151), n (%)
Right apex69 (46)
Right mid-gland64 (42)
Right base57 (38)
Left apex62 (41)
Left mid-gland64 (42)
Left base62 (41)

Each SRP specimen was evaluated for the location and grade of prostate cancer (Fig. 1; Table 3). Most (110/193 [57%]) of the cancers were multifocal. In multifocal cancers, the most common cancer location was the right mid-gland (55/88 [63%]) and right apex (53/88 [60%]), whereas non-multifocal tumours were most commonly located in the right apex (28/61 [46%]). The base was the least likely prostate cancer location.

Table 3. Positive SRP pathological results
Pathological locationAll patients (N = 150), n (%)Unifocal (n = 61), n (%)Multifocal (n = 88), n (%)
Right apex82 (55)28 (46)53 (60)
Right mid-gland79 (53)23 (38)55 (63)
Right base49 (33)10 (16)38 (43)
Left apex69 (46)21 (34)47 (53)
Left mid-gland73 (49)22 (36)50 (57)
Left base49 (33)15 (25)33 (38)

Comparison of PRB with SRP Pathology

Based on final pathology, prostate cancers were categorised as multifocal (92) or unifocal (55). Tumours that were multifocal had higher frequency of positive PRBs among all regions of the prostate. In patients with multifocal disease, the left mid-gland was the most common source of positive PRBs (46/92 [50%]), whereas the right apex was the most common location of positive PRBs in unifocal tumours (23/55 [42%]).

Table 4 shows the agreement rates between PRB and SRP results. Agreement rates ranged from 62% to 76% between different regions of the prostate. Undetected cancer rates from PRBs ranged from 12% to 26%, with the right mid-gland having the greatest frequency of missed cancers (Table 5). The positive predictive value of PRB was 69% with a negative predictive value of 70%. There was a high rate of upgraded tumour grades from PRB to SRP specimen, with 58% (92/158) of patients upgraded from PRB, including 20% primary Gleason grade (31/158), 20% (32/158) secondary Gleason grade, and 18% (29/158) where both the primary and the secondary Gleason grades were increased. Only 3% (4/158) of PRBs were downgraded (4/158). In patients with unilaterally localised positive biopsies, final pathology was unilateral in only half (Table 6).

Table 4. PRB results compared with SRP with agreement rates
LocationPRBSRP, nAgreement
+Overall (n = 121), n (%)Multifocal (n = 75), n (%)Unifocal (n = 45), n (%)
apex+47792 (76)54 (72)37 (82)
mid-gland+361476 (62)45 (60)30 (67)
base+281790 (74)57 (76)32 (71)
apex+341780 (66)47 (63)32 (71)
mid-gland+421482 (68)55 (73)30 (67)
base+252879 (65)46 (61)32 (71)
Table 5. Locations of undetected prostate cancers by PRB (negative prostate biopsy results with cancer present at SRP)
LocationRight, n (%)Left, n (%)
Apex22 (18)24 (20)
Mid-gland31 (27)21 (17)
Base14 (12)14 (12)
Table 6. Results of unilateral positive PRB
 Right unilateral positive (n = 64), n (%)Left unilateral positive (n = 56), n (%)
Ipsilateral positive35 (55)26 (46)
Contralateral positive1 (1)4 (7)
Bilateral positive28 (44)26 (46)


The goal of the present study was to describe the accuracy of PRB to identify prostate cancer in a given location of the prostate before SRP. In a study comparing 250 non-irradiated RP specimens with prostate biopsies, the greatest accuracy was at the base (64%), followed by the mid-gland (53%), with the apex having the lowest accuracy (43%) [2]. We showed greater rates of accuracy for all regions of the prostate than previously reported for non-irradiated prostates. This may be because of the higher frequency of high-grade multifocal disease in patients after RT, or perhaps as a result of better sampling of the typically small post-radiation gland. In another study comparing 281 prostate biopsies with RP specimens, the only predictor that increased the accuracy of the prostate biopsy was the absence of multifocality [7]. In that study, multifocal tumours were present in 77% of patients, whereas tumours were multifocal in only 57% in the present study. The accuracy of PRB in the present study was slightly greater in four of six anatomical regions from tumours that were non-multifocal compared with multifocal tumours.

The cohort described in the present series is the largest single-institution series of SRPs. Pathological analysis of these previously irradiated prostates may further our knowledge of potentially radiation-resistant regions of the prostate. In this cohort, the most common area to harbour prostate cancer was the right apex in which more than half of prostates (55%) had persistent prostate cancer. This suggests that the apex may be particularly difficult to target with RT either because of shielding by the pelvis or because it receives less radiation to prevent damage to the rhabdosphincter. The least common location for tumours was the base (33% for both sides). A prior study of 47 SRPs similarly found the base to be the least common location of radiation recurrent prostate cancer [8].

In patients who have had prior RT, the role of PRB remains controversial. Currently, the American Society for Therapeutic Radiology and Oncology does not recommend routine PRB, but rather recommends following PSA levels as a marker for recurrent disease [9]. In an unbiased cohort (in regard to PSA), the rate of positive PRB was at 50% at 13 months but decreased to 30% at 30 months [10]. The management of patients with a positive biopsy result but normal PSA levels/kinetics remains controversial. The average time from RT to treatment of prostate cancer relapse is 5 years if BCR is dependent on the use of PSA by the American Society for Therapeutic Radiology and Oncology criteria [11]. Thus, those selecting SRP may benefit from earlier intervention to reduce the rates of advanced disease [12].

The rate of upgrading at SRP was high at 58%, with 20% increasing at least one primary Gleason grade compared with the non-irradiated prostate, with a reported range of 25% to 57% [13, 14]. All samples, both PRBs and SRPs, were reviewed by a dedicated genitourinary pathologist at a high-volume institution. Previous investigations of upgrading suggest that the highest rate of upgrading occurs when the biopsy and prostatectomy are performed at separate institutions and, therefore, graded by different pathologists [15]. An important variable that is poorly understood is the effect of RT on the Gleason grading system of prostate biopsy. As the largest series of SRP specimens, the high rate of upgrading confirms the difficulty in making a Gleason grade on PRBs compared with final RP specimens. Two factors may have affected the rate of upgrading. First, there was a high rate of Gleason score ≥ 8 (35%) at SRP. Second, almost 80% of prostates had residual radiation treatment effects.

The rate of downgrading from prostate biopsy to prostatectomy is often lower than the rate of upgrading, but may be highly variable with a range between 5% and 25% [15]. In the present study, only 3% of biopsies were downgraded at SRP, suggesting a far greater likelihood that larger sampling of the tumour would show high-grade disease. This may also be explained by the difficulty in discerning low-grade tumours from normal prostate glands with altered architecture from RT.

The management of patients with recurrent prostate cancer after RT remains a difficult situation. Because most recurrences are local, salvage treatment directed at the prostate gland should be considered by the patient and physician because uncontrolled local disease has a high rate of metastatic spread [16]. One possibility for salvage therapy could be focal therapy directed at biopsy-predicted focal regions of the prostate. In the present series, unilateral biopsy predicted the location of the tumour in about half of cases. Thus, half-gland-directed therapy would result in missed tumours half of the time. For younger patients, the advantage of whole-gland removal by SRP remains a viable option of staging and cancer control [6].

Several limitations of the present study deserve mention. First, the location of the tumours was categorised by trained genitourinary pathologist, but interobserver variation in categorising the tumour location may vary by pathologist. Second, the grade of pathology may also be subject to interpreter variability. The PRBs recorded are standard 12 core template biopsies and more thorough mapping-guided biopsies, potentially through perineal approach, with or without MRI assistance may detect higher rates of cancer. Finally, the patients who were able to undergo SRP bias the study because they may have less aggressive disease. Men with more advanced disease may not be candidates for surgery and could alter the accuracy of PRB.

In conclusion, the accuracy of PRBs ranges from 62% to 74%, with the rate of previously undetected tumours ranging from 12% to 26% at SRP. Multifocal tumours occur in more than half of patients, and unilateral biopsy results are inaccurate in half of prostates after RT. Almost 60% of tumours are upgraded at SRP. Although PRB is useful for detection of recurrent prostate cancer after RT, far more accurate information is obtained at SRP. Thus, patients who elect salvage focal therapy may have higher rates of disease recurrence compared with those who choose whole-gland treatment or removal.

Conflict of Interest

None declared.