Elevated serum prostate specific antigen levels in conjunction with an initial prostatic biopsy negative for carcinoma: who should undergo a repeat biopsy?


Mr D. Greene Department of Urology, Sunderland Royal Hospital, Kayll Road, Sunderland SR4 7TP, UK.



To determine the outcome of repeated prostatic biopsies in men attending with suspected prostate cancer but an initial negative biopsy.

Patients and methods

Patients who had undergone two or more transrectal ultrasonography (TRUS)-guided prostate biopsies were identified from the Hospital Information Support System database. Indications for TRUS were a raised prostate-specific antigen (PSA) level (>4.0 ng/mL), with or without an abnormal digital rectal examination (DRE). Sextant prostate biopsies plus biopsies of any suspicious hypoechoic area or area of DRE abnormality were obtained for histology. Forty-eight patients underwent repeat TRUS-guided prostatic biopsies (mean age 67.5, sd 7.25, range 53–82 years).


The mean (sd, median, range) PSA level was 16.9 (13.5, 11.6, 5.2–61.8) ng/mL. Fifteen patients (31%) had carcinoma on repeat biopsy, 11 after the second and four after a third biopsy. The positive repeat biopsy rate was 24% where the PSA level was 4.0–9.9 ng/mL, 33% if the level was 10.0–19.9 ng/mL and 39% if it was ≥20.0 ng/mL. There was no significant difference in age or initial PSA concentration between those men with positive and those with negative repeat biopsies. However, patients with cancer had significantly higher PSA levels before repeat biopsy than at first biopsy (P=0.0043) and had greater PSA velocities than had patients with no diagnosis of cancer (P=0.0067).


Where sufficient clinical suspicion exists, despite an initial negative biopsy, repeat TRUS-guided prostate biopsies should be carried out to exclude carcinoma of the prostate.


Awareness of prostate cancer has increased among patients and GPs because of the increase in prostate cancer-related publications in the medical literature and general press in recent years. While the natural history of prostate cancer is not yet fully known, the availability of PSA testing, combined with a greater awareness of prostate cancer, has led to a significant increase in the referral of patients with an elevated PSA for evaluation. Despite a recommendation against screening for the UK population [1] it is likely that more patients with LUTS will have their PSA level measured, because of concerns by the patient and/or GP about underlying cancer. This will lead to more patients being diagnosed with stage T1c prostate cancer. This factor, combined with the increased incidence of prostate cancer, will provide a considerable challenge for urologists, although the scale of the problem in the UK is unlikely to parallel that in the USA where, as a result of screening, 316 000 new cases of prostate cancer were predicted in 1996 [2].

While almost 60% of patients with a serum PSA value of >10 ng/mL have prostate cancer [3], many other conditions, e.g. BPH, prostatitis, prostatic infarct and prostatic intraepithelial neoplasia (PIN), cause an elevation of serum PSA. In contrast, only a quarter of patients with a serum PSA level of 4–10 ng/mL are found to have cancer [4] and many patients will undergo unnecessary prostate biopsies as a result. In an attempt to refine the sensitivity of PSA as a predictor of prostate cancer, several PSA derivatives have been studied, including PSA density, PSA velocity and free to total PSA ratio. Despite these advances, considerable difficulties may still arise in the detection of prostate cancer, especially in the PSA range of 4–10 ng/mL [5].

When evaluating a patient with suspected prostate cancer, the accepted diagnostic investigations include a DRE, the estimation of serum PSA, and TRUS of the prostate [4,6]. Candidates for such management must stand to benefit significantly from the diagnosis and subsequent treatment of their prostate cancer, whether in terms of improved quality of life, reduced morbidity or prolonged survival. However, many of these patients have no discrete hypoechoic areas visible on TRUS. In such cases, the standard procedure has been to take sextant biopsies, sampling the base, mid-prostate and apex on each side [6]. Unfortunately, despite an elevated PSA level, abnormal DRE or both, in a significant number of cases no cancer is detected on TRUS-guided prostate biopsy [7[8][9]–10]. As the volume of prostate sampled by the standard sextant biopsy technique is relatively small, the possibility remains that these individuals may harbour prostate cancer, despite an initial negative biopsy. As yet, no parameters have been agreed that would prevent a patient in this situation undergoing an unnecessary repeat biopsy. The urologist must weigh carefully the potential morbidity and associated anxiety of repeat prostatic biopsy against the benefit to the patient of a confirmed diagnosis. This issue has been addressed by several studies from the USA [7[8][9]–10], but has not been widely discussed in urological literature from the UK. Thus, the aim of the present study was to investigate the outcome of repeat prostatic biopsies in men referred with a clinical suspicion of prostate cancer and persistent elevation of PSA, but an initial negative prostate biopsy, and to evaluate which clinical variables are most helpful in predicting cancer on repeat biopsy in this group.

Patients and methods

Between May 1995 and November 1997, 1059 patients attended for TRUS at our institution. We reviewed the Hospital Information Support System database and identified 48 men (mean age 67.5 years, sd 7.25, range 53–82) with suspected carcinoma of the prostate who had undergone two or more TRUS-guided prostatic biopsies during this period. The indications for TRUS in this subgroup of patients were an elevated PSA (>4.0 ng/mL) with or without an abnormal DRE. A Bruel and Kjaer system 3535 diagnostic ultrasound unit with a 7 MHz biplanar transrectal probe (model 8551) was used in all examinations. Using the biplanar probe, the prostate was scanned in both transverse and sagittal planes. Sextant biopsies were routinely obtained and a further two biopsies taken from any area of DRE abnormality or any suspicious hypoechoic area. Biopsies were obtained using an automatic biopsy gun (Biopty® , Bard, Covington, USA) and 18 G biopsy needles. All TRUS examinations were carried out by the same operator (D.R.G.), either personally or in direct supervision of a Higher Urological Trainee.

All biopsy material was submitted to the Pathology Department at Sunderland Royal Hospital and graded by Consultant histopathologists. All patients were given a metronidazole (1 g) suppository after biopsy and were treated for 3 days with amoxicillin/clavulinic acid, or ciprofloxacin if allergic to penicillin. PSA was measured in all cases by the Department of Chemical Pathology, Sunderland Royal Hospital, using the Hybritech Tandem-E assay (Hybritech Inc, San Diego, CA, USA). Each patient’s medical record was carefully reviewed and the patient’s age, prebiopsy PSA values, PSA velocity, histopathological diagnosis of each biopsy, DRE findings and outcome of each TRUS examination entered into a database and analysed statistically. Data are presented as mean (sd) and distributions compared using Student’s t-test, with significance assumed at P<0.05.


Of the 48 men who underwent repeat TRUS-guided prostatic biopsies, where their initial biopsy failed to reveal carcinoma of the prostate, 12 underwent three and two underwent five biopsies. Overall, 15 (31%) patients had carcinoma on repeat biopsy, 11 after a second and four after a third biopsy. Histologically, 12 of 15 tumours detected were Gleason grade 3+3, while the three remaining tumours were Gleason grades 3+2, 4+3 and 4+5, respectively. The positive biopsy rate was 23% (11/48) at the second and four of 14 at the third biopsy. All patients had an abnormal serum PSA value (>4.0 ng/mL). Their mean PSA was 16.9 (13.5) (median 11.6, range 5.2–61.8) ng/mL at the time of their initial biopsy. Fourteen of the 48 men (29%) had an abnormal DRE, 15 of 48 (31%) had an abnormal TRUS examination and in eight of 48 (17%) both the DRE and TRUS were abnormal. Based on the outcome of their repeat biopsies, seven of the 15 men with a positive (cancer detected) repeat biopsy had an abnormal DRE, 10 had an abnormal TRUS and seven had both, whereas of the 33 with a negative (cancer not detected) repeat biopsy, seven (21%) had an abnormal DRE, five (15%) had an abnormal TRUS and only one had both abnormalities.

The histological diagnoses at the first biopsy fell into five main categories, as shown in Table 1; BPH was by far the commonest diagnosis on first biopsy. In five of the original 1059 cases (0.5%), the initial biopsies were reported as ‘inadequate for diagnosis’, necessitating repeat biopsies. The outcome of the second set of biopsies in relation to the initial biopsies are also shown in Table 1. There was no significant difference in patient age between those with a negative or positive repeat biopsy, with mean (sd) age of 69.6 (8.6) and 67.3 (7.1) years, respectively (P=0.3311, two-tailed t-test).

Table 1.  Outcome of first biopsy, and of the second biopsy in relation to histology Thumbnail image of

The probability of a positive repeat biopsy was 24% where the serum PSA level was 4.1–9.9 ng/mL, 33% where it was 10.0–19.9 ng/mL and 39% where it was ≥20.0 ng/mL. There was no significant difference between the PSA level at the time of first biopsy in those with a negative or a positive repeat biopsy, with mean levels of 15.7 (13.3) and 18.9 (15.1) ng/mL, respectively (P=0.505, two-tailed t-test). However, those men with cancer on repeat biopsy had significantly higher mean (sd, range) PSA values before repeat biopsy than at the time of their initial biopsy, at 24.7 (18.3, 5.9–61.6) ng/mL and 18.9 (15.1, 4.9–50.4) ng/mL, respectively (P=0.0043, two-tailed t-test). Also, those patients with a positive repeat biopsy had significantly higher mean PSA velocities than those with a negative repeat biopsy, at 7.4 (8.9) ng/mL per year and 1.0 (6.7) ng/mL per year (P=0.0067, two-tailed t-test, Fig. 1).

Figure 1.

Comparative PSA velocity (ng/mL per year) for those with positive (green) and negative (red) repeat biopsies. P=0.0067.

In the 25 men with BPH on first biopsy, nine (36%) had cancer on repeat biopsy. There was no difference between the positive and negative repeat biopsy groups in initial PSA values, at 16.8 (14.4) and 17 (12.9) ng/mL, respectively (P=0.486). Of those patients initially in the ‘inadequate for diagnosis’ category, two of five had carcinoma on the second and one had carcinoma on the third biopsy. Two patients had inadequate second biopsies and of these, one had carcinoma on his next biopsy. Overall, four of seven patients with inadequate biopsies were subsequently found to harbour cancer. Five patients had their second biopsy reported as ‘suspicious (but not diagnostic) of malignancy’, and of those, two had cancer on their third biopsy. Interestingly, two of eight patients with low-grade PIN (LGPIN) were found to have cancer on later biopsies, but no patient with high-grade PIN (HGPIN) was subsequently found to have cancer.


At present, no set criteria exist to guide the urologist in the management of patients with persistently elevated serum PSA levels (with or without an abnormal DRE), where the initial set of prostate biopsies have been negative. In the present series, 31% of patients were found to harbour cancer on repeat biopsy. This value compares with the series by Fleshner et al. [7], where 30% of patients had positive repeat biopsies, but is higher than the 19–23% quoted by other studies [8[9]–10]. All authors ultimately recommend repeat prostatic biopsy where there is a high level of clinical suspicion, despite an initial negative biopsy, given the significant initial false-negative biopsy rate. Fleshner et al. defined a subgroup of patients deemed to be at lowest risk for cancer (PSA<10 ng/mL, PSA density<0.15, PSA velocity<0.75 ng/mL per year, no PIN and a negative DRE, TRUS and no family history) and found that five of 21 (24%) patients in this group had carcinoma on repeat biopsy [7]. Using these parameters, a subset of patients cannot, as yet, be identified for whom repeat biopsy is not required.

All the present patients underwent systematic para-sagittal TRUS-guided sextant biopsies, as well as directed biopsies of any DRE abnormality or hypoechoic lesion. In view of the high false-negative biopsy rate, this may not be the optimal approach. Eskew et al. [11] described a systematic five-biopsy technique and recommended 12–15 biopsies where the prostate volume was <50 mL, and up to 18 biopsies where it was ≥50 mL; this method increased their cancer detection rate by 35%. Similarly, Levine et al. found that by performing two consecutive sets of TRUS-guided sextant biopsies at a single visit, an additional 30% of cancers were detected in the second set of biopsies [12]. In both cases, the procedures were well tolerated, with acceptable complication rates. Using a stochastic computer simulation of ultrasound-guided biopsies, using mathematically reconstructed radical prostatectomy specimens, Chen et al. [13] found that sextant biopsies only detected cancer in 73% of cases where the total tumour volume was>0.5 mL. Mapping the tumour foci not detected by the sextant method revealed that these foci were in areas not biopsied by the sextant technique (transition zone, mid-peripheral zone and anterior horn of peripheral zone). They found that a 10-core biopsy scheme incorporating these additional areas detected 96% of cancers. On this evidence, taking more biopsies and from more areas of the prostate may reduce the number of repeat biopsies required in the future.

In the present series, routine transition zone biopsies were not taken; there would appear to be some justification for this approach in patients with an initial negative biopsy, although this issue remains controversial. Lui et al. [14] found that nine of 17 patients with a positive repeat biopsy had cancer confined to the transition zone. This finding conflicts with the results of several other studies that question the usefulness of routine transition zone biopsy. Epstein et al. [15] found that the transition zone biopsy alone was positive in only 2% of cases, while Bazinet et al. [16] concluded that the low additional yield of transition zone biopsies precludes their routine use in the detection of prostate cancer. Keetch et al. [17] took transition zone biopsies in addition to repeat peripheral zone biopsies in men who had one or two sets of previous negative biopsies and found that transition zone biopsy detected few additional tumours in this group of patients. They suggested that cancers located in the transition zone do not usually cause persistently elevated PSA levels in the setting of negative prostate biopsies. A large randomized prospective trial should resolve this issue.

Of interest in the present series was the lack of association between HGPIN and carcinoma on repeat biopsy. Four of 48 (8%) patients had HGPIN initially, but none were subsequently found to have cancer. However, two of eight patients with LGPIN had cancer on repeat biopsy. These findings are similar to those of Roehrborn et al. [8], who found that no patient with PIN yielded cancer on second biopsy, but contradict Ellis and Brawer [9], who found that all five patients in their series with grade II/III PIN (but none of nine with grade I PIN) had cancer on repeat biopsy. In any event, the present patients with HGPIN are being followed closely in our clinic and will be re-biopsied if there is any further significant increase in their serum PSA levels.

In an attempt to reduce the number of unnecessary repeat biopsies, several investigators have studied various PSA derivatives as a means of refining PSA sensitivity. PSA density was not calculated in the present series, nor was free PSA measured, hence their potential impact on the outcome of repeat biopsies cannot be assessed. There was a significant correlation between PSA velocity and a positive repeat biopsy (P=0.0067), but as this was calculated by subtracting the PSA level at repeat biopsy from that at initial biopsy, and adjusting the result to the rate of change over 12 months, rather than recording serial measurements of PSA over 18 months to calculate velocity, this conclusion may be unreliable. Keetch et al. [18] found that although the combination of PSA density (threshold>0.15) and PSA velocity (>0.75 ng/mL per year) could be useful in reducing the number of unnecessary biopsies, they were not sufficiently sensitive for routine use in cancer detection.

Free PSA has been evaluated recently as a predictor of prostate cancer in several studies. In a multicentre trial, Thiel et al. [19] found that a free PSA value of <7% was highly suspicious of prostate cancer, whereas a level of >25% suggested benign disease. In patients with persistently elevated PSA levels, a normal DRE and two prior negative biopsies, Morgan et al. [20] found that a free PSA value of <10% was 91% sensitive and 86% specific in predicting the presence of cancer, and could help distinguish between BPH and malignancy in such patients. Similarly, Catalona et al. [21] found that free PSA thresholds of 28% and 30% detected 90% and 95% of cancers, respectively, in men with PSA values of 4.1–10.0 ng/mL, a normal DRE and a prior negative biopsy. These studies show that measurement of free PSA provides predictive information about the presence of prostate cancer in men with initial negative biopsies and helps differentiate between benign and malignant disease. It may prove to be a useful tool in reducing the number of unnecessary repeat biopsies in such patients in the future.

In conclusion, the false-negative initial biopsy rate of 31% in this selected series is cause for concern and in agreement with other authors, as current clinical practice stands, we recommend that all patients with persistent elevation of PSA and prior negative biopsies should undergo repeat biopsy, to exclude carcinoma of the prostate. In view of concerns raised about cancer detection by sextant biopsy it would seem appropriate to recommend a more extensive biopsy regimen for repeat TRUS-guided biopsies. There are several variables that warrant further evaluation and that may prove useful in reducing the need for repeat biopsies; of these, free PSA shows greatest promise. However, until large randomized clinical trials are conducted to answer these questions, clinicians have no choice but to continue to re-biopsy all patients where a high clinical suspicion of cancer exists, despite a negative initial biopsy, particularly in the case of rising PSA velocity.