For men without a prostate cancer diagnosis or symptoms that might indicate prostate cancer what is the incremental value of performing a digital rectal examination (DRE) in addition to PSA testing in detecting any prostate cancer? (PICO question 6)
Background
DRE, in combination with measurement of serum prostatic acid phosphatase, was the standard method for establishing the clinical suspicion of prostate cancer prior to the introduction of PSA testing and systematic biopsy of the prostate. However, men were often reluctant to have a DRE and remain so today. Other problems were that a significant volume of cancer needed to be present before a DRE abnormality could be identified, and that there was significant observer variation. Therefore, in an era when PSA testing is increasingly offered to men concerned about the possibility of prostate cancer, with the aim of identifying much smaller foci of cancer, it is important to ask the question: Does DRE still have an important role in the detection of asymptomatic prostate cancer?
Evidence
Five studies30-34 were identified that examined the benefits and harms of using DRE in addition to total PSA levels as initial tests to identify men likely to have prostate cancer. All the studies were assessed to have a moderate risk of bias. The search strategy, inclusion and exclusion criteria, and quality assessment are described in detail in the Technical report.
The most important data were provided by the Prostate Cancer Prevention Trial,34 a randomised controlled trial comparing finasteride with placebo, in which men underwent screening for 7 years. This was the largest relevant screening study identified, and the only one in which men were biopsied regardless of DRE result or PSA level (i.e. screen negatives as well as screen positives were biopsied). Therefore, this study was able to provide reliable estimates of differences in sensitivity and
specificity, as well as estimates of increases in cancers detected and unnecessary biopsies. The study was generally well conducted, with potential verification bias investigated and shown not to be an issue.22 However, the risk of bias was considered to be moderate because the authors did not state whether DRE, PSA tests and pathologist review of biopsy specimens were performed blind. To avoid potential bias due to any possible effects of finasteride, only data from the placebo arm were
examined in this review.
The use of DRE in addition to PSA thresholds resulted in a moderate increase in the detection of prostate cancer.34 However, the incremental gain in cancer detection was at the cost of biopsy referrals for men without prostate cancer (false positives), the rate of which increased with decreasing PSA threshold. The rate of false positives was 1.91 for every additional cancer using a PSA threshold of 4.0 ng/mL, 1.99 for every additional cancer using a threshold of 3.0 ng/mL, and 2.44 for every
additional cancer using a threshold of 2.0 ng/mL. At a threshold of 3.0 ng/mL, adding DRE resulted in a relative increase in sensitivity of 12 percentage pointsaccompanied by a specificity decline of 7 percentage points. In absolute terms, this would mean that for every 1000 men repeatedly screened, 26 more cancers would be found, but 52 more false positives would be referred for biopsy. At a PSA threshold of 4.0 ng/mL, there was a 14 percentage-point increase in sensitivity and a 7 percentage- pointdecline in specificity. In absolute terms, 30 more cancers would be detected but 58 men would undergo unnecessary biopsies per 1000 men screened. Importantly, the same increase in cancer detection rate could have been achieved without DRE, but instead by using a lower PSA threshold (Figure 2.2).
Figure 2.2 Trade-off between detecting true positives and adding false positives for PSA alone and in combination with DRE
Rates of true and false positive results for PSA only (blue line) and PSA + DRE (orange line) Source: data derived from Thompson et al (2007)34
The other four studies30-33 examined the addition of DRE to a PSA threshold of 4.0 ng/mL. The results of these studies were roughly in agreement as to the direction and magnitude of accuracy of the incremental gain. The number of false positives for every additional cancer detected was even higher in these studies, despite the use of more extensive biopsies in one study,32 and the fact that DRE was performed by urologists or urologic residents in three of these studies.30, 32, 33 However, differences in populations, the degree of screening, and verification prevent pooling of the data and limit direct comparison.
Four studies30, 31, 33, 34 reported the effects of adding DRE to a screening protocol with PSA threshold of 4.0 ng/mL on cancer yield stratified by Gleason Score:
Data from the placebo arm of the Prostate Cancer Prevention Trial34 show that, for every 1000 men screened, adding DRE to a screening protocol with PSA threshold of 4.0 ng/mL would detect 3 additional cancers with Gleason Score > 7 and 7 additional cancer with Gleason Score > 6. The proportion of higher-grade cancers amongst the additional cancers detected with DRE was lower than, or similar to, that detected using PSA alone: 23.2% cancers with Gleason Score > 6 and 9.0% cancers with Gleason Score > 7, compared with rates detected by using PSA alone: 35.2% cancers with Gleason Score > 6 and 10.1% cancers with Gleason Score > 7.
A study conducted among US veterans31 reported that 34.0% of the additional cancers detected by DRE were Gleason Score > 6 and 13.6% were Gleason Score > 7.
In a large US community screening study,30 3.3% of additional cancers detected by DRE were Gleason Score > 7.
In a small Mexican screening study33 the single additional cancer detected by DRE had a Gleason Score of 7.
However, based on the data from the Prostate Cancer Prevention Trial,34 the addition of DRE to PSA increased sensitivity for cancers with Gleason Score > 7 by 25.4 percentage points, while specificity was reduced by 8.6 percentage points. For cancers with Gleason Score > 6, the addition of DRE to PSA gained a 15.0 percentage-point increase in sensitivity at the cost of a 8.5 percentage-point reduction in specificity.
The findings of the Prostate Cancer Prevention Trial34 may not be generalisable to the Australian primary care setting because the trial cohort was men over 55 years old who had undergone previous screening (initial normal DRE and PSA < 3 ng/mL on entry to the study). In comparison, an PSA testing in Australia covers a broader range of men. In addition, the trial investigators may have benefited from specific training and have had greater experience in performing DRE, compared with clinicians who perform DRE in Australian primary care. Therefore, the benefits of adding DRE to PSA testing in Australia may be fewer than those reported.
Evidence summary and recommendations
Evidence Summary Level References
There is evidence from one large moderate-quality study that the addition of DRE to PSA testing provided an incremental gain in prostate cancers detected, but at a cost of two or more extra false positives per cancer detected. The study also showed that similar gains could be made by lowering the PSA threshold. DRE accuracy is likely to be lower outside the trial setting of this study.
III-2 30-34
The sensitivity for detecting high-grade cancers was increased when DRE was added to PSA testing. However, the gain in
detecting higher-grade cancers by adding DRE was generally not greater than that for lower-grade cancers.
III-2 30, 31, 33, 34
Evidence-based recommendation
In asymptomatic men interested in undergoing testing for early diagnosis of prostate cancer, digital rectal examination is not recommended as a routine test in the primary care setting. Grade C
Practice point:
Although DRE is not recommended as a routine test for men who, after
advice, wish to be t ested for the presence of prost ate cancer, it still has an important role in assessing the prostate prior to biopsy in a specialist setting.
Health system implications of these recommendations Clinical practice
Current guidelines for preventive care in general practice35 recommend both DRE and PSA for men who choose to undergo prostate cancer screening after being fully informed of the risks, benefits and uncertainties. Therefore, implementation of this recommendation would alter current practice. Resourcing
Implementation of this recommendation would have no significant resource implications. It may slightly reduce the consultation time for men attending primary care.
Barriers to implementation
No barriers to the implementation of this recommendation are foreseen.