Current issues of ACP Journal Club are published in Annals of Internal Medicine


Serum prostate specific antigen was better than serum prostate acid phosphatase for diagnosing prostate cancer

ACP J Club. 1993 May-June;118:88. doi:10.7326/ACPJC-1993-118-3-088

Source Citation

Cooke RR, Nacey JN, Beeston RE, Delahunt B. The efficacy of serum prostate specific antigen as a tumour marker in prostatic carcinoma: a comparison with serum acid phosphatase. N Z Med J. 1992 Sep 9;105:345-7.



To compare the value of serum prostate specific antigen (PSA) with serum prostate acid phosphatase (ACP) in the diagnosis of prostatic cancer.


Blinded comparison of serum PSA and serum ACP results with histologic assessment of prostatic samples.


Urology department of the Wellington Hospital, New Zealand.


349 men having either prostatic biopsy or prostatectomy between May 1990 and April 1991.

Description of tests and diagnostic standard

Serum samples were taken before digital rectal examination. Serum PSA levels were determined using a 2-site immunoradiometric assay (Tandem-R PSA, Hybritech, San Diego, California). Serum ACP was measured using thymolphthalein monophosphate as substrate. The diagnostic standard was routine histologic assessment of tissue obtained from either prostatic biopsy or prostatectomy. Tissue analysis was done by a genitourinary pathologist who was blinded to serum test results.

Main outcome measures

Sensitivity and specificity of the tests.

Main results

97 patients had histologically proven carcinoma, and 227 had benign prostatic hyperplasia. The remainder of the samples were from patients previously seen and 1 patient with prostatitis. The test properties for PSA and ACP at various cut-off points are summarized in the Table. The area under the receiver-operator curve for PSA was 0.81 and for ACP was 0.72 {statistical significance was not determined}*.


Serum prostate specific antigen measurement was superior to serum acid phosphatase measurement in the diagnosis of prostatic cancer. Serum prostate specific antigen levels between 1 ng/mL and 23 ng/mL did not allow for adequate discrimination between prostatic cancer and benign prostatic hyperplasia and may require further testing.

Source of funding: Not stated.

For article reprint: Mr. R. Cooke, Radiochemistry Laboratory, Wellington Hospital, Private Bag, Wellington South, New Zealand. FAX 64-4-385-5856.

*Information supplied by authors.

Table. Test properties of prostate specific antigen (PSA) and prostate acid phosphatase (ACP) for diagnosing prostate cancer

Cut-off point Sensitivity Specificity +LR† -LR†
PSA 1.0 93% 26% 1.27 0.27
PSA 4.0 73% 73% 2.70 0.37
PSA 10.0 61% 93% 8.71 0.42
PSA 23.0 48% 98% 24.0 0.53
ACP 0.2 95% 10% 1.06 0.5
ACP 0.4 79% 44% 1.41 0.48
ACP 0.8 48% 90% 4.8 0.58
ACP 1.5 30% 98% 15.0 0.71

†+LR = likelihood ratio for the presence of disease if the test is positive; -LR = likelihood ratio if the test is negative. Both calculated from data in article.


Cooke and colleagues set out to determine if serum PSA is more useful in discriminating various forms of prostatic disease than serum ACP. Among 349 patients who subsequently had prostatic biopsy or surgery, PSA did slightly better than ACP. Improvement in accuracy, however, is accompanied by a higher cost (U.S. $45.35 vs U.S. $21.65 at our hospital).

These findings raise a number of questions, the most important of which is whether either test is suitable for screening of the general population. Using reported sensitivity and specificity figures and the prevalence of prostate cancer in the general United States population, a PSA level of ≥ 23 ng/mL yields a positive predictive value of 1.7%. In other words, for every 1000 men with a PSA level ≥ 23 ng/mL, only 17 will subsequently be found to have prostate cancer. The sensitivity for detecting cancer at this high threshold is poor (48%) but could be improved by lowering the level, which would trigger further work-up. This, however, would also increase the number of patients subjected to unnecessary, expensive, and potentially invasive investigations. Thus, neither test, in my opinion, is suitable for screening.

Both tests have a higher predictive value when used to evaluate men with evidence of prostatic disease. What is, however, the clinical meaning of a given serum level? To rely on the test result, one needs high sensitivity (to rule out the disease at a low serum level) or high specificity (to rule in cancer at a high level). For PSA, 90% sensitivity is crossed only when its level is < 1 ng/mL, and 98% specificity is reached at the level > 23 ng/mL. Thus, very high and very low levels of PSA are useful. Any other result will change the post-test probability of prostatic cancer, but not enough to change the course of action.

The remaining question is how reliably changes in the PSA or ACP level can be used to follow people with treated prostatic cancer. Further information is needed.

Joseph L. Lyon, MD
University of UtahSalt Lake City, Utah, USA