Click CC to turn on closed captioning.
Published: October 2009Print Record of Viewing
Dr. Saenger provides an overview of the methods available for evaluating prostate-specific antigen (PSA) levels and discusses the clinical implications of standardizing PSA methods.
Presenter: Amy K. Saenger, PhD
- Division of Clinical Core Laboratory Services at Mayo Clinic
Questions and Feedback
Welcome to Mayo Medical Laboratories Video Hot Topics. These presentations provide short discussion of current topics and may be helpful to you in your practice.
Our presenter for this program is Amy K. Saenger, PhD, Director of the Central Clinical Laboratory in the Department of Laboratory Medicine & Pathology at Mayo Clinic in Rochester, Minnesota. Dr. Saenger will provide an overview of the methods available for evaluating prostate-specific antigen levels and discuss the clinical implications of standardizing PSA methods.
Today I will be discussing issues related to prostate specific antigen testing and touch on the controversial topic of screening for prostate cancer. This is a subject that has received much attention in both the clinical and laboratory realm due to several large clinical trials recently published and the fact that there are major implications for measuring an accurate and precise PSA.
Elevated PSA Result on Screening
Over the past two decades the incidence of prostate cancer has increased dramatically. Prostate cancer is the most common cancer diagnosed in North American males, excluding skin cancers. It is the second leading cause of cancer death in males, second only to lung cancer.
Use of serum prostate specific antigen (or PSA) for prostate cancer screening is largely responsible for the increases, and epidemiological data suggests it is not only a higher frequency of cancer diagnosis but mortality has decreased because the cancer is being diagnosed at an earlier stage, allowing for earlier treatment. An elevated PSA result obtained upon screening can signify several things. An increased PSA could signal prostate cancer, with adenocarcinoma the most frequent type.
Other types of prostate cancer include small cell undifferentiated carcinoma, which is a rare, life-threatening form; transitional cell carcinomas, sarcomas of stromal origin, or can originate as metastases from other organs. It is important to recognize that the differential diagnosis of the type of prostate cancer cannot be made with PSA alone. An elevated PSA can also be due to benign prostatic hyperplasia, or BPH, which is thought of as more of a quality of life disease. BPH is relatively common, with an estimated incidence of approximately one quarter of males over the age of 40 and up to 80% of men 80 years or older.
There are other reasons a PSA is elevated and include prostatitis, a condition which lacks clear diagnostic criteria, or there could be analytical issues. The challenge for physicians is to distinguish between these various conditions and determine the cause of an elevated PSA. In addition, the decision to perform a prostate biopsy is often made based upon the PSA screening result.
Reasons for Ordering PSA
One study closely examined the reasons physicians were ordering PSA tests. As clearly shown in the graph, a majority of orders were for prostate cancer screening, followed by monitoring cancer reoccurrence, following suspected cancers, monitoring patient’s cancer progression, and less often used in managing BPH. Although this study was published nearly one decade ago, similar practices are observed in clinics and medical facilities across the country and PSA testing is often the highest volume tumor marker laboratories perform.
PSA Screening in the News
There have been two large clinical trials published very recently in the New England Journal of Medicine that reported results from randomized trials which evaluated the effectiveness of PSA screening.
One of those trials was the American Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial, which recruited approximately 77,000 men between the ages of 55 and 74 years. Participants were placed in either a control group or a group which received a PSA test and digital rectal exam every year for six years. The group that had PSA screening had a 20% increase in the diagnosis rate of cancer; however, the stage and grade of cancer diagnosed was similar to that of the control group. There was no reduction in the rate of prostate cancer deaths due to screening, and there was study contamination due to the fact that about half of the men in the control group had already had a PSA test before or during the trial at some point.
The European Randomized Study of Screening for Prostate Cancer reported on approximately 162,000 men followed for an average of nine years. In this study there were 70% more cases of metastatic cancer and more than twice as many locally advanced cases in the control group compared to the group that underwent regular screening. There was a 20% reduction in death from prostate cancer in the screening group, however the reduction was minimal. The chance of dying from prostate cancer after nine years was reduced from 0.36% to 0.29%.
Furthermore a recent Dartmouth study suggests there have been an estimated 1 million excess diagnoses since 1986, yet the incidence of prostate cancer remains well above the levels that existed prior to widespread PSA screening.
Recommendations for Screening
Several clinical groups have advocated the benefits of screening for prostate cancer. These groups include the American Urological Association, who recommend screening men beginning at age 50 and have a life expectancy of at least 10 years, and even earlier in African-American men or those who have a family history of prostate cancer.
The American Cancer Society has similar guidelines to the American Urological Association, although include a digital rectal exam with the PSA. The National Comprehensive Cancer Network recommends using a PSA cutoff of 2.5 ng/mL, which is lower than the 4.0 ng/mL cutoff traditionally used. Endorsement of prostate cancer screening also comes from the National Academy of Clinical Biochemistry, and all groups recommend that a positive PSA be repeated prior to prostate biopsy.
Arguments for Screening for Prostate Cancer
There are several reasons these groups suggest the benefits outweigh the risks with PSA screening. Prostate cancer can be viewed as a significant health burden in the United States, with an estimated 192,280 new cases and 27,360 prostate cancer-related deaths occurring in the United States in 2009. PSA has a high sensitivity and negative predictive value, depending on what cutoffs are used. Screening can identify localized disease, before tumors have metastasized, and although controversial, it is argued by some that the cancer detected with screening is significant and likely to progress. Lastly, countries that screen for prostate cancer have observed a decrease in mortality from the disease itself but have an increased number of individuals diagnosed and subjected to treatment.
Recommendations for Not Screening
In contrast, there are also clinical groups who argue against screening for prostate cancer. These include the American Academy of Family Physicians, American College of Physicians, American College of Preventive Medicine, Canadian Task Force in Preventive Healthcare, and the US Preventive Services Task Force who all cite a lack of sufficient data proving the benefits of screening or believe that the decision to screen lies solely with the patient.
Arguments Against Screening for Prostate Cancer
The arguments against prostate cancer screening focus on the high rate of detection of indolent cancers with PSA and the overtreatment of tumors unlikely to cause mortality. There is a great amount of lead time bias such that it gives an illusion of improved survival.
Furthermore, because PSA does not have high specificity, a large number of prostate biopsies are performed to completely rule out cancer and there is morbidity associated with the cancer treatment, and less often, with the biopsies themselves. As noted in the recent clinical trials there is no conclusive evidence that screening improves patient outcomes.
Finally, there is significant cost associated with the downstream effect of an elevated PSA, including biopsies, repeat testing, and increased patient anxiety, all of which place further burden on the economic and healthcare system. While there are strong arguments for and against screening, at present it is apparent that for now PSA will continue to be utilized.
PSA Sensitivity and Specificity
The well known total PSA cutoff of 4.0 ng/mL was first proposed in 1986 from a study conducted by the company Hybritech, who evaluated a small population of 472 men without a history of prostate cancer. A further screening study performed at Washington University involved 6630 men aged 50-74 years, and led to the approval of PSA by the US Food and Drug Administration for prostate cancer screening. This study clearly demonstrated the efficacy of the 4.0 ng/mL cutoff, which at the time, was considered an aggressive stance. A cutoff of 10.0 ng/mL was used commonly for triggering prostate biopsies in the early 1990’s.
Consequently, 4.0 ng/mL became well recognized as “the” biopsy cutoff, even though it was well known that men could have cancer with PSA values less than 4.0 ng/mL, and that values greater than 4 produced a 6/2% false-positive rate and detected only 20% of cases.
High-Grade Prostate Cancer is Not Rare When PSA =4.0 ng/mL
Since that time it has been shown by numerous studies that many high-grade prostate cancers are missed using the 4.0 ng/mL cutoff, and a Memorial Sloan Ketterling Cancer Center Study demonstrated that as many as 20% of PSAs that are over 4 ng/mL will eventually return to “normal” within a four year follow up window of time. Other strategies were needed to further optimize the use of PSA.
Increase Specificity Using PSA Velocity
One way to increase the specificity of PSA is to modify how it is interpreted, and evaluate the change in PSA over time (in months or years), termed PSA velocity. If the patient has metastatic cancer the PSA will change rapidly over time, to a greater extent than if there is localized cancer contained within the prostate or to an even a greater extent compared to a control population or men with BPH.
The National Cancer Comprehensive Network guidelines recommend calculation of PSA velocity over a one year period. When the total PSA is less than 4, an increase of greater than or equal to 0.35 ng/mL/year is considered suspicious and when the PSA is between 4-10 ng/mL an increase of greater than or equal to 0.75 ng/mL/year is considered suspicious. Ideally the clinician would use three specimens over an 18 to 24 month period to adequately assess the velocity.
Optimizing Clinical Sensitivity and Specificity: Age/Ethnic Reference Intervals
Use of age and ethnic specific reference intervals was also suggested to improve the specificity and sensitivity of PSA. Many studies have noted that men without prostate cancer will have higher PSA values as they age. Use of age adjusted cutoffs could improve sensitivity in younger males, allowing their cancer to be detected earlier, and improve specificity in older males so the false-positive rate is reduced and less biopsies are performed. This approach has not been uniformly accepted and the most recent National Academy of Clinical Biochemistry guidelines in 2008 do not recommend age-specific reference ranges.
Utilization of Free/Total PSA Ratio
Free PSA can also be measured in serum and increased levels of free PSA were first noted to be associated with benign disease and decreased probability of cancer. Reporting of the free PSA as a free to total PSA ratio was first approved by the FDA in 1998. If the ratio is high (>25%) the probability of cancer is greatly decreased, and similarly if the ratio is low (<10%) it indicates a greater probability of cancer. A limitation of this ratio is that most patients will be somewhere in the middle (the gray zone), and this would still lead to biopsies. Nevertheless, free PSA is used to aid biopsy decisions.
Why Aren't PSA Results Interchangeable?
A controversial topic besides using PSA for screening involves analytical issues/measurements with PSA and often leads us to ask: why aren’t PSA results interchangeable?
Development of PSA Standards
Perhaps it may help to examine the evolution of PSA testing even further. The first PSA assay was a dual monoclonal radioimmunoassay manufactured by Hybritech and referred to as Tandem-R assays. The concentration of the purified PSA used to standardize the Hybritech assay back in 1984 was determined based on Lowry protein methods, which were what was available at that time. In 1991, the Hybritech assay became an automated immunoassay on the Beckman Access and first used in prostate cancer screening after further FDA approval in 1994. Efforts to standardize PSA assays were initiated in 1992 by Dr. Thomas Stamey at the First Stanford Conference. At the Second Stanford Conference on International Standardization of Prostate-Specific Antigen in 1994, Dr. Stamey and colleagues proposed use of a primary calibrator consisting of 90% purified PSA-complexed to alpha1-antichymotrypsin and 10% free PSA (later termed the 90:10 standard) on a molar basis.
However, in 1995, Stamey and colleagues published an article showing that the extinction coefficient for PSA was actually 1.84 mL/mg/cm, and they determined this using quantitative amino acid analysis. The previous extinction coefficient used for PSA standards with the Hybritech assay was 1.42 mL/mg/cm, approximately a 23% difference, with the Hybritech concentration reading higher than the more accurately determined standard. At that time Stamey hypothesized this finding would lead to the Hybritech Tandem-R assay reading 10-20% “too high.” In 1999, the World Health Organization (WHO) established the First International Reference Preparation for PSA (WHO standards 96/670 for total and 96/668 for free PSA). This standard is a recombinant DNA-derived human PSA, which was standardized using the new extinction coefficient.
Although many of the commercial PSA methods used at that time yielded values that were higher than the WHO-assigned value, very little attention came from the medical community about the potential impact of this new standard on prostate cancer screening. Manufacturers then began to standardize PSA assays to the WHO standard, as recommended. Finally, in 2003 it was noted that the change or differences between the Hybritech and WHO standards may have an effect on the medical decision levels.
Effect of Analytical Bias on Classification Based on Fixed Criteria
If we simply examine the effect of analytical bias alone on classification that is based on a fixed criteria, it may be more obvious what this means. Let’s say these curves represent a distribution of patients who have their PSA measured. On the right side of the curve we have our fixed decision threshold (in this case we’ll use 4.0 ng/mL) and the marked area represents our patients with a disease (prostate cancer) we want to detect. If we have an analytical shift in the calibration then we move the curve to the left by 20%. If the cutoff does not shift to match the lower patient results, then the number of patients we detect simply decreases. This, in a very simplistic explanation, is what we see between Hybritech and WHO assays if the cutoff does not change.
Analytical Difference: Results per 1000 Patients Tested
This was also simulated by Dr. George Klee here at the Mayo Clinic, based on the distribution of 20,000 consecutive patient PSA results that were tested for prostate cancer screening purposes. Dr. Klee calculated the effect of different levels of analytical bias on the number of results that exceeded the clinical decision level of 4.0 ng/mL. If there is a positive 6% analytical bias there is an 11% increase in elevated results, and an increased frequency of biopsies. With a negative 6% analytical bias there is a 14% decrease in detection rate, and the patients would not receive the recommended follow-up testing and/or biopsy. Similarly, it is clear with a very large bias of 20% the number of patients categorized based on a cutoff greatly increases. This data suggests that laboratories and manufacturers have to be aware of the great impact of analytical bias in PSA testing and ensure that they reduce this source of error to the lowest possible levels.
Hybritech vs. WHO Standardized Assays
Summarizing these purely analytical differences, it is notable that the WHO standardized calibration yields PSA values between 10 to 20% lower than the Hybritech assays, a difference that is likely due to the different mass assignments of the standards. The initial impact of this calibration change was not realized by the medical community but awareness is certainly increasing.
A study published in Clinical Chemistry in 2006 elegantly demonstrates that an increase of total PSA of 0.5 or 0.75 ng/mL/year, which are values used in the calculation of PSA velocity, can be the result simply of a change in the assay system used rather than an actual change in PSA concentrations. Comparing five different immunoassays for total PSA produced variable results in that study, and all assays agreed within 15% of one another. However, some manufacturers, even though technically traceable to the WHO standard, produced patient results that mirrored the Hybritech assay results, which can be confusing and misleading.
CAP Proficiency Testing
Evaluation of CAP proficiency testing results from 2008 and 2009 also demonstrates the variability between manufacturers. The percent bias calculated is the absolute percent bias which deviates from the Access/Hybritech method, chosen as the predicate or gold standard method. The variation ranges from less than 5% up to 40%. It is clear that further efforts need to be made to produce results which have greater harmonization.
WHO 96/670 Total PSA Preparations
Interestingly, Dr. Klee at Mayo also evaluated the effect of matrix bias back in 2004 and showed that how or what the WHO standard is diluted with has an effect on the PSA results from different assays. In that study, the WHO material was diluted into six different matrices, including PBS buffer with bovine serum albumin, 2 and 5% human serum albumin, manufacturer specific diluents and a synthetic matrix consisting of BSA, BGG (bovine gamma globulin), gelatin, and mannitol in PBS.
The graph on the left shows the total PSA standard at 4 ng/mL, and the graph on the right similar data but at 10 ng/mL. Both graphs show the Access (which used the Hybritech calibrator) giving results that are higher than the Centaur (which used a WHO standard). Variations in the standardization procedure for diluting the WHO standard, which is conducted by manufacturers, may account for some of the differences noted between assays that are traceable back to the same standard.
WHO Calibration/Concordance at 3.1 ng/mL Cutoff
Going back to the original screening study that led to FDA approval of the 4.0 ng/mL cutoff with the Hybritech assay, using concordance calculations alone it can be shown that a 4.0 translates to a 3.1 ng/mL cutoff if the WHO standardized assay is used. Use of this different cutoff yields a clinically equivalent patient distribution with the Hybritech cutoff.
Analyzing a smaller subset of patients, it is again clear that use of different cutoffs for the calibration leads to the same clinical concordance, with no patients being misclassified as elevated or missed using the fixed cutoff.
WHO Calibration/Concordance at 4.0 ng/mL Cutoff
However, if we were to use the same 4.0 cutoff for a Hybritech calibrated and a WHO standardized assay then we would not miss diagnosing any men with prostate cancer but we would have 38 men (or 18%) who would be considered positive on the Hybritech assay but below the 4.0 cutoff with a WHO standardized assay. This signifies those 38 men would not have received follow-up if the same cutoff was used and suggests manufacturers who offer WHO standardized assays need to provide assay-specific cutoffs that logically would not be 4.0 ng/mL, but some concentration below 4.0.
Clinical Differences in PSA Screening
In a study from 2004 that was one of the first clinical reports of the impact of PSA standardization, over 2000 men were tested with the Hybritech assay and two other WHO standardized assays. The study found that the WHO assays yielded results that were 23% lower. Using a cutoff of PSA greater than 4.0 ng/mL, 57 patients (2.5%) would have been candidates for biopsy based on Access but not Centaur data. If a lower cutoff of PSA greater than 2.5 ng/mL had been used, this number would have increased to 107 patients (4.6%). If all men who had a PSA greater than 2.5 ng/mL were considered, 19% of those individuals would have a PSA that was at or above the 4.0 ng/mL cut-off. This represents a “high prevalence of cancer” group and is dramatically similar to the 18% figure derived from the data seen on the previous slide.
The Clinical Difference
In the NPCC (National Prostate Cancer Coalition) grass-roots screening study performed in Chicago, 1916 men (approximately half of who were Caucasian) were compared using the Access which uses Hybritech standard and Siemens Bayer Centaur which is a WHO standard. The findings of the study show a negative bias of greater than 20% in the Centaur values. Using a PSA threshold of 2.5 ng/mL, biopsy would have been recommended for 94 (5%) men using one assay (the Hybritech) but not the other (the Centaur). At 4.0 ng/mL biopsy would have been recommended for 87 men (or 4.5%) using the Hybritech assay but not the Centaur. This is in close agreement with the study on the previous slide, where it was found that 4.6% of men would not have been recommended for biopsy.
Fixed Thresholds Produce Problems for Biopsy Recommendations
A recent study in Clinical Chemistry reported the results of an ‘in-silico’ (or computer modeled) study where 106 samples were used to establish the bias between Hybritech and WHO calibrations. Data from 5865 tests were then compared for biopsy recommendation and cancer detection. In remarkable agreement with the analysis of the previous studies this study found a 19% decrease in the biopsy recommendation rate. These studies demonstrate the ~20% lower results from a WHO calibration result in missed opportunities for follow-up care for men who may be at risk for prostate cancer. Clearly use of a fixed threshold can produce problems for further referrals for biopsy.
Effect on "Watchful Waiting"
Visually we can see this difference again in the figure shown. After some amount of time a man’s PSA becomes elevated or greater than the 4.0 ng/mL cutoff with a Hybritech assay. Using that same cutoff in a WHO standardized assay, the PSA is now less than 4.0 ng/mL or may be considered “normal”. After some period of time, perhaps an abnormal result would be obtained but if the increase in PSA is due to cancer there is a greater risk in increasing the stage and grade of prostate cancer with the watchful waiting approach.
Lowering that cutoff allows the curves to be superimposed, and a positive result to occur at the same time point regardless of whether a Hybritech or WHO assay is used. Obviously this is a very generic representation of the clinical impact but is a concrete illustration of the analytical differences.
Adding Biological Variability into the Mix
One further additive issue to the analytical issues deals with biological variability, which should be considered for proper interpretation. In men over the age of 50 the biological variability from one day to the next is approximately 20%. This variability holds true across the range of total PSA concentrations of 0.2-20 ng/mL. Therefore a change of 30%, which accounts for biological and analytical variability, can be considered normal for some men. Other studies suggest that the baseline PSA actually has to change by 50% to be considered significant.
Futures in Prostate Cancer Testing?
Although many controversies surround PSA there are several other biomarkers on the horizon that are being evaluated for use in prostate cancer. One of them is hK2, which is localized to the prostate and shares a high structural homology (up to 80%) with PSA (which is hK3). Use of hK2 is hypothesized to increase the precision of PSA-based prediction in patients with a history of PSA levels below the standard cutoff thresholds for biopsy, and may be useful in determining clinically insignificant prostate cancers.
Another marker is PCA3 (originally called DD3) and there is one company which markets that assay. PCA3 is a urine assay that measures PCA3 mRNA after a digital rectal exam. Both the PCA3 and PSA mRNA are separately quantified and the ratio of the two, or the PCA3 score, is calculated. A higher PCA3 score indicates a greater likelihood of prostate cancer. The FDA has not approved PCA3 as a diagnostic test but there are several reference laboratories in the US that perform this test. PCA3 is reported to have a sensitivity of 69% and specificity of 79%.
Two other serum biomarkers under investigation are proPSA and BPSA (otherwise called BPH/benign-related PSA), which are different molecular forms of free PSA. These two isoforms act essentially in an opposite manner. BPSA is found at higher levels in hyperplastic tissue from patients with BPH and may be a useful indicator of transition zone prostate volume. proPSA, specifically the truncated -2proPSA, is a molecular form of PSA present in prostate cancer and measurement may help distinguish cancer from benign conditions or assess how aggressive the cancer is. Use of proPSA with BPSA may help reduce the number of unnecessary biopsies. Both proPSA and BPSA are research use only immunoassays at the present time. Further investigation is warranted to help define the role of these and other biomarkers in prostate cancer screening and management.
PSA Testing at Mayo
At the Mayo Clinic, total and free PSA are performed on the Beckman Coulter DXI using the Hybritech calibration. There is a Total PSA Screen which is used for screening purposes and a Total PSA Diagnostic which is useful for monitoring or diagnosing prostate cancer. There is no difference between the total PSA used for screening or diagnostic purposes; however, Medicare does reimburse the tests differently based on the CPT code that is submitted. Total and free PSA can also be measured together and a free to total ratio is reported only if the total PSA is between 2-10 ng/ml. In the near future, we plan to move to a WHO-calibrated method and there are ongoing discussions about the strategy we will use.
In conclusion, PSA assays remain problematic due to a lack of true standardization between WHO calibrated assays, which are needed for screening and diagnostic tests. If a lab changes from a Hybritech assay to a WHO standardized assay, use of different cutoffs should be utilized and education of physicians about the change is critical. Furthermore, many new biomarkers are on the horizon for earlier detection and prediction of prostate cancer. There are many clinical decisions made off of PSA measurements, and there will continue to be until we can give clinicians something that is superior to PSA.