EditorialStockholm3 in a Multi-Ethnic Cohort: Optimizing Prostate Cancer Screening to Reduce Harm and Improve Equity

The Takeaway

In the article accompanying this editorial, Vigneswaran et al.¹ demonstrated that the Stockholm3 test, used to determine which men with elevated serum PSA levels can be safely excluded from further diagnostic evaluation, performs equally well in Asian, Black, Hispanic and non-Hispanic White men. By using Stockholm3 followed by MRI, nearly half of men would not have to undergo prostate biopsy, thereby diminishing the harms of PSA screening.

Prostate cancer screening through routine measurement of serum prostate specific antigen (PSA) has been mired in controversy since its inception in the late 1980s. After years of campaigns encouraging men to “Know your PSA”, the U.S. Preventive Services Task Force in 2012 gave PSA testing a D-rating based on data showing little or no benefit to aggressively diagnosing and treating early staged prostate cancer^2,3 and evidence of significant harms from biopsy and treatment. In 2017, this was revised to a C-rating⁴ based on European randomized trials showing modest survival benefits^5–7 and a mitigation of harm through the increasing use of active surveillance for low risk disease⁸. Adding to this controversy was the heavy burden of prostate cancer in Black men, for whom cancer incidence and death rates are much higher, and more rather than less screening seemed most appropriate⁹. Furthermore, Asian, Black, and Hispanic men remain underrepresented in most clinical trials, including prostate cancer screening.

Key to the PSA controversy is its poor performance as a screening tool, with a specificity of approximately 25% and sensitivity of 85%, resulting in nearly 1.4 million prostate biopsies performed each year in the U.S. to diagnose approximately 300,000 cases^9,10. Complications of prostate biopsies, including urinary tract infections, pain, bleeding, psychological distress, overdiagnosis and overtreatment of indolent cancers^4,11,12, has provided impetus for the development of ancillary tests to blunt the harms of PSA testing. These tests help identify men at low risk for clinically significant prostate cancer (csPCa) (defined as < Grade Group 2 or Gleason grade 3+4=7), who have a low risk for prostate cancer mortality¹³ such that they can be safely excluded from biopsy. In the article accompanying this editorial, Vigneswaran et al.¹ tested the performance of Stockholm3, a reflex test performed in men with a screening PSA ≥ 4 ng/ml, in a prospective multicenter observational trial in 2,129 men, of which 350 were Asian (16%), 505 were Black (24%), 305 were Hispanic (14%), and 969 were non-Hispanic or non-Latino and White (46%, 969). The Stockholm3 test measures PSA, free PSA, KLK2, GDF15, and PSP94 serum protein levels, a polygenic risk score calculated from over 100 germline risk SNPs, the HOX13 G84E germline variant, and clinical parameters including age, first degree family history of prostate cancer (if available), and history of a previous prostate biopsy^14,15. In the overall cohort, the Stockholm3 test resulted in a predicted 48% decrease in biopsies while maintaining sensitivity for identifying csPCa. Importantly, the test performance was maintained in all racial and ethnic groups, maintaining a high sensitivity for csPCa in Asian, Hispanic, Black and White patients, with similar reductions in the need for biopsies across all groups. The SEPTA study provides crucial evidence of validity in multiple racial and ethnic groups and will bolster arguments for broader adaptation in the U.S. and other racially and ethnically diverse countries.

Since the Stockholm3 test was developed in a Swedish population, validation in diverse populations was essential. Stockholm3 has been validated in a large European population outside of Scandinavia¹⁶, and the test is now available throughout Europe, and parts of Africa and South America. Prior to this report, there were legitimate concerns about the generalizability of the Stockholm3 test to diverse populations. The application of polygenic risk prediction scores developed largely in White men to other racial and ethnic populations where unique risk alleles have been identified could dramatically affect performance of the test¹⁷. Likewise, differences in the distribution of somatically acquired mutations and identification of somatically acquired private mutations in different racial and ethnic groups suggests the existence of underlying germline differences that could drive genomic differences between these groups and erode the effectiveness of Stockholm3. No such effects were observed, as the performance of the test in identifying csPCa and eliminating unnecessary biopsies was as good for Asian, Black, and Hispanic men as for White men.

With these findings Stockholm3 can be added to the growing list of reflex tests available for men with an elevated PSA. These tests include clinical nomograms, blood-based tests including the 4Kscore^® (which includes several of the serum proteins and clinical parameters in Stockholm3) the Prostate Health Index (PHI), percent free PSA, as well as urine based tests (ExoDX^™, SelectMDx^®, Progensa^® Prostate Cancer Antigen 3, MyProstateScore¹⁸) and a tissue-based test for men with previous negative biopsies¹⁹. Typically, these assays target a single analyte (protein, RNA, or methylated DNA) and include clinical variables, while Stockholm3 utilizes clinical, protein and germline DNA-based markers (that need be measured only once). Many of these reflex tests have been validated in diverse populations and appear to have similar performance in triaging patients who do not need to be biopsies. However, there have been no rigorous comparisons of these tests with each other.

The elephant in the room for triage of patients with an elevated PSA is multiparametric MR prostate imaging. Several randomized trials show the effectiveness of MRI at eliminating unnecessary biopsies, with miss-rates for csPCa that compares well with the tests listed above^20–22. MRI is becoming standard of care in North America and Europe because it has the advantages of visualizing discrete targets within the whole gland that can be targeted for biopsy. MRI also accurately measures total prostate and transition zone (BPH) size for calculations of PSA density which contribute to interpretation of elevations of PSA²³. For these reasons, in most large urologic oncology practices, including ours, MRI has become the default reflex test in men with an elevated PSA and guidelines in the U.S. are changing to reflect this practice.

Given the utility and performance of MRI, many clinician investigators have questioned the need for ancillary tests²⁴. However, as a reflex test MRI adds significant cost to screening. Given the high prevalence of men with PSA elevations and clinically localized prostate cancer, MRI use has skyrocketed, leading to significant scheduling backlogs at MR imaging facilities. Similar backlogs occur in biopsy scheduling due a shortage of fusion biopsy-trained urologists and ultrasound fusion equipment available, particularly in rural and underserved areas. Likewise, significant variations in interpretation of scans by radiologists can affect diagnostic accuracy²⁵.

With all the available tests, how do we best screen for prostate cancer? In the current health care system, it is common to see patients who have had several of these tests, many times with conflicting results, who are confused and anxious about the best path forward (Figure 1). This market-based approach has led to worries that the plethora of tests will merely end up being cost-added, rather than value-added in the diagnostic space since biopsy is the only way to resolve the issue²⁶. The Early Detection Research Network (EDRN) is recruiting men prior to biopsy for collection of blood and urine samples with pre-biopsy MRI with the goal of testing screening tests, singly and in combination, to determining the optimal combination and sequence to optimize detection protocols (NCT03784924).

Figure 1:

Of the many available tests for selection of patients for prostate biopsy (top), we do not know when to deploy these tests (before or after MRI), whether they provide overlapping information, whether they are generalizable, and how to interpret conflicting results. Deploying Stockholm3 after PSA, followed by MRI (bottom) reduces biopsy rates by 42–52% across racial/ethnic groups, without compromising detection of significant cancers.

Rather than waiting for these studies to sort out the optimal approach, several groups, most notably the Stockholm3 investigators, are proceeding with testing algorithms that involve baseline PSA testing, followed by one of the biomarker tests for selection of men for MRI and later biopsy^27,28. This approach has the advantage of using low resource intensive blood or urine tests to reduce the use of expensive and resource intensive MRI scanning^29,30. Stockholm3 followed by MRI has been demonstrated to be non-inferior to PSA testing alone or PSA+MRI in all men with PSA≥ 3.0 ng/ml, and significantly decreased MRI use and unnecessary biopsies³¹. The investigators are now testing whether Stockholm3 can be used without PSA triage to select men for MRI and biopsy³². The ProScreen population-based randomized trial in Finland compares PSA testing, reflex 4Kscore^® for PSA≥ 3.0 ng/ml and referral for MRI and targeted biopsy. The 15,201 men invited to screen (of whom 7744 attended screening sessions) showed a higher incidence of csPCa compared to 45,544 men not invited (1.13% vs. 0.62%)³³. Therefore, these approaches appear to achieve the goal of maintaining (or possibly improving) detection of csPCa, decreasing detection of low grade prostate cancer and decreasing biopsies in men without prostate cancer. While these approaches appear to better use resources and decrease the harms of PSA testing, they ultimately need to be compared to PSA testing alone in large prospective randomized trials to ensure that the small number of csPCa (and larger group of low risk prostate cancers) missed by these staged approaches do not erode prostate cancer specific survival benefits from screening⁶. Given the similar performance of the reflex tests and the shortages of MRI capacity, U.S.-based investigators should be encouraged to carry out similar staged clinical trials of biomarker driven selection of patients for MRI compared to PSA alone or PSA and MRI alone.

Lastly, while laudable that the investigators have shown excellent performance of Stockholm3 across diverse populations, additional work is necessary to make sure these tests are used fairly. Broadly speaking, cancer screening and treatment of localized disease are unevenly applied in the U.S. leading to significant inequalities of outcomes³⁴. In prostate cancer, use of prognostic genetic tests on biopsies for treatment planning shows significant disparities in use. In part, these inequities can be attributed to socioeconomic factors, including insurance status, as observed in Hispanic men who show lower rates of definitive treatment for localized high risk prostate cancer³⁵. Structural barriers, such as access to specialty care and expertise, transportation, and distance to the nearest medical facility are undoubtedly other important factors. Finally, physician attitudes and beliefs likely contribute to differences in care, evidenced by the low and decreasing use of active surveillance in Black men with low risk prostate cancer³⁶ despite evidence of identical outcomes compared to White men³⁷. A structured and validated screening approach could help address these inequities and provide a benchmark for evaluating screening practices.