Precision Medicine in Prostate Cancer: The ProGRESS Study and Beyond

The ProGRESS Study and Beyond

Pathology · Seminar week 38 · released October 5, 2026 · includes a discussion video

The ProGRESS study utilizes a precision screening approach to prostate cancer by integrating polygenic risk scores with genetic ancestry and family history, moving beyond…

Learning Objectives

By the end of this seminar, learners will be able to:

1. Distinguish population-based PSA screening from risk-adapted precision screening.
2. Explain how P-CARE integrates polygenic risk, genetic ancestry, and family history.
3. Interpret the complementary roles of polygenic risk scores and pathogenic germline variants.
4. Apply calibration, discrimination, and clinical-utility principles when evaluating risk models.
5. Identify structural, clinical, and genomic contributors to disparities in prostate cancer outcomes.
6. Integrate PSA, prostate MRI, biomarkers, and biopsy findings into a coherent diagnostic pathway.
7. Recognize the evidentiary, ethical, and implementation requirements for translating ProGRESS findings into practice.

Introduction to Prostate Cancer Screening

%%FIG0%% Prostate cancer screening presents an unusually difficult balance: the disease is common and potentially lethal, yet many screen-detected tumors would never become symptomatic during the patient’s lifetime. Prostate-specific antigen, or PSA, is organ-specific but not cancer-specific. Benign prostatic hyperplasia, urinary retention, infection, recent instrumentation, and normal biological variation can all elevate PSA. Conversely, clinically significant cancer can occur at a PSA below traditional biopsy thresholds. A single cutoff therefore compresses a continuous, context-dependent risk into an artificial binary result.

Randomized trials established both the potential benefit and the limitations of PSA screening. In the European Randomized Study of Screening for Prostate Cancer, screening reduced prostate cancer mortality, but the absolute benefit emerged over prolonged follow-up and required many invitations, tests, and diagnoses (ERSPC; PMID: 25108889). In contrast, the CAP trial found that a one-time PSA invitation increased low-risk cancer detection without demonstrating a prostate cancer mortality reduction at 10 years (PMID: 29509864). Differences in screening intensity, contamination, follow-up, diagnostic protocols, and treatment complicate direct comparisons among trials.

Teaching Point: PSA is a risk marker, not a diagnosis. The clinical question is not simply whether PSA is “normal,” but whether the patient’s probability of clinically significant cancer—generally Grade Group 2 or higher—is sufficient to justify MRI, biomarker testing, or biopsy.

Before escalating evaluation, confirm a newly elevated PSA. Review prior values, laboratory variability, prostate volume, medications, and reversible causes. Febrile urinary infection or acute bacterial prostatitis warrants treatment based on clinical findings, followed by delayed reassessment after recovery; empiric antibiotics should not be prescribed merely to lower PSA in an asymptomatic patient. Avoid testing immediately after urinary retention, catheterization, cystoscopy, or prostate biopsy. Finasteride and dutasteride typically reduce PSA by approximately 50% after six to twelve months, so the measured value and its trajectory require adjustment. PSA velocity should not serve as the sole indication for biopsy.

The AUA/SUO early-detection framework supports shared decision-making, a baseline PSA at approximately 45–50 years for average-risk patients, and consideration of initiation at 40–45 years for people at increased risk, including those with a strong family history, Black ancestry, or a relevant germline pathogenic variant. For many adults aged 50–69 years who choose screening, an interval of two to four years can be individualized according to baseline PSA and risk. Screening should not continue reflexively when limited life expectancy makes benefit improbable.

Framework: A precision pathway moves through four questions: What is the patient’s baseline risk? Is the PSA result reproducible and interpretable? What is the probability of Grade Group 2 or higher disease? Would finding cancer improve this patient’s length or quality of life?

The harms extend beyond false-positive tests. Biopsy can cause pain, bleeding, infection, sepsis, and urinary retention. Diagnosis may create anxiety and lead to treatment-related urinary, sexual, or bowel dysfunction. The 15-year ProtecT results showed low prostate cancer mortality across active monitoring, surgery, and radiotherapy for localized PSA-detected disease, although metastases and progression were more frequent with monitoring (PMID: 36912538). Screening must therefore be connected to high-quality risk classification and active surveillance; otherwise, improved detection can merely produce more treatment.

Precision screening replaces a uniform PSA threshold with a layered assessment incorporating age, health status, PSA history, prostate volume, family history, inherited variation, ancestry, imaging, and patient preferences. P-CARE, as evaluated through ProGRESS, advances this concept by integrating polygenic risk scores, genetic ancestry, and family history (PMID: 41588240).

MUST ACT: Do not send a patient directly from one mildly elevated PSA to biopsy without confirming the result and estimating clinically significant cancer risk, unless examination or other findings indicate urgent evaluation.

Nuance: Precision screening does not mean testing everyone more intensively. Its intended value is bidirectional: earlier or more sensitive evaluation for high-risk patients and less frequent testing or fewer invasive procedures for those at low absolute risk.

Audience Poll: A healthy 52-year-old has a first PSA of 3.2 ng/mL, no urinary symptoms, and a normal examination. Which should come next: immediate biopsy, repeat PSA with risk review, empiric antibiotics, or discharge from screening?

Development of the P-CARE Model: Components and Validation

%%FIG1%% P-CARE represents a transition from isolated risk factors to an integrated estimate of susceptibility. In the ProGRESS framework, its central inputs are polygenic risk, genetic ancestry, and family history (Vassy et al., Nature Cancer, 2026; PMID: 41588240). Each captures a different domain. A polygenic risk score summarizes common inherited alleles; genetic ancestry helps characterize population structure and model transportability; family history captures shared rare variants, polygenic effects, exposures, screening behavior, and outcomes that may not be represented in a genotype panel.

Family history must be elicited with enough specificity to be clinically useful. “Prostate cancer in the family” is inadequate. Record the relative’s degree of relatedness, age at diagnosis, metastatic or lethal phenotype, and whether multiple relatives are affected. Ask about breast, ovarian, pancreatic, colorectal, endometrial, upper urinary tract, and other cancers that could suggest hereditary breast–ovarian cancer or Lynch syndrome. Maternal and paternal lineages matter equally. Family history is dynamic and should be updated over time.

Framework: Model development and model validation are different tasks. Development estimates relationships between predictors and outcome. Validation asks whether those relationships remain accurate in patients who were not used to fit the model.

A credible validation program should evaluate discrimination, calibration, and clinical utility. Discrimination describes how well a model ranks people who will and will not experience the outcome; the area under the receiver-operating-characteristic curve is common but does not reveal whether predicted probabilities are numerically accurate. Calibration compares predicted and observed risk across the full range and within clinically important subgroups. A model can discriminate reasonably while systematically overestimating risk in one ancestry group and underestimating it in another. Decision-curve analysis then asks whether using the model at realistic action thresholds produces more benefit than screening everyone or no one.

Internal validation through bootstrapping or cross-validation estimates optimism within the development dataset. External validation is more demanding: testing across different institutions, geographic regions, recruitment mechanisms, genotyping platforms, age distributions, and ancestry groups. Temporal validation is also necessary because diagnostic pathways evolve. A model trained when systematic transrectal biopsy was standard may behave differently in an MRI-first, transperineal-biopsy era.

ProGRESS is important because it focuses attention on implementation across diverse populations rather than assuming that a genomic score has universal meaning (PMID: 41588240). However, “validated” should never be treated as a permanent label. Model performance depends on the outcome definition, follow-up period, missing-data strategy, genotyping quality, population, and downstream diagnostic protocol. Prediction of any prostate cancer is less clinically useful than prediction of Grade Group 2 or higher cancer, metastatic disease, or prostate cancer death.

Teaching Point: A statistically significant association is not sufficient for clinical deployment. The model must change decisions at thresholds acceptable to patients and clinicians.

The P-CARE output should be interpreted as an absolute risk over a defined period, not merely a percentile. Being in the top decile may sound alarming, but the decision depends on the underlying incidence, age, life expectancy, and competing mortality. Conversely, a “below-average” result does not confer immunity. Risk communication should state the endpoint, time horizon, comparison population, uncertainty, and proposed action.

A practical workflow could use P-CARE before or alongside baseline PSA to individualize the age of initiation and testing interval. A higher estimated inherited risk might support earlier PSA testing, closer follow-up, or a lower threshold for MRI. A lower estimate might support longer intervals after a reassuring PSA. Exact thresholds require prospective clinical-utility data and should not be reverse-engineered from a publication abstract or substituted for guideline-based judgment.

Decision Point: Before acting on a score, ask whether the assay and model were validated for this patient’s ancestry, whether the endpoint matches the proposed decision, and whether the result adds information beyond age, PSA, prostate volume, and family history.

MUST ACT: Use genetic counseling or appropriately trained clinicians when testing may reveal a high-penetrance hereditary cancer syndrome, create implications for relatives, or generate uncertain findings. A susceptibility score is not equivalent to diagnostic germline testing.

Audience Poll: Which failure would concern you most: a modest C-statistic, poor calibration in Black patients, unclear action thresholds, or absence of evidence that model-guided care improves patient outcomes?

Polygenic Risk Scores vs. Monogenic Variants in Risk Assessment

%%FIG2%% Polygenic and monogenic risk are complementary rather than competing concepts. Most inherited susceptibility to prostate cancer is distributed across many common variants, each exerting a small effect. A polygenic risk score weights and combines these variants to position an individual along a continuum of relative risk. By contrast, monogenic assessment seeks rare pathogenic or likely pathogenic variants with larger effects and recognizable hereditary implications.

Large genome-wide association studies have identified hundreds of prostate cancer susceptibility loci. Multi-ancestry analyses have improved risk-score construction and demonstrated that men in the highest polygenic-risk strata can have several-fold greater risk than those near the middle of the distribution (for example, the 269-variant analysis by Conti et al.; PMID: 33398198). Such stratification is attractive for determining when screening begins and how often it occurs. It does not establish that everyone with a high score needs MRI or biopsy.

Nuance: Most prostate cancer polygenic scores predict diagnosis more strongly than lethal disease. If a score primarily enriches for indolent tumors, using it to intensify biopsy could worsen overdiagnosis. Endpoints such as Grade Group 2 or higher cancer, metastasis, and prostate cancer mortality are therefore essential.

Monogenic variants answer different questions. BRCA2 pathogenic variants confer a substantial risk of prostate cancer and are associated with earlier onset and more aggressive disease. BRCA1, ATM, CHEK2, PALB2, mismatch-repair genes, and HOXB13 may also be relevant, although penetrance and phenotype vary by gene and variant. The recurrent HOXB13 G84E variant was associated with hereditary prostate cancer in the original report (PMID: 22236224), but its frequency differs markedly across populations. In metastatic prostate cancer, inherited DNA-repair variants are sufficiently prevalent to affect patients and relatives; Pritchard and colleagues identified germline DNA-repair alterations in approximately 12% of men with metastatic disease (PMID: 27433846).

The IMPACT study provides prospective evidence for targeted PSA screening in BRCA1/2 carriers. Interim results showed that BRCA2 carriers were more likely to be diagnosed with clinically significant disease, supporting systematic screening in this population (PMID: 31537406). In practice, many guidelines recommend beginning shared decision-making and PSA screening at age 40 for BRCA2 carriers and considering a similar approach for other high-risk pathogenic variants.

A pathogenic BRCA2 result also has consequences beyond screening: cascade testing for relatives, breast and pancreatic cancer risk assessment, and potential treatment implications if cancer develops. Polygenic scores generally do not create the same syndrome-level implications. They should not be used to exclude a hereditary syndrome, and a low polygenic score should never override a pathogenic high-penetrance variant.

Framework: Think of monogenic testing as identifying a high-impact pathway and polygenic scoring as estimating background susceptibility. Family history supplies phenotype and context. The integrated result is stronger than any component in isolation.

Testing strategy should follow the clinical question. A patient with metastatic, regional, very-high-risk, high-risk, or suggestive familial disease may meet criteria for germline multigene-panel testing independent of a polygenic score. Pretest counseling should address possible positive, negative, and variant-of-uncertain-significance results. A variant of uncertain significance is not actionable and should not drive intensified screening, prophylactic procedures, or testing of unaffected relatives.

Polygenic scores create additional interpretation problems. Performance can degrade when allele frequencies, linkage disequilibrium patterns, or effect estimates differ between the development cohort and the tested population. Converting a European-ancestry percentile directly into absolute risk for a patient with predominantly African, Asian, Indigenous American, or admixed ancestry can misclassify risk. Updated multi-ancestry models reduce but do not eliminate this problem.

Decision Point: If a patient has three relatives with early or metastatic prostate cancer, pursue hereditary-cancer evaluation even if his polygenic score is average. If he has no suggestive pedigree but a high validated polygenic score, consider risk-adapted screening—not syndrome labeling.

Teaching Point: Neither test determines whether cancer is currently present. PSA, examination, biomarkers, MRI, and biopsy answer the near-term diagnostic question.

Risk results should be communicated numerically when possible: estimated absolute risk, relative risk, reference population, and uncertainty. Avoid deterministic language such as “the cancer gene” or “genetically safe.” The psychological and insurance implications of germline testing should also be discussed, including the protections and limitations of applicable genetic nondiscrimination laws.

Audience Poll: Would you manage a high polygenic-risk score differently from a pathogenic BRCA2 variant, and what specific action would differ?

Addressing Health Disparities in Prostate Cancer Outcomes

%%FIG3%% Black men in the United States experience higher prostate cancer incidence and mortality than White men. The disparity reflects interacting forces: inherited susceptibility, tumor biology, environmental exposures, comorbidity, access to longitudinal primary care, screening patterns, diagnostic delay, insurance, geographic availability of MRI and specialists, treatment quality, and structural racism. A precision model can help identify risk, but it cannot by itself repair inequitable care delivery.

MUST ACT: Do not use race as a biological shortcut. Self-identified race, genetic ancestry, and exposure to racism are related in population data but are not interchangeable variables.

Genetic ancestry estimates patterns of inherited similarity to reference populations. They can help correct population stratification and improve calibration of genomic models. Race is a social and political classification that also captures lived experience, discrimination, neighborhood conditions, and access to care. Substituting ancestry for race can erase social mechanisms; treating race as purely genetic can reinforce biological essentialism. P-CARE’s inclusion of genetic ancestry is valuable when it improves the validity of genomic risk estimates, but equitable implementation requires simultaneous attention to social determinants (PMID: 41588240).

Historically, genome-wide association datasets have overrepresented people of European ancestry. A score can appear technically precise while providing less accurate probabilities for underrepresented groups. The direction of error matters. Underestimation may delay screening or MRI in populations already experiencing adverse outcomes. Overestimation may increase unnecessary biopsies and complications. Reporting one overall performance statistic can conceal both problems.

Framework: Equity evaluation should include representation, measurement, calibration, access, action, and outcomes. Who was enrolled? How were ancestry and race measured? Is the model calibrated within groups? Can patients obtain the recommended next test? Do clinicians act consistently? Does implementation reduce advanced disease without disproportionate harm?

Family history can also be measured inequitably. Smaller family size, estrangement, early deaths from competing causes, adoption, limited access to diagnoses, and nondisclosure can produce an apparently “negative” pedigree. Clinicians should distinguish “no affected relatives” from “family history unknown or uninformative.” A missing history should not automatically lower estimated risk.

Access to prostate MRI illustrates the difference between predictive accuracy and real-world effectiveness. A model may appropriately recommend MRI, but the benefit disappears if the nearest scanner has a six-month wait, the study is unaffordable, or no experienced reader is available. Similar barriers affect transperineal biopsy, genetic counseling, confirmatory testing, active surveillance, radiation, and surgery. Safety-net and rural systems need referral pathways, transportation support, tele-genetics, and quality monitoring—not merely licenses for a genomic calculator.

Teaching Point: Equal thresholds do not necessarily produce equitable outcomes when baseline risk and access differ. Risk-adapted screening can be equitable if thresholds are validated, resources are available, and harms are measured across groups.

Implementation teams should prospectively audit who is offered testing, who completes it, turnaround time, rates of inadequate samples, MRI use, biopsy route, Grade Group 2 or higher detection, complications, treatment, and loss to follow-up. Results should be stratified by race, ancestry, rurality, socioeconomic measures, language, and health system. Community advisory input is particularly important when consent materials discuss ancestry, data sharing, or future genomic research.

The language used during counseling matters. A higher-risk result should be framed as an opportunity for tailored prevention, not as a racial destiny. Conversely, clinicians should not avoid discussing elevated population risk out of discomfort; doing so may perpetuate delayed detection. Shared decision-making must be culturally responsive and supported by understandable absolute-risk estimates.

Decision Point: If a precision pathway identifies a patient as high risk but MRI access is delayed, the correct response is not passive waiting. Escalate navigation, consider validated secondary biomarkers when appropriate, repeat PSA according to clinical urgency, and involve urology.

Nuance: An algorithm can amplify inequity without using race explicitly. ZIP code, health-care utilization, missingness, and prior testing may encode unequal access. Fairness evaluation must examine outcomes, not merely the model’s input list.

Audience Poll: Which investment would most improve equity in your setting: more diverse genomic data, earlier PSA outreach, MRI capacity, patient navigation, or standardized follow-up of abnormal results?

Integration of MRI and AI in Precision Screening

%%FIG4%% Multiparametric MRI has changed the diagnostic pathway by allowing clinicians to localize suspicious lesions, estimate the likelihood of clinically significant cancer, and target biopsy. It is best viewed as a triage and localization test after risk assessment, not a universal replacement for PSA or histology.

A conventional multiparametric examination includes high-resolution T2-weighted imaging, diffusion-weighted imaging with apparent-diffusion-coefficient maps, and dynamic contrast enhancement. Biparametric protocols omit contrast and may reduce time and cost when performed and interpreted well. PI-RADS v2.1 standardizes acquisition and assigns lesions a score from 1 to 5, but interpretation remains influenced by scanner quality, artifacts, prostate zone, reader experience, and inflammatory or benign mimics.

PROMIS established MRI’s value as a prebiopsy triage test by comparing multiparametric MRI and transrectal ultrasound-guided biopsy against a detailed mapping-biopsy reference standard (PMID: 28110982). PRECISION subsequently demonstrated that an MRI pathway with targeted biopsy detected more clinically significant cancer and fewer clinically insignificant cancers than standard systematic biopsy in biopsy-naïve men; approximately one quarter of patients in the MRI arm avoided biopsy (PMID: 29552975). Göteborg-2 extended the evidence into a screening context, showing that restricting biopsy to MRI-visible lesions reduced detection of clinically insignificant cancer, although the balance between avoided diagnoses and missed significant disease remains central (PMID: 36477032).

Teaching Point: A negative MRI lowers risk; it does not reduce risk to zero. MRI-invisible Grade Group 2 or higher cancers occur, particularly with small-volume, infiltrative, or technically obscured lesions.

MRI should be interpreted with pretest probability. PSA density—the serum PSA divided by MRI- or ultrasound-derived prostate volume—is particularly useful. A negative MRI and low PSA density may justify surveillance in an average-risk patient, whereas a negative MRI may not be sufficiently reassuring in a BRCA2 carrier, a patient with strongly abnormal digital rectal examination, a high-risk P-CARE profile, or persistently rising PSA. Thresholds such as 0.15 ng/mL/cc are commonly used, but they should not be treated as universal biological boundaries.

When biopsy is indicated, MRI-targeted sampling can be performed by cognitive targeting, software fusion, or in-bore techniques. For an MRI-visible lesion, obtain multiple targeted cores according to lesion size and local protocol. Whether to add systematic cores depends on biopsy history, MRI findings, baseline risk, and the consequence of missing disease outside the target. Combined targeted and systematic biopsy usually maximizes detection but also increases diagnosis of low-grade tumors. Transperineal biopsy is increasingly favored because it markedly reduces infectious complications and samples anterior regions effectively; local anesthesia commonly permits an outpatient procedure.

MUST ACT: Verify that the MRI is technically adequate and read by an experienced radiologist before allowing a “negative MRI” label to terminate evaluation in a high-risk patient.

Artificial intelligence may assist prostate segmentation, lesion detection, PI-RADS classification, image quality assessment, registration, and targeted-biopsy planning. Computer-aided detection can highlight candidate lesions; computer-aided diagnosis can estimate malignancy probability. Radiomics and deep-learning systems may integrate T2, diffusion, PSA density, age, and prior biopsy data. The attractive endpoint is not merely higher sensitivity, but reliable detection of Grade Group 2 or higher cancer while reducing unnecessary biopsy and reader variability.

Nuance: Retrospective performance on curated images is not equivalent to clinical benefit. Algorithms may fail after changes in scanner vendor, field strength, coils, acquisition parameters, reconstruction software, or patient mix. External and prospective validation are mandatory.

AI also creates automation bias. A radiologist may discount a subtle lesion because the algorithm did not flag it or overcall a benign transition-zone nodule because it received a high score. Human–AI performance, workflow integration, reading time, subgroup calibration, and failure detection should be evaluated—not just standalone model accuracy. Institutions need version control, monitoring for performance drift, and a defined process for discordance between radiologist and algorithm.

Decision Point: For a PI-RADS 3 lesion, do not let the category alone dictate biopsy. Integrate PSA density, lesion location and size, prior biopsy, family history, germline status, P-CARE risk, age, and patient preferences.

Framework: Precision screening is sequential Bayesian updating: inherited risk establishes the prior probability; PSA and biomarkers modify it; MRI refines localization and likelihood; biopsy establishes histology. Each test should be ordered only if its result could change the next decision.

Audience Poll: Would you defer biopsy after a technically excellent PI-RADS 2 MRI in a patient with persistently elevated PSA density and a pathogenic BRCA2 variant?

Future Directions and Clinical Application of Findings from the ProGRESS Study

%%FIG5%% The ProGRESS study advances a compelling hypothesis: prostate cancer screening can be made more efficient and equitable by integrating polygenic risk, genetic ancestry, and family history rather than relying on age and PSA alone (PMID: 41588240). Translation into routine care, however, requires evidence that model-guided decisions improve outcomes meaningful to patients.

The next evidentiary step is prospective clinical utility. A randomized or carefully designed pragmatic trial could compare a P-CARE-guided pathway with guideline-concordant usual care. Important endpoints would include detection of Grade Group 2 or higher cancer, interval and metastatic presentations, unnecessary MRI and biopsy, Grade Group 1 overdiagnosis, biopsy complications, treatment burden, anxiety, decisional conflict, cost, and prostate cancer mortality. Because mortality requires long follow-up, intermediate endpoints must be chosen carefully and linked to later lethal disease.

Framework: The translational sequence is analytic validity, clinical validity, clinical utility, implementation validity, and population impact. A model can succeed at the first two stages and still fail in practice.

Analytic validity includes genotype call quality, imputation, ancestry estimation, reproducibility across arrays, and handling of missing variants. Clinical validity includes discrimination and calibration for relevant endpoints. Clinical utility asks whether the information improves decisions. Implementation validity tests whether real clinics can deliver the pathway reliably. Population impact assesses whether advanced disease and mortality fall without unacceptable overdiagnosis, cost, or inequity.

A near-term application may be personalization of screening initiation and interval. Rather than testing every patient on the same schedule, a validated model could identify those who merit a baseline PSA in their early forties and those who can safely wait or undergo longer intervals after a low result. Later in the pathway, P-CARE could be combined with PSA density, kallikrein-based or urine biomarkers, MRI, and clinical risk calculators to estimate the probability of Grade Group 2 or higher disease.

Decision Point: Adding a test is justified only if it changes management. If every patient receives the same PSA schedule regardless of P-CARE result, testing adds cost and complexity without clinical utility.

Electronic health-record integration should present concise, actionable information: absolute risk over a specified interval, the reference population, uncertainty, major contributors, and a recommended action linked to guideline logic. Alerts should be limited to decisions that require attention. The system must also recognize when the model is outside its validated population or when key data—such as an informative family history—are missing.

Consent and governance are central. Patients should understand whether testing is clinical or research, what results will be returned, whether samples or data will be retained, and who may access them. Germline findings may affect relatives who did not consent. Recontact policies are needed when variant classifications or risk models change. Institutions must protect against unauthorized data use while enabling responsible recalibration and outcome monitoring.

MUST ACT: Do not deploy a static polygenic score as though it were a lifetime truth. Version the assay and model, record the reference population, and reassess recommendations when evidence or family history changes.

Economic evaluation must include more than the price of genotyping. A successful pathway may reduce repeated PSA testing and unnecessary biopsy but increase MRI, counseling, navigation, and early diagnostic activity. Coverage policies should be evaluated for their effect on access. If patients with fewer resources cannot complete the recommended downstream pathway, precision testing may widen rather than narrow disparities.

Clinicians will also need education on risk communication. Relative-risk percentiles should be translated into absolute probabilities and paired with concrete options. Uncertainty must be explicit. A patient may rationally choose more intensive screening because avoiding metastatic cancer is the dominant priority, while another may prioritize avoiding biopsy and overdiagnosis.

Teaching Point: P-CARE should augment shared decision-making, not automate it. Patient age, health, life expectancy, values, and willingness to undergo downstream testing remain decisive.

Research should evaluate whether polygenic risk modifies the penetrance or screening response of monogenic variants, whether models predict aggressive rather than merely incident disease, and whether serial biomarkers or imaging can update inherited risk estimates. Prospective recruitment must deliberately include Black, Asian, Latino, Indigenous, admixed, rural, and socioeconomically diverse populations.

Nuance: The most successful future model may not be the most complex. A simpler tool with transparent thresholds, good calibration, inexpensive inputs, and reliable follow-up may outperform a marginally more accurate model that clinicians and patients cannot use.

Audience Poll: What evidence would you require before adopting P-CARE routinely: external calibration, fewer biopsies, more Grade Group 2 detection, reduced metastatic disease, cost-effectiveness, or all of these?

Case Scenario: Integrating ProGRESS Screening in Clinical Practice

Presentation

A 56-year-old Black man and military veteran presents for preventive care. He has no urinary symptoms and an estimated life expectancy exceeding 20 years. His father developed metastatic prostate cancer at 63, and a paternal uncle was treated for prostate cancer in his early sixties. No relative is known to have undergone germline testing.

His first PSA is 3.8 ng/mL. Digital rectal examination reveals a mildly enlarged, smooth prostate without nodules. He has no fever, dysuria, pyuria, urinary retention, or recent instrumentation. He is not taking finasteride or dutasteride.

Decision Point: The result should not trigger immediate biopsy or empiric antibiotics. Repeat PSA under standardized conditions, clarify the pedigree, and estimate baseline and near-term risk.

Six weeks later, PSA remains elevated at 3.7 ng/mL. The clinician explains that the differential includes benign enlargement, biological variation, inflammation, and prostate cancer. His age, Black race as a marker of population-level disparity, and family history already justify careful evaluation. P-CARE adds a high polygenic-risk estimate using a model validated for his ancestry, while genetic ancestry is incorporated for calibration rather than treated as a deterministic explanation for risk (PMID: 41588240).

Because his family history includes multiple affected close relatives and metastatic disease, he is referred for genetic counseling. A clinical germline panel finds no pathogenic variant and one variant of uncertain significance.

Teaching Point: The uncertain variant does not explain the pedigree and must not guide management. A negative panel also does not erase familial or polygenic risk; current panels do not capture every inherited mechanism.

Prostate MRI shows a 44-mL gland and a 9-mm PI-RADS 4 lesion in the left posterolateral peripheral zone without extracapsular extension. PSA density is approximately 0.084 ng/mL/cc. Although the density is not markedly elevated, the MRI lesion and high integrated baseline risk support biopsy.

He undergoes an outpatient transperineal biopsy with targeted cores from the MRI lesion plus systematic sampling. Pathology identifies acinar adenocarcinoma, Grade Group 2, Gleason 3+4, in targeted cores, with a small proportion of pattern 4 and no cribriform morphology. Systematic cores are otherwise negative.

Clinical Reasoning

The diagnostic sequence illustrates the proper role of precision screening. P-CARE did not diagnose cancer. It modified the prior probability and supported timely escalation after a reproducibly elevated PSA. MRI localized the lesion and refined biopsy planning. Histology established the diagnosis.

The next decision is not automatically radical treatment. Staging and management depend on PSA, clinical stage, tumor volume, Grade Group, adverse histologic features, MRI findings, life expectancy, and patient preferences. Selected patients with favorable intermediate-risk, low-volume Grade Group 2 disease may consider active surveillance, while surgery and radiotherapy remain appropriate definitive options. Germline and polygenic risk can inform counseling, but they should not independently dictate treatment in the absence of validated evidence.

The patient reviews active surveillance, prostatectomy, and radiotherapy in a multidisciplinary visit. If he chooses surveillance, a structured protocol should include serial PSA, clinical review, repeat MRI when indicated, and confirmatory tissue assessment rather than unstructured “watchful waiting.” If he chooses definitive therapy, counseling should compare cancer-control expectations with urinary, sexual, and bowel effects.

MUST ACT: Close the loop on every abnormal screening result. Precision screening fails if a high-risk result generates a referral but the patient is lost between primary care, MRI, biopsy, pathology review, and treatment counseling.

Counterfactuals

If his repeat PSA had normalized and remained low, P-CARE could still justify a shorter screening interval without immediate MRI. If MRI had been PI-RADS 2, the team would integrate MRI quality, PSA density, persistent PSA behavior, family history, and inherited risk before deferring biopsy. If a pathogenic BRCA2 variant had been identified, screening intensity, counseling, cascade testing, and eventual treatment considerations would change substantially.

Nuance: This case demonstrates improved stratification and diagnostic coherence, not proof of improved survival. Whether P-CARE-guided pathways reduce metastatic disease or mortality requires prospective outcome data.

Audience Poll: At which point did precision information most alter this patient’s care: screening initiation, interpretation of PSA, MRI referral, biopsy strategy, or postdiagnosis counseling?

Tonight on Shift

1. Confirm before escalating. Repeat a newly elevated PSA and assess infection, retention, instrumentation, prostate volume, medications, prior results, health status, and life expectancy.

1. Define the inherited-risk question. Use family history, an ancestry-validated polygenic score, and indicated germline testing for complementary purposes; never treat a variant of uncertain significance as actionable.

1. Estimate clinically significant cancer risk. Focus on Grade Group 2 or higher disease rather than any cancer, and communicate absolute risk, time horizon, uncertainty, and the intended decision.

1. Use MRI in context. Combine PI-RADS, MRI quality, PSA density, prior biopsy, examination, inherited risk, and patient preference. A negative MRI is not an automatic stop signal in a high-risk patient.

1. Choose biopsy deliberately. When tissue is needed, consider transperineal sampling, appropriate MRI targeting, and selective systematic cores while balancing missed significant cancer against low-grade overdiagnosis.

1. Audit the whole pathway. Track access, completion, delays, complications, clinically significant cancer detection, low-risk overdiagnosis, and loss to follow-up across racial, ancestry, geographic, and socioeconomic groups.

Read this seminar as Markdown · All seminars · Lecture library · Question bank