📄 medicine

Utility of PSA density velocity in diagnosing prostate cancer on prostate MR

This study found that while a higher PSAD velocity is associated with clinically significant prostate cancer in univariate analysis, it does not serve as an independent predictor when evaluated alongside other factors such as PI-RADS score, age, and PSAD in a multivariate model.

Original authors: Sabarish Narayanasamy, Yasuhisa Kurata, Ayumu Kido, Hirotsugu Nakai, Hiroaki Takahashi, Jordan LeGout, Adam Froemming, Akira Kawashima, Abhinav Khanna, Takahashi Naoki

Published 2026-07-29

📖 3 min read☕ Coffee break read

CC BY 4.0

Original authors: Sabarish Narayanasamy, Yasuhisa Kurata, Ayumu Kido, Hirotsugu Nakai, Hiroaki Takahashi, Jordan LeGout, Adam Froemming, Akira Kawashima, Abhinav Khanna, Takahashi Naoki

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are a detective trying to solve a mystery inside a man's body: Is there a sneaky, dangerous intruder hiding in the prostate gland? For years, the main clue has been a chemical called PSA (Prostate-Specific Antigen). Think of PSA as a smoke alarm. When it goes off, it means something is burning, but it doesn't tell you what is burning. Is it a tiny, harmless candle (a benign condition) or a raging forest fire (cancer)? Because the alarm is so sensitive, it often goes off for harmless reasons, leading to unnecessary panic and painful investigations (biopsies).

To get a better look, doctors use a super-powered camera called an MRI. This camera takes detailed pictures of the prostate, giving the detective a visual clue called a "PI-RADS score," which rates how suspicious a spot looks. Another helpful clue is "PSAD" (PSA Density). If you imagine the prostate as a sponge, PSAD asks: "How much smoke is there for every inch of sponge?" A high amount of smoke in a small sponge is more worrying than the same amount in a giant sponge. But what if the smoke alarm isn't just loud, but getting louder? That's where "velocity" comes in. Scientists wondered if watching how fast these numbers change over time—like checking if the smoke alarm is screaming faster and faster—could be the ultimate clue to spotting the dangerous cancer before it causes trouble.

This study, conducted by a team of researchers from Mayo Clinic and Kyoto University, set out to test a specific idea: Does the speed at which PSA Density changes (called PSAD-velocity) help doctors predict if a patient has clinically significant prostate cancer (csPCa)? They gathered data from over 1,000 men who had at least two MRI scans over a period of time. They compared the men who turned out to have dangerous cancer against those who didn't, looking at their age, their MRI scores, their static PSA numbers, and how fast those numbers were moving.

The results were a bit like finding a clue that looked promising at first glance but turned out to be a red herring. When the researchers looked at the data one variable at a time, they found that men with faster-rising PSAD-velocity were indeed more likely to have cancer. It seemed like a useful hint. However, the story changed when they put all the clues together in a big puzzle. Once they accounted for the other, stronger clues—like the patient's age, the specific MRI score (PI-RADS), and the current PSA Density—the "speed" of the change (PSAD-velocity) stopped being a significant predictor.

In simple terms, the study found that while a fast-rising PSAD-velocity is associated with cancer, it doesn't offer any new or independent information that the other tools don't already provide. The other factors, particularly the MRI score and the current PSA Density, were so powerful that they overshadowed the velocity metric. The researchers concluded that adding PSAD-velocity to the mix doesn't help doctors make better decisions about who needs a biopsy. It's like having a detective who checks the speed of the smoke alarm; while it's interesting, the detective already knows the answer just by looking at the size of the fire and the type of building. The study suggests that sticking to the established tools—MRI scores and current density—is the most reliable way to catch the dangerous cancer without causing unnecessary worry.

Technical Summary: Utility of PSA Density Velocity in Diagnosing Prostate Cancer on Prostate MRI

Problem Statement
Prostate cancer remains a leading malignancy in men, yet the specificity of Prostate-Specific Antigen (PSA) screening for detecting clinically significant prostate cancer (csPCa) is limited. While multiparametric MRI (mpMRI) and the PI-RADS scoring system have become standard for diagnosis, there is a need to refine risk prediction models to reduce unnecessary biopsies and overdiagnosis. Prostate-Specific Antigen Density (PSAD) has emerged as a valuable complementary metric to MRI, particularly for equivocal lesions. However, the utility of dynamic changes in these metrics over time—specifically PSAD-velocity (the rate of change in PSAD)—remains unexplored. Unlike static measurements, PSAD-velocity accounts for temporal trends, potentially offering a more nuanced predictor of csPCa. This study investigates whether PSAD-velocity serves as an independent predictor of csPCa in patients undergoing prostate MRI, comparing its efficacy against established markers such as PI-RADS score, patient age, PSA, and static PSAD.

Methodology
This was a three-center, retrospective observational study approved by the institutional review board (IRB# 23-008038), covering data from 2017 to 2023.

Cohort Selection: From 23,244 prostate MR exams, the study identified 1,174 unique patients who met strict inclusion criteria:
- Underwent two or more MR exams.
- Had a "baseline" exam performed 0.3 to 2 years prior to an "index" exam.
- Underwent pathological confirmation (biopsy) within one year of the index exam.
- Patients with known csPCa at the time of the MR were excluded; the cohort consisted of screening patients and those on active surveillance (AS) for Gleason score 6 cancer.
Data Collection: Variables extracted included PI-RADS score (v2.0/v2.1), patient age, indication (screening vs. AS), PSA, prostate volume, and derived metrics (PSAD, PSA-velocity, PSAD-velocity).
Calculations:
- Prostate Volume: Calculated via automated segmentation (DynaCAD) with radiologist correction or the ellipsoid formula.
- PSAD: PSA divided by prostate volume.
- PSAD-Velocity: Defined as $(PSAD_{index} - PSAD_{baseline}) / \text{time interval}$ . For patients with multiple baseline exams, linear regression was used to determine the slope.
- PSA-Velocity: Calculated similarly from PSA values.
Pathological Confirmation: csPCa was defined as a Gleason Score (GS) of 7 or higher. Biopsies utilized MRI-ultrasound fusion for PI-RADS 3+ lesions and systematic cores.
Statistical Analysis: Patients were stratified into csPCa and non-csPCa groups. Univariate analysis (t-tests, chi-square) assessed individual variable associations. Significant variables ( $p < 0.05$ ) were entered into a multivariate logistic regression model to identify independent predictors.

Key Results

Cohort Demographics: Of the 1,174 patients (mean age 65.9), 550 (47%) were diagnosed with csPCa. The median interval between baseline and index MR was 1.1 years.
Univariate Analysis: Several factors were significantly associated with csPCa:
- PI-RADS Score: Strongly predictive ( $p < .0001$ ).
- Age: Higher in csPCa group (67.0 vs. 64.9 years, $p < .0001$ ).
- Indication: Active surveillance exams were more common in the csPCa group (67% vs. 59%, $p = 0.004$ ).
- PSA: Higher in csPCa group (7.5 vs. 6.6 ng/mL, $p < .001$ ).
- PSAD: Significantly elevated in csPCa group (0.16 vs. 0.12 ng/mL/mL, $p < .0001$ ).
- PSAD-Velocity: Higher in csPCa group (0.023 vs. 0.012 ng/mL/mL/year, $p = 0.005$ ).
- PSA-Velocity: Did not reach statistical significance ( $p = .4$ ).
Multivariate Analysis: When controlling for other variables, PI-RADS score, age, AS indication, PSA, and static PSAD remained significant independent predictors ( $p < .0001$ for all). PSAD-velocity lost statistical significance ( $p = .1$ ) in the multivariate model.

Key Contributions and Claims
The primary contribution of this work is the first evaluation of PSAD-velocity as a predictor of csPCa in the context of prostate MRI. The study claims the following:

Association vs. Independence: While a higher PSAD-velocity is associated with the presence of csPCa in isolation (univariate analysis), it does not function as an independent predictor when combined with established markers like PI-RADS score, age, and static PSAD.
Redundancy of Dynamic Metrics: The lack of independent significance for PSAD-velocity suggests that the predictive information it contains is largely captured by the static PSAD and PSA values. The study posits that PSAD-velocity is inherently correlated with these stronger predictors, offering little additional diagnostic value in a multivariate context.
Measurement Variability: The authors suggest that the diminished significance of PSAD-velocity may be exacerbated by error propagation. Since PSAD-velocity relies on the difference between two measurements (PSA and volume) over time, uncertainties from biological fluctuations (e.g., prostatitis, sexual activity) and measurement variability (e.g., differences between ellipsoid formula and segmentation software) accumulate, potentially weakening its predictive power compared to single-timepoint metrics.
Validation of Static PSAD: The study reinforces the utility of static PSAD, confirming it as a robust independent predictor that complements PI-RADS scoring, consistent with prior literature.

Significance and Limitations
The paper concludes that PSAD-velocity does not offer added diagnostic information over static PSAD for predicting csPCa. Consequently, it may not be necessary to incorporate PSAD-velocity into routine risk assessment models for biopsy decisions when PI-RADS, age, and static PSAD are already available.

The authors acknowledge several limitations:

Retrospective Design: Inherent biases in data collection.
Timepoint Restriction: PSA and velocity calculations were based solely on time points coinciding with MR exams, potentially missing broader temporal trends.
Measurement Variability: The study could not fully assess inter-observer variability in prostate volume measurements, which may impact the precision of velocity calculations.
Selection Bias: Despite being multi-center, the study design may introduce selection bias.

Ultimately, the study provides a modest but clear finding: while dynamic changes in PSAD correlate with cancer presence, they do not independently improve diagnostic accuracy beyond what is already provided by standard clinical and imaging markers.

Technical Summary: Utility of PSA Density Velocity in Diagnosing Prostate Cancer on Prostate MRI

More like this