Introduction

Figure. Kaplan-Meier curves for grade reclassification by baseline prostate health index (PHI) risk category (N=382). Asterisk indicates risk categories consisted of: category 1, scores 0-26.9; category 2, scores 27.0-35.9; category 3, scores 36.0-54.9; and category 4, scores >54.9.

While active surveillance (AS) for low-risk prostate cancer is now considered standard of care, how to monitor patients on AS is still very much a matter of debate.^1,2 It has become increasingly clear that surveillance protocols should be tailored to each individual patient based on their risk of upgrading during AS. For this purpose, liquid biomarkers such as the prostate health index (PHI) could play an important role. We sought to determine the clinical value of a baseline PHI at the initiation of AS and the clinical value of repeated PHI testing during AS in predicting upgrading after confirmatory biopsy.

Methods

We identified 382 AS patients from the prospectively maintained Johns Hopkins AS database with no greater than grade group (GG) 1 prostate cancer on diagnostic and confirmatory biopsy, at least 1 multiparametric MRI of the prostate, and PHI test; 241 of these patients had at least 2 PHI tests. We defined grade reclassification (GR) as identification of any amount of GG2 or higher prostate cancer during a surveillance biopsy. PHI was divided into 4 risk categories as defined by the manufacturer: category 1, scores 0-26.9; category 2, scores 27.0-35.9; category 3, scores 36.0-54.9; and category 4, scores >54.9; the categories corresponded to a 9.8%, 16.8%, 33.3%, and 50.1% risk of GG1 or higher prostate cancer, respectively. Associations between baseline PHI risk category or change in PHI risk categories over time or PSA density (PSAD) changes over time and GR were evaluated with multivariable Cox proportional hazard regression models adjusted for age, PI-RADS (Prostate Imaging Reporting & Data System) score and number of positive cores.

Results

We found that even after confirmatory biopsy, men who started AS with a high baseline PHI score had lower rates of GR-free survival. At 5 years, the GR-free rates (95% CI) were 77% (63, 86), 75% (63, 84), 42% (27, 56), and 38% (19, 57) for men with baseline PHI in risk categories 1, 2, 3, and 4, respectively (log-rank P < .001; see Figure). Similarly, the men who experienced an increase in PHI risk category during surveillance or had a persistently high PHI risk category had lower rates of GR-free survival (log-rank P = .032). On multivariable regression, baseline PHI risk category was associated with upgrading (risk category 4 [vs 1] HR 2.74, 95% CI 1.32-5.66; P = .002; model C-index 0.764), as were PHI risk category changes over time (risk category 4 [vs 1] HR 4.20, 95% CI 1.76-10.05; P = .002; C-index 0.759). Of note, models incorporating PHI as a continuous variable instead of a categorical variable with 4 risk categories yielded similar results. In a separate model with PSAD changes over time instead of PHI changes over time, PSAD was also associated with upgrading (C-index 0.733).

Conclusions

We found that baseline PHI at the initiation of AS provides clinically useful patient risk stratification for upgrading. We also found that both a baseline PHI and PHI changes over time are associated with upgrading during surveillance. Interestingly the model with baseline PHI performed better than the model with PHI changes over time, suggesting that in our relatively low risk AS cohort a baseline PHI might be sufficient for risk stratification. In addition, the model with PSAD changes over time performed almost as well as the model with PHI changes over time, suggesting that after a baseline PHI, PSAD might be sufficient for risk stratification. Yet, the model with PHI changes over time was disadvantaged by having significantly fewer data points that the one with PSAD. It is possible that in a higher-risk cohort and with more PHI data points over time, PHI would provide superior risk stratification, which we believe warrants more investigation.