Introduction:

A baseline prostate-specific antigen (PSA) in a man’s 40s can assess subsequent risk for clinically significant prostate cancer (PCa)/PCa death at a time in life when the PSA level should be low with few confounders. Combined with other risk factors, such as family history and race, a baseline PSA obtained in this population may allow for risk stratification for primary care providers (PCPs) to potentially trigger an early evaluation by a urologist.  In 2017, Duke Health developed and implemented such an algorithm as a clinical decision support tool in the Electronic Health Record (EHR) to standardize the risk assessment and notify PCPs of PSA values over age-related thresholds.  We examined the initial longitudinal outcomes of men in their forties who were assessed and found to have a baseline elevated PSA.   

Methods:

On February 1, 2017, the Duke Prostate Cancer Screening Algorithm was integrated into two areas of the EHR: as a screening reminder in Health Maintenance and as a message with the PSA results.  After obtaining IRB approval, our prospective database was queried for all patients (49,980 men) between the ages of 40-49 who had a PSA level drawn at the time of algorithm implementation.  Those with elevated PSA (> 1.5 ng/ml in the 40s) were included in the study cohort (564 patients).  Charts were reviewed for information pertaining to patient demographics, PSA results, timings of subsequent referrals, and oncologic outcomes.  Statistical analysis via T-test was used to compare mean baseline PSAs between those who did and did not receive a Urology referral.  Multivariable logistic regression was utilized to identify potential factors associated with detecting clinically significant prostate cancer in this population.  All statistics were performed with SPSS v. 27. 

Results:

The cohort contained 564 men, of whom 303 (54%) were immediately referred for suspicion of PCa based on the algorithm, while 27 had deferred referral with 25-month median delay (range 4-43).  The remainder of the cohort (n=234) were either not referred to urology or were referred for reasons unrelated to PCa suspicion.  Overall, 49 patients (8.7%) received biopsies, and 20 of these men were diagnosed with PCa. A total of 11 (55%) had Gleason Grade Group (GGG) 2 or higher, including three patients with > GGG 4.  Patients referred for PCa suspicion had significantly higher mean PSA values than those who were not (2.97 vs 1.98, p=0.001).  Multivariable analysis revealed PSA to be associated with subsequent clinically significant PCa on biopsy (OR 1.065, p<0.05), while timing of referral and family history or AA race were not (p > 0.05) (Table 2). 

Conclusion:

The Duke Health risk stratified PSA screening algorithm led to the identification of men in their forties who may benefit from further urologic evaluation, including a small portion of men harboring aggressive prostate cancer.  Interestingly, despite finding a PSA above the algorithm threshold and an EHR prompt, a significant proportion of men did not receive a referral to Urology. Nevertheless, lack of immediate referral did not predict finding higher grade disease at biopsy. Further longitudinal follow up may help determine optimal time for biopsy in men with abnormal baseline PSA and rising kinetics. Overall, these findings support the utility of obtaining a baseline PSA in men in their 40s and the need to balance coordination between PCPs and Urologists for optimal patient care. 

Funding: This work was supported by grants from the National Cancer Institute (P30CA014236) and the Duke Institute for Health Innovation (DIHI).