Low Blood Levels of LRG1 Before Radical Prostatectomy Identify Patients with High Risk of Progression to Castration-resistant Prostate Cancer

Abstract

Background: After radical prostatectomy (RP), depending on stage, up to 40% of patients with prostate cancer (PCa) will experience biochemical failure (BF). Despite salvage therapy, approximately one-third of these patients will need permanent hormone therapy (pHT) and are at risk of progression to castration-resistant PCa (CRPC). Prognostic markers herald the need for neoadjuvant, adjuvant, or multimodal treatment.

Objective: To evaluate the added value of blood LRG1 in predicting treatment failure in patients who have undergone radical prostatectomy (RP).

Design setting and participants: We quantified LRG1 in serum or plasma sampled before radical prostatectomy from patients from the Martini-Klinik (Martini; n = 423), the Danish CuPCa cohort (CuPCa; n = 182), and Oslo University Hospital (OUH; n = 145).

Outcome measurements and statistical analysis: The endpoints were BF, pHT, and CRPC. The association between LRG1 and survival outcomes was evaluated using Kaplan-Meier estimation and Cox proportional-hazards modeling. The added predictive value of LRG1 in nested models was estimated using the concordance index, time-dependent area under the receiver operating characteristic curve, and decision curve analysis.

Results and limitations: In multivariable Cox models using preoperative characteristics, LRG1 was associated with an estimated lower risk of BF in the Martini cohort (adjusted hazard ratio [aHR] 0.68, 95% confidence interval [CI] 0.52-0.90) and in the CuPCa cohort (aHR 0.47, 95% CI 0.30-0.73). Using preoperative prognostic variables, our data showed that doubling of LRG1 was also associated with a lower risk of pHT receipt in the CuPCa cohort (aHR 0.43, 95% CI 0.20-0.93) and of CRPC development in the OUH cohort (aHR 0.32, 95% CI 0.15-0.69). Similar aHR values were observed using either preoperative or postoperative variables for all endpoints.

Conclusions: PCa patients with high blood LRG1 are at lower risk of BF, pHT receipt, and progression to CRPC. Since LRG1 adds value to established prognostic models, new prognostic factor combinations including LRG1 should be considered in future studies.

Patient summary: We measured concentrations of the blood-based protein LRG1 before surgery for prostate cancer. Patients with high LRG1 levels had better disease-free survival, suggesting that LRG1 can help in predicting prognosis.

Keywords: Biomarkers; Castration resistance; Hormone treatment; LRG1; Noninvasive; Prostate cancer; Radical prostatectomy; Surgery; Treatment resistance.

Fig. 1

Study outline. LRG1 levels were determined in preoperative blood samples from men undergoing radical prostatectomy in three independent institutions. Propensity score matching of postoperative characteristics was used to generate a 1:2 case-control cohort of 423 serum samples from patients operated on at the Martini-Klinik (Martini). The endpoint for the Martini cohort was biochemical failure. Next, LRG1 in plasma from patients with postoperative high-risk, locally advanced disease or lymph node metastasis included in the Copenhagen uPAR prostate cancer (CuPCa) study was analyzed using biochemical failure and the start of permanent hormone therapy (pHT) as endpoints. Lastly, LRG1 was measured in 145 serum samples from the Prostate Biobank at Oslo University Hospital (OUH) enriched for patients starting permanent HT on prostate cancer recurrence. The primary endpoint for the OUH cohort was initiation of medication approved for castrate-resistant prostate cancer (CRPC). Figure created with BioRender.com.

Fig. 2

Kaplan-Meier biochemical failure (BF)-free survival among patients undergoing radical prostatectomy stratified by preoperative LRG1 level for (A) the Martini and (B) the CuPCa cohort. In both cases the median for the Martini cohort was used as the stratification cutpoint for low versus high LRG1. (C) Summary of multivariable Cox proportional-hazards preoperative and postoperative clinical models: adjusted hazard ratio (aHR) for BF after inclusion of LRG1 in the models, with model performance in terms of the concordance index (C index) for the Martini cohort (422 patients, 141 events) and CuPCa cohort (173 patients, 78 events). CI = confidence interval; ΔC = change in C index.

Fig. 3

(A) Kaplan-Meier survival curves for the time from surgery to permanent HT in the CuPCa cohort stratified by preoperative plasma LRG1. (B) Summary of multivariable Cox proportional-hazards preoperative and postoperative clinical models: aHR for perm.HT receipt after inclusion of LRG1 in the models, with model performance in terms of the concordance index (C index) in the CuPCa cohort. (C) Kaplan-Meier survival curves for the time from surgery to CRPC in patients stratified by preoperative serum LRG1 in the OUH cohort. (D) Summary of multivariable Cox proportional-hazards preoperative and postoperative clinical models: aHR for CRPC after inclusion of LRG1 in the models, with model performance in terms of the C index in the OUH cohort. In all cases, the median for the Martini cohort was used as the cutpoint for low versus high LRG1. aHR = adjusted hazard ratio; CRPC = castration-resistant prostate cancer; pHT = permanent hormone therapy