Establishment and validation of serum lipid-based nomogram for predicting the risk of prostate cancer

Abstract

Background: This study aimed to explore the value of combined serum lipids with clinical symptoms to diagnose prostate cancer (PCa), and to develop and validate a Nomogram and prediction model to better select patients at risk of PCa for prostate biopsy.

Methods: Retrospective analysis of 548 patients who underwent prostate biopsies as a result of high serum prostate-specific antigen (PSA) levels or irregular digital rectal examinations (DRE) was conducted. The enrolled patients were randomly assigned to the training groups (n = 384, 70%) and validation groups (n = 164, 30%). To identify independent variables for PCa, serum lipids (TC, TG, HDL, LDL, apoA-1, and apoB) were taken into account in the multivariable logistic regression analyses of the training group, and established predictive models. After that, we evaluated prediction models with clinical markers using decision curves and the area under the curve (AUC). Based on training group data, a Nomogram was developed to predict PCa.

Results: 210 (54.70%) of the patients in the training group were diagnosed with PCa. Multivariate regression analysis showed that total PSA, f/tPSA, PSA density (PSAD), TG, LDL, DRE, and TRUS were independent risk predictors of PCa. A prediction model utilizing a Nomogram was constructed with a cut-off value of 0.502. The training and validation groups achieved area under the curve (AUC) values of 0.846 and 0.814 respectively. According to the decision curve analysis (DCA), the prediction model yielded optimal overall net benefits in both the training and validation groups, which is better than the optimal net benefit of PSA alone. After comparing our developed prediction model with two domestic models and PCPT-RC, we found that our prediction model exhibited significantly superior predictive performance. Furthermore, in comparison with clinical indicators, our Nomogram's ability to predict prostate cancer showed good estimation, suggesting its potential as a reliable tool for prognostication.

Conclusions: The prediction model and Nomogram, which utilize both blood lipid levels and clinical signs, demonstrated improved accuracy in predicting the risk of prostate cancer, and consequently can guide the selection of appropriate diagnostic strategies for each patient in a more personalized manner.

Keywords: LDL cholesterol; Nomograms; Prostate cancer; Serum lipid; Triglyceride.

Fig. 1

Flow diagram of study design. tPSA, free prostate-specific antigen; TG, Triglyceride; LDL, Low-density lipoprotein; DRE, digital rectal exam; TRUS, Transrectal Ultrasonography;

Fig. 2

Receiver operating characteristic curve (ROC) of Model 1 and Model 2 for predicting prostate cancer risk. The y-axis represents the true positive rate of the risk prediction, the x-axis represents the false positive rate of the risk prediction. The thick blue line represents the performance of the predictive model and the light blue dotted line represents the 95% CI in the training set of Model 1(A) and Model 2(B) and the validation set of Model 1(C) and Model 2(D)

Fig. 3

Nomogram predicts the probability of PCa.

Fig. 4

Calibration curves of the predictive prostate cancer risk nomogram. The y-axis represents actual diagnosed cases of prostate cancer, the x-axis represents the predicted risk of prostate cancer. The diagonal dotted line represents a perfect prediction by an ideal model, and the solid line represents the performance of the training set (A) and validation set (B), with the results indicating that a closer fit to the diagonal dotted line represents a better prediction

Fig. 5

Decision curve analysis for the prostate cancer risk nomogram. The y-axis measures the net benefit. The thick solid line = net benefit when all patients have no prostate cancer, the thin solid line = net benefit when all patients have prostate cancer, the solid green line = PSA, the solid red line = Model 1, solid blue line = Model 2. A from the training set, B from the validation set