A total of 110 patients who underwent transrectal ultrasound-guided PBx were grouped based on histological finding of the biopsy into BPH (n=35 patients) and PCa (n = 75). The mean age of BPH cohort was 70 years and 67 years for PCa. About half of the total participants had normal BMI: BPH (51.8%) and PCa (47.4%). The majority of the patients recruited for the study were hypertensive (75% of BPH, and 80% of PCa).
The mean serum PSA of the PCa group (176.2 ± 53.44 ng/ml) was significantly higher than the BPH group (22.55 ± 4.08 ng/ml). Except for haemoglobin (Hb) concentration which was significantly lower in the PCa group compared with the BPH group, there were no significant differences in the full blood count differentials between the two groups (Tab. 1). Furthermore, the differences in the systemic negative acute inflammatory marker, Albumin (ALB) was significantly lower in the PCa compared with the BPH group. On the other hand, higher mean values of the positive acute inflammatory marker, C-reactive protein, was recorded in the PCa compared with BPH group indicating a differential response of these systemic inflammatory marker responses to BPH and PCa (Tab. 1).
In light of the significant differences observed in serum ALB, Hb concentration and CRP between PCa and BPH groups, further analyses were carried out to investigate ratios involving PSA and the three biomarkers to obtain threshold values that differentiate significantly between PCa and BPH.
Here we report on Ratios involving circulating negative and positive acute inflammatory proteins for prognostic threshold, such that the outcome is effectively stratified to differentiate between BPH and PCa. Below in table 2 are findings on the ratios with potential utility in predicting PCa and BPH. The mean PSA-to-ALB ratio was significantly lower in the BPH (0.53 ± 0.10) group compared with the PCa group (2.867 ± 0.60) at p ˂ 0.0001. Similarly, mean PSA-to-Hb ratio was significantly lower in the BPH group (1.75 ± 0.31) compared with the PCa group (11.41 ± 2.91) at p ˂ 0.0001. Finally, PSA-to-CRP values were significantly lower in the BPH group (181.0 ± 38.81) compared with the PCa group (435.7 ± 86.03) at p-value = 0.014. The Receiver Operating Characteristics (ROC) curve analyses were carried out to assess the ability of the ratios to differentially predict PCa or BPH based on sensitivity, specificity, positive predictive value and negative predictive values. Therefore, based on the Area Under the ROC (AUROC) curves analyses, the cut-off points of PSA-to-ALB, PSA-to-CRP and PSA-to-Hb ratios were 1.00, 250 and 2.5 respectively (Tab. 3a). Multivariate analysis showed that PSA-to-CRP (odds ratio (OR) = 2.01 (0.80 - 5.03), PSA-to-Hb (OR = 5.87 (2.27 - 15.16)) and PSA-to-ALB ratio (OR = 12.86 (3.62 - 45.71)) were independent predictors of PCa (Table 3b). The sensitivities, specificities, positive predictive values, and negative predictive values using PSA/CRP cut-off, PSA/Hb cut-off, and PSA/Alb cut-off are shown in Table 3b. Specificity and positive predictive value of PSA-to-ALB ratio was the highest (93% and 91% respectively), followed by PSA-to-Hb ratio (86 %, 80 % respectively) and the least was PSA-to-CRP (78 % and 77 % respectively) (Table 3b).
Analysis of PSA-to-ALB ratio resulted in much significant disparity between PCa and BPH patients with specificity of 93% (81 - 98) and positive predictive value of 91% (77 - 98) at a cut-off of PSA/ALB ≥ 1.0 compared with either PSA alone with specificity of 55% (43 - 68) and PPV of 17% (7 - 34) or Albumin alone with specificity of 61% (49 - 72) and PPV of 17% (7 - 34) from our studies (Tab. 3b). Similarly, a ratio of PSA-to-Hb concentration resulted in much larger discriminating values of specificity 86% (74 - 94) and PPV of 80% (61 - 92) at a cut-off of 2.5 for PCa prediction compared with the individual variables which showed relatively less specificity and PPV (Tab. 3b). Furthermore, our assessment of PSA-to-CRP ratio revealed much significant discriminating values with higher specificity and PPV compared with the individual measured variables. Therefore, it can be concluded that PSA-to-ALB, PSA-to-Hb and PSA-to-CRP resulted in higher disparities between PCa and BPH such that they were effective in stratifying PCa from BPH in our cohort of patients.
Table 1. Comparison of serum markers and Full blood count between BPH and PCa groups
Table 1: Comparison of serum PSA, Full blood count and selected serum inflammatory markers between benign prostate hyperplasia (BPH) group and prostate cancer (PCa) group. Results are mean ± SD, * signifies p-value ≤ 0.05.
Table 2. A comparison of ratios of serum PSA to selected systemic inflammatory markers between Benign prostate hyperplasia (BPH) and Prostate cancer (PCa).
PSA/Blood Inflammation Marker Ratio
|
BPH
|
95% CI
|
PCa
|
95% CI
|
P- value
|
PSA/ALB
|
0.53 ± 0.10
|
0.33 – 0.73
|
2.867 ± 0.60
|
1.67 – 4.07
|
< 0.0001
|
PSA/CRP
|
181.0 ± 38.81
|
102.1 – 259.9
|
435.7 ± 86.03
|
264.3 – 607.1
|
0.0145
|
PSA/Hb
|
1.75 ± 0.31
|
1.13 – 2.42
|
11.41 ± 2.91
|
6.94 -16.93
|
< 0.0001
|
Table 2: A comparison of ratios of serum PSA to selected systemic inflammatory markers between Benign prostate hyperplasia (BPH) and Prostate cancer (PCa). PSA/CRP ratio, serum prostate-specific antigen to C-reactive protein ratio; PSA/ALB – serum prostate-specific antigen to serum Albumin ratio; PSA/Hb – serum prostate-specific antigen to Haemoglobin ratio. PCa (n= 75), BPH (n=35). Results are mean + SD, * signifies p-value ≤ 0.05.
Table 3a. AUROC for variables and PSA-ratios predicting Prostate cancer Biopsy.
Variable
|
Area (± SEM)
|
95% CI
|
P value
|
PSA
|
0.67 (±0.05)
|
0.57 - 0.77
|
0.004
|
|
|
|
|
ALB
|
0.68 (±0.04)
|
0.58 - 0.78
|
0.003
|
CRP
|
0.69 (±0.05)
|
0.59 - 0.80
|
0.001
|
Hb
|
0.69 (±0.05)
|
0.59 - 0.80
|
0.001
|
|
|
|
|
Ratio
|
Area (± SEM)
|
95% CI
|
P value
|
PSA/ALB
|
0.7667 (± 0.05)
|
0.68 - 0.86
|
< 0.0001
|
PSA/CRP
|
0.6453 (± 0.058)
|
0.53 - 0.76
|
0.014
|
PSA/Hb
|
0.7794 (± 0.04)
|
0.69 - 0.87
|
< 0.0001
|
Table 3a: Data on the Area Under the Receiver Operating characteristic curves (AUROC) for PSA, ALB, CRP and Hb compared with PSA/CRP, PSA/Hb and PSA/ALB ratios to predict PCa. PSA/Hb and PSA/ALB ratios resulted in higher AUROC than the individual variables. PSA - prostate-specific antigen, ALB - Albumin, CRP - C-reactive protein and Hb - haemoglobin concentration
Table 3b. An AUROC analysis and unconditional logistic regression analyses on individual variable and their ratio with PSA to predict prostate cancer biopsy.
Table 3b: AUROC analysis on individual variable and their ratio with PSA to predict prostate cancer biopsy in the study cohort. Unconditional logistic regression analyses assessing the association of individual variables and ratios with Prostate cancer. These studies involved PCa (n=75) cases and BPH (n=35) cases. The specificities, positive predictive values and other predictive values of the ratios showed higher predictive values compared with that of the individual variables. Results are shown as predictive value (95% CI).