Normalized periprostatic adipose tissue thickness: an imaging marker associated with prostate biopsy outcomes among patients with PI-RADS and PSA double gray zone

ABSTRACT

Periprostatic adipose tissue (PPAT) is a potential factor closely associated with prostate cancer (PCa) development. This study aimed to introduce normalized PPAT thickness, a novel imaging biomarker, as a PCa predictor for patients within the diagnostic ‘double gray zone’, defined as the combination of Prostate Imaging Reporting and Data System (PI-RADS) score 3 lesions and serum prostate-specific antigen (PSA) levels of 4–10 ng/mL. A total of 219 patients were retrospectively enrolled. PPAT thickness was measured on pre-biopsy MRI. Pearson correlation analysis was performed to assess the relationship between BMI and PPAT thickness. Independent predictors of PCa were investigated by logistic regression analysis. Normalized PPAT thickness was defined as the ratio of PPAT thickness to prostate volume, and its predictive performance was compared with PSA density (PSAD). Restricted cubic spline analysis was performed to determine its optimal threshold for PCa prediction. A negative correlation was observed between PPAT thickness and BMI (ρ = -0.154, p = 0.023), suggesting that PPAT is less affected by overall obesity. PPAT thickness was significantly higher in PCa patients (0.52 vs. 0.36 cm, p < 0.001) and was identified as an independent PCa predictor (OR 1.523, 95% CI 1.252–1.853, p < 0.001). Normalized PPAT thickness outperformed PSAD for predicting clinically significant PCa (csPCa) (AUC 0.819 vs. 0.690, p = 0.003), and its threshold of 14 outperformed the traditional PSAD threshold > 0.15 (csPCa detection rate 39.2% vs. 21.7%). In conclusion, we proposed normalized PPAT thickness as a novel PCa predictor in the diagnostically challenging ‘double gray zone’ cohort.

Introduction

The potential role of adipose tissue as a tumour-promoting factor has attracted considerable research attention in recent years [1]. Several studies have demonstrated that obesity is a risk factor for colorectal, ovarian, and breast cancers [2–4]. This phenomenon may be driven by a combination of systemic metabolic dysregulation and a chronic inflammatory environment [5]. Multiple adipokines secreted by adipose tissue, such as leptin, resistin, and visfatin, have been implicated as potential mediators of carcinogenesis and cancer progression [6–8]. In addition, intentional weight loss has been shown to reduce cancer risk, providing additional evidence for the tumorigenic effect of obesity [9].

However, the relationship between obesity and prostate cancer (PCa), the second most prevalent male malignancy worldwide, remains controversial [10,11]. Although higher BMI predicts earlier biochemical recurrence after radical prostatectomy, paradoxically, it is associated with prolonged overall survival of metastatic PCa patients [12,13]. This contradiction may result from the inability of BMI to reflect the heterogeneous distribution and functional diversity of adipose tissue depots. The mechanisms through which adipose tissue influences PCa progression remain unclear.

Periprostatic adipose tissue (PPAT), roughly defined as the adipose tissue surrounding the prostate gland, has received increasing attention for its role in PCa development [14]. Previous studies have measured the thickness of adipose tissue anterior to the prostate gland and posterior to the symphysis pubis, and found that this thickness was positively associated with a more aggressive PCa phenotype and a poorer prognosis [15–18]. As a readily measured imaging marker, PPAT thickness holds promise for pretreatment risk stratification and patient evaluation in PCa.

Currently, PCa screening relies heavily on MRI and prostate-specific antigen (PSA) testing [19,20]. This dependence may lead to a critical challenge when patients fall into the ‘gray zone’ of both MRI and PSA results, which was defined as a combination of Prostate Imaging Reporting and Data System (PI-RADS) score 3 and serum PSA 4–10 ng/mL [21,22]. After adjusting for prostate volume (PV), PSA density (PSAD) outperforms PSA alone in detecting PCa [23–25]. However, the key threshold of PSAD > 0.15 for biopsy can still risk overlooking 15%–30% of clinically significant PCa (csPCa), thereby underscoring the need for complementary biomarkers [26]. Our team has previously reported that PPAT thickness independently predicted PCa in patients receiving systematic prostate biopsy and that a smaller PV also served as a strong PCa predictor [27]. This prompted us to develop a novel imaging biomarker by integrating PPAT thickness to PV and explore its value for PCa prediction.

In this study, we introduced the concept of normalized PPAT thickness (nPPAT thickness), defined as the ratio of PPAT thickness to PV, for predicting PCa in patients within the diagnostically challenging ‘double gray zone’. We also explored its optimal cut-off value to refine biopsy selection criteria in this diagnostically equivocal population.

Materials and methods

Patient selection and data collection

This study builds upon our preliminary findings, which were previously communicated as a conference paper [28]. We retrospectively analysed the records of patients who underwent transperineal ultrasonography-guided systematic prostate biopsy at Beijing Friendship Hospital between March 2016 and September 2025 (shown in Figure 1). Biopsy decision-making was based on a combination of clinical factors, including patient age, serial PSA changes, digital rectal examination findings, and patient preference. All patients underwent pre-biopsy serum PSA testing within the ‘gray zone’, which was defined as 4–10 ng/ml, and presented with PI-RADS category 3 lesions assigned by multiparametric MRI (mpMRI).

Figure 1.

Patient inclusion and exclusion criteria.

ISUP: International Society of Urological Pathology; PCa: prostate cancer; ciPCa: clinically insignifcant prostate cancer; csPCa: clinically signifcant prostate cancer; PI-RADS: Prostate Imaging Reporting and Data System; PSA: prostate-specific antigen.

The following exclusion criteria were applied: (1) incomplete clinical data; (2) conditions that influenced PSA, including acute urinary tract infections, acute prostatitis, or indwelling catheters; (3) poor mpMRI image quality that precluded determination of lesion volume or PI-RADS v2.1 score; and (4) prior surgical treatment for benign prostate hyperplasia.

All prostate biopsies were performed by urologists with over 5 years of experience, obtaining at least 10 systematic cores per patient under transperineal ultrasonography guidance. The highest Gleason score among all biopsy cores was recorded as the patient’s biopsy Gleason score and was stratified into International Society of Urological Pathology (ISUP) grade groups [29]. If a patient underwent repeated biopsies, only data from the most recent procedure were entered into the analysis. Based on pathological results, patients were categorized into the PCa group and non-PCa group. The PCa group was further divided into the csPCa group (ISUP grade group ≥ 2) and the clinically insignificant PCa (ciPCa) group (ISUP grade group 1). The following variables were collected: age, BMI, the latest serum PSA levels, free-to-total PSA ratio (f/t ratio) and PSAD.

This retrospective study was conducted in accordance with the Declaration of Helsinki (as revised in 2013) and was approved by the Institutional Review Board of Beijing Friendship Hospital, Capital Medical University (NO. 2023-P2-329), which waived the requirement for informed consent.

Image analysis

All patients underwent standard mpMRI at our institution within three months prior to biopsy. MRI examinations were conducted using a 3.0-Tesla scanner with a phased-array abdominal coil. All lesions classified as PI-RADS 3 were subjected to a two-stage re-evaluation process: first assessed independently by a radiologist with five years of experience in genitourinary imaging, and subsequently confirmed by a senior radiologist with 10 years of expertise in urological imaging, adhering to the PI-RADS v2.1 guidelines [19].

In accordance with the method of our previous study, PPAT thickness was defined and measured as the shortest perpendicular distance from the pubic symphysis to the prostate on a single midsagittal T2WI (indicated by the red line in Figure 2(A)) [27]. On the same slice, subcutaneous fat thickness was determined as the shortest perpendicular distance from the pubic symphysis to the skin (indicated by the blue line Figure 2(A)). All measurements of adipose tissue thicknesses were performed by two urologists with at least five years of experience. All radiologists and urologists involved in image analysis were blinded to the clinical and pathological information, including serum PSA levels and biopsy results during the entire evaluation process. The average of the measurements was used in the statistical analysis.

Figure 2.

Analysis of PPAT thickness. (A) Measurement of PPAT thickness (marked in red) and subcutaneous fat thickness (marked in blue) on a single midsagittal T2-weighted imaging. (B) Pearson correlation analyses between PPAT thickness and subcutaneous fat thickness with BMI. (C–E) Adipose tissue thicknesses and BMI across study groups. (F–G) Distribution of overall PCa and csPca across different intervals of (F) PPAT thickness and (G) PSA. (H–I) ROC curves for PCa prediction. PPAT thickness yielded significantly higher discriminatory power than PSA for both (H) overall PCa (AUC 0.741 vs. 0.525, p < 0.001) and (I) csPca (AUC 0.751 vs. 0.517, p = 0.002).

*p < 0.05.AUC: area under curve; PCa: prostate cancer; ciPCa: clinically insignifcant prostate cancer; csPCa: clinically signifcant prostate cancer; PPAT: periprostatic adipose tissue; PSA: prostate-specific antigen; ROC: receiver operating characteristic.

PV was calculated according to the traditional prolate ellipse formula, where volume = 0.52 × longitudinal diameter (on sagittal T2WI) × anteroposterior diameter (on sagittal T2WI) × transverse diameter (on axial T2WI) [30]. PSAD was then calculated as the ratio of serum PSA level to PV. Similarly, normalized PV (nPV = 1000 cm³ / PV) and nPPAT thickness (PPAT thickness × 1000 cm² / PV) were computed to correct for PV.

Statistical analysis

The normality of continuous variables was assessed using the Kolmogorov-Smirnov test. Normally distributed numerical variables were expressed as mean values with standard deviations (SD) and compared using student t test. Non-normally distributed continuous variables were expressed as medians with interquartile ranges (IRQ) and compared using the Mann-Whitney U test. The inter-observer reliability of adipose tissue thickness measurements was evaluated using intraclass correlation coefficient (ICC), categorized as poor (<0.5), moderate (0.5–0.75), good (0.75–0.9), and excellent (≥0.9). The Pearson correlation coefficient (ρ) was employed to assess the relationships of BMI with PPAT thickness and subcutaneous fat thickness.

Sample size calculation was performed based on our preliminary study data, which showed a mean difference of 0.17 cm in PPAT thickness between csPCa and non-csPCa patients with a standard deviation of 0.18 cm. Using a two-tailed Student’s t-test with a 1:7 case-to-control ratio, α = 0.05 and power = 0.90, a minimum of 85 patients was required.

Receiver operating characteristic (ROC) curve analysis was performed to analyse the predictive values of different factors for overall PCa and csPCa, and the areas under curve (AUCs) were compared using DeLong test. Logistic regression analyses were used to identify independent predictors. Variables with p < 0.1 in the univariate analysis were further included in the multivariate analysis.

The improvement in using nPPAT thickness against PSAD for predicting overall PCa and csPCa was evaluated by calculating the net reclassification improvement (NRI), defined as the sum of gains in sensitivity and specificity for given risk thresholds: ≤0.1 for low risk, 0.1–0.3 for intermediate risk and >0.3 for high risk. Decision curve analysis was performed to assess the net benefits of both predictors. The optimal threshold of nPPAT thickness for predicting csPCa was determined using two complementary methods: (1) the Youden index, which maximizes the sum of sensitivity and specificity, and (2) the maximum slope of the restricted cubic spline curve, which identifies the point where the rate of cancer risk is highest. Sankey plots were employed to visually demonstrate the flow paths and proportions of patients from different subgroups to final biopsy outcomes.

SPSS 26.0, GraphPad Prism Version 9.0 and R 4.4.3 were utilized for statistical analysis. All statistical tests were two-tailed, and p < 0.05 was considered significant for all parameters.

Results

Patient characteristics

A total of 219 patients meeting the inclusion and exclusion criteria (detailed in Figure 1) were enrolled in this study. The baseline characteristics of all patients were shown in Table 1 and Supplementary Table S1. Patients with malignant pathological results (n = 70) were included in the PCa group, and those with a Gleason score more than 6 (ISUP grade group ≥ 2) (n = 29) were included in the csPCa group. The study population had a median PSA level of 6.59 ng/ml, a median prostate volume of 48.80 cm³, and a median PSAD of 0.131. Compared with the non-PCa group, PSA level showed no significant elevation in the PCa group (6.87 vs. 6.64 ng/mL, p = 0.525). Similarly, no significant difference in PSA level was observed between patients with csPCa and those without (6.88 vs. 6.60 ng/mL, p = 0.765).

Table 1.

Comparison of baseline characteristics.

Variables	Total (n = 219)	Non-PCa group (n = 149)	PCa group (n = 70)	p a	non-PCa + ciPCa group (n = 190)		csPCa group (n = 29)		pb
Age (year), median (IRQ)	68 (62–72)	67 (62–71)	69 (65–74)	0.028	68 (63–72)		68 (61–74)		0.896
BMI (kg/m2), median (IRQ)	24.43 ± 3.00	24.48 ± 3.01	24.32 ± 3.00	0.712	24.33 ± 3.00		25.11 ± 2.96		0.189
PV (cm3), median (IRQ)	48.80 (36.94–64.58)	54.92 (42.06–71.06)	40.38 (28.83–53.45)	<0.001	51.61 (40.72–68.35s)		33.42 (24.93–45.89)		<0.001
nPVc	20.49 (15.48–27.07)	18.21 (14.07–23.77)	24.77 (18.71–34.69)	<0.001	19.38 (14.63–24.56)		29.93 (21.43–40.35)		<0.001
PSA (ng/mL), median (IRQ)	6.59 (5.23–8.00)	6.64 (5.17–7.94)	6.87 (5.23–8.17)	0.525	6.60 (5.20–7.98)		6.88 (5.48–8.28)		0.765
PSAD, median (IRQ)	0.131 (0.096–0.181)	0.122 (0.090–0.160)	0.170 (0.106–0.229)	<0.001	0.129 (0.092–0.174)		0.175 (0.116–0.273)		0.001
f/t ratio (%), median (IRQ)	17.72 (12.42–22.30)	18.97 (14.27–23.52)	15.05 (10.73–20.16)	0.001	18.09 (12.52–22.89)		15.59 (11.68–19.11)		0.076
PPAT thickness (cm), mean ± SD	0.41 ± 0.19	0.36 ± 0.17	0.52 ± 0.17	<0.001	0.39 ± 0.18		0.56 ± 0.18		<0.001
nPPAT thicknessd	7.75 (4.89–13.60)	5.80 (4.42–10.68)	13.56 (8.39–20.41)	<0.001	6.49 (4.68–11.48)		19.19 (10.88–22.34)		<0.001
Subcutaneous fat thickness (cm), mean ± SD	2.93 ± 1.08	2.91 ± 1.10	2.98 ± 1.05	0.683	2.91 ± 1.08		3.06 ± 1.13		0.511

IQR: interquartile range; PCa: prostate cancer; ciPCa: clinically insignifcant prostate cancer; csPCa: clinically signifcant prostate cancer; f/t ratio: free-to-total PSA ratio; PPAT: periprostatic adipose tissue; nPPAT: normalized periprostatic adipose tissue; PSA: prostate-specific antigen; PSAD: PSA density; PV: prostate volume; nPV: normalized prostate volume.

^aThe p-values were compared between non-PCa group and PCa group.

^bThe p-values were compared between non-PCa + ciPCa group and csPCa group.

^cCalculated using the formula: nPV = 1000 cm³ / PV.

^dCalculated using the formula: nPPAT thickness = PPAT thickness × 1000 cm³ / PV.

Bold numbers indicate statistically significant p-values (p < 0.05).

Adipose tissue measurements

The mean PPAT and subcutaneous fat thicknesses were 0.41 cm and 2.93 cm, respectively. The assessment of inter-observer reliability showed an ICC of 0.969 for PPAT thickness and 0.947 for subcutaneous fat thickness, which indicated excellent inter-observer reproducibility. Subcutaneous fat thickness demonstrated a positive correlation with BMI (ρ = 0.695, p < 0.001). However, a negative correlation was observed between PPAT thickness and BMI (ρ = −0.154, p = 0.023) (shown in Figure 2(B)). Moreover, PPAT thickness, but neither subcutaneous fat thickness nor BMI, was significantly elevated in both ciPCa and csPCa patients (Table 1 and Figure 2(C–E)), highlighting the unique property of PPAT.

PPAT thickness outperforms PSA in predictive accuracy

The proportions of overall PCa and csPCa exhibited an upward trend with increasing PPAT thickness (Figure 2(F)), with 60.5% (23/38) of overall PCa and 31.6% (12/38) of csPCa in the high interval (PPAT thickness > 0.6 cm). This uptrend was more evident than that observed for PSA (Figure 2(G)). When applied separately, PPAT thickness yielded significantly higher predictive power for both overall PCa (AUC 0.741 vs. 0.525, p < 0.001) and csPCa (AUC 0.751 vs. 0.517, p = 0.002) than PSA (shown in Table 2 and Figure 2(H,I)). Logistic regression analyses (Table 3) showed PPAT thickness could independently predict overall PCa (OR 1.523, 95% CI 1.252–1.853, p < 0.001) and csPCa (OR 1.492, 95% CI 1.133–1.965, p = 0.004). Collectively, these results suggested PPAT thickness outperformed PSA level in PCa prediction among patients within ‘double gray zone’.

Table 2.

ROC analysis of PSA and PPAT thickness for predicting overall PCa and csPca.

Parameters	AUC	95% CI	pa
Overall PCa
PSA level	0.525	0.443–0.611	Reference
PPAT thickness	0.741	0.673–0.809	<0.001
csPCa
PSA level	0.517	0.406–0.629	Reference
PPAT thickness	0.751	0.659–0.843	0.002

AUC: area under curve; CI: confidence interval; PCa: prostate cancer; csPCa: clinically significant PCa; PPAT: periprostatic adipose tissue; PSA: prostate-specific antigen.

^aThe p-values were calculated using DeLong test.

Table 3.

Univariate and multivariate logistic regression analyses for prediction of overall PCa and csPca.

	Prediction of overall PCa				Prediction of csPCa
	Univariate analysis		Multivariate analysis		Univariate analysis		Multivariate analysis
Variables	OR (95% CI)	p	OR (95% CI)	p	OR (95% CI)	p	OR (95% CI)	p
Age (year)	1.054 (1.012–1.098)	0.012	1.074 (1.024–1.127)	0.003	1.014 (0.959–1.071)	0.630
BMI (kg/m2)	0.982 (0.893–1.080)	0.710			1.068 (0.932–1.225)	0.343
PSA (ng/mL)	1.065 (0.897–1.264)	0.471			1.108 (0.870–1.411)	0.404
f/t ratio (%)	0.942 (0.904–0.981)	0.004	0.949 (0.907–0.994)	0.027	0.950 (0.897–1.007)	0.084	1.000 (0.943–1.061)	0.995
PV (cm3)	0.971 (0.957–0.987)	<0.001	0.981 (0.965–0.997)	0.021	0.944 (0.916–0.972)	<0.001	0.954 (0.927–0.982)	0.002
PPAT thickness (cm)	1.650 (1.372–1.983)	<0.001	1.523 (1.252–1.853)	<0.001	1.577 (1.243–1.999)	<0.001	1.492 (1.133–1.965)	0.004
Subcutaneous fat thickness (cm)	1.006 (0.979–1.032)	0.681			1.010 (0.973–1.048)	0.607

OR: odds ratio; CI: confidence interval; PCa: prostate cancer; BMI: body mass index; csPCa: clinically significant PCa; f/t ratio: free-to-total PSA ratio; PPAT: periprostatic adipose tissue; PSA: prostate-specific antigen; PV : prostate volume.

p < 0.1 in the univariate analysis and p < 0.05 in the multivariate analysis are indicated by boldface.

Analysis of nPPAT thickness

Considering the limited performance of PSA, the predictive value of PSAD for PCa was reassessed against that of the nPV. Notably, ROC analysis demonstrated that PSAD yielded an AUC comparable to that of nPV for overall PCa (AUC 0.667 vs. 0.700, p = 0.111), and a lower AUC for csPCa (AUC 0.690 vs. 0.756, p = 0.002) (Table 4 and Figure 3(A,B)). Based on these results, we introduced nPPAT thickness, defined as the ratio of PPAT thickness to PV, to predict PCa in ‘double gray zone’ patients. ROC analysis demonstrated that nPPAT yielded a significantly higher AUC than PSAD for both overall PCa (AUC 0.774 vs. 0.667, p = 0.001) and csPCa (AUC 0.819 vs. 0.690, p = 0.003). Decision curve analysis showed that nPPAT thickness provided higher net benefits than PSAD for predicting overall PCa at probability thresholds below 0.6, and csPCa at thresholds below 0.5 (Figure 3(C,D)). The improvement in predictive performance was further evidenced by overall NRI values of 0.181 (95% CI 0.074–0.303, p = 0.003) for overall PCa and 0.344 (95% CI 0.144–0.537, p < 0.001) for csPCa.

Table 4.

ROC analysis of PSAD, nPV and nPPAT thickness for predicting overall PCa and csPca.

Parameters	AUC	95% CI	p a
Overall PCa
PSAD	0.667	0.584–0.751	Reference
nPV	0.700	0.623–0.777	0.111
nPPAT thickness	0.774	0.706–0.841	0.001
csPCa
PSAD	0.690	0.577–0.802	Reference
nPV	0.756	0.659–0.854	0.002
nPPAT thickness	0.819	0.738–0.900	0.003

AUC: area under curve; CI: confidence interval; PCa: prostate cancer; csPCa: clinically significant PCa; nPPAT: periprostatic adipose tissue; PSAD: prostate-specific antigen density; nPV: normalized prostate volume.

^aThe p-values were calculated using DeLong test.

Figure 3.

Predictive value of nPPAT thickness versus PSAD for PCa prediction. (A–B) ROC curves for PCa prediction. nPPAT thickness demonstrated significantly better discrimination than PSAD both for (A) overall PCa (AUC 0.774 vs. 0.667, p = 0.001) and (B) csPca (AUC 0.819 vs. 0.690, p = 0.003). (C–D) decision curve analysis for predicting (C) overall PCa and (D) csPca. (E–F) Restricted cubic spline curves for (E) overall PCa and (F) csPca. Vertical dashed lines mark the steepest slope, indicating the proposed cutoff. (G–H) Sankey plots illustrating patients flow and biopsy outcomes in subgroups defined by (G) nPPAT thickness and (H) PSAD. (I) Risk distribution map of overall PCa and csPca for each subgroup categorized by PPAT thickness and PSAD.

AUC: area under curve; PCa: prostate cancer; ciPCa: clinically insignifcant prostate cancer; csPCa: clinically signifcant prostate cancer; nPPAT: normalized periprostatic adipose tissue; PSAD: prostate-specific antigen density; nPV: normalized prostate volume; ROC: receiver operating characteristic.

To explore the optimal cut-off of nPPAT thickness for PCa, we found the Youden index reached its maximum value at an nPPAT thickness of 14.36 (sensitivity = 0.690, specificity = 0.858) when predicting csPCa. Restricted cubic spline curves were used to depict the detection rates for overall PCa and csPCa, with the maximum slope observed at approximately nPPAT thickness = 14 (Figure 3(E,F)). As a result, a threshold of 14 was selected for nPPAT thickness. Patients were then categorized into two subgroups using this threshold, and Sankey plots were used to depict the flow of patients and their biopsy outcomes (Figure 3(G)). Compared with the widely used threshold of PSAD > 0.15 (Figure 3(H)), our nPPAT thickness-based criterion (>14) identified more csPCa (39.2% vs. 21.7%), and maintained a high avoidance rate of unnecessary biopsies (78.0% vs. 76.5%). Four distinct risk subgroups were generated by combining nPPAT thickness and PSAD, and the two subgroups defined by nPPAT thickness > 14 demonstrated obviously higher PCa detection rates than in the remaining two subgroups (Table 5 and Figure 3(I)). These results further indicated that nPPAT thickness could outperform PSAD in predicting biopsy outcomes among patients within ‘double gray zone’.

Table 5.

Risk stratification table constructed by integrating subgroups of nPPAT thickness and PSAD. The observed rates of overall PCa and csPca were calculated for each risk subgroup.

Subgroup		Biopsy outcomes
nPPAT thickness	PSAD	Overall PCa detection, n (%)	csPCa detection, n (%)
≤14	≤0.15	26/127 (20.5%)	7/127 (5.5%)
≤14	>0.15	11/41 (26.8%)	2/41 (4.9%)
>14	≤0.15	6/9 (66.7%)	4/9 (44.4%)
>14	>0.15	25/42 (59.5%)	14/42 (33.3%)

PCa: prostate cancer; csPCa: clinically significant prostate cancer; PPAT: periprostatic adipose tissue; PSAD: prostate-specific antigen density.

Discussion

In this study, we focused on PPAT, a special adipose depot, and analysed the predictive values of PPAT-based imaging markers for PCa. In the most diagnostically challenging ‘double gray zone’ cohort, which was defined by a PI-RADS score of 3 and serum PSA 4–10 ng/mL, PSA lost its conventional predictive value for PCa, whereas PSAD exhibited lower discriminative performance than PV alone. Based on these observations, we devised nPPAT thickness by integrating PPAT thickness and PV. This novel marker significantly improved PCa detection in the ‘double gray zone’. Moreover, a normalized nPPAT thickness > 14 is proposed as a complementary biomarker to PSAD > 0.15 for risk stratification in this specific population.

Despite advancements in PCa screening, persistent challenges remain in cases with ambiguous findings from both imaging and serum biomarkers. According to a meta-analysis, PI-RADS category 3 lesions demonstrate limited diagnostic accuracy, with detection rates of only 33% of overall PCa and 17% of csPCa [21]. This diagnostic uncertainty extends to patients with serum PSA levels in the 4–10 ng/mL range, among whom PCa detection rates show a marked decline [22]. These clinical ambiguities frequently result in excessive biopsy recommendations, contributing to increased risks of procedure-related complications and psychological burdens on patients [31,32].

Recent studies have identified PPAT as a metabolically active adipose tissue, distinguishing it from other adipose depots [33,34]. PPAT exhibits a chronic hypoxic state, which is associated with inflammation and fibrosis [33]. Its lipid metabolism is associated with metabolic alterations of prostate cells and can be disturbed during PCa development [34]. Several imaging features of PPAT have been explored for PCa prediction, including dynamic contrast enhancement patterns and second-order radiomic features, which reflect cancer-associated biological characteristics [35]. We observed a weak and inverse relationship between PPAT thickness and BMI, indicating that PPAT is relatively independent of overall obesity. Furthermore, PPAT thickness was significantly higher in PCa patients and showed superior predictive accuracy compared with that of PSA. These results replicated our previous report and provided additional evidence that PPAT is closely associated with PCa development [27]. Notably, PPAT thickness can be easily measured on standard MRI without specialized equipment, making it highly convenient for clinical use. However, the underlying biological mechanisms require further investigation in basic research studies.

It is noteworthy that our results suggested that PV should also be an independent predictor for both overall PCa and csPCa. Among ‘double gray zone’ patients, the traditional PSAD still retained some discriminatory ability for PCa, yet the predictive power of this derivative came from PV rather than PSA. This result was in accordance with our previous study on another cohort, suggesting that larger prostates are less likely to be related to PCa [27]. Several hypotheses have been proposed to explain the inverse correlation between PV and PCa incidence. First, the density of biopsy cores may be diluted by a larger PV, leading to potential missed detection [36,37]. Second, benign prostatic hyperplasia may generate mechanical stress fields which may restrict PCa growth [38]. This theory also explains the increased proportion of invasive PCa observed among 5-α reductase inhibitor users [39]. Third, larger prostates may cause more serious lower urinary tract symptoms and lead to more frequent physical examinations. These combined effects may contribute to the inverse correlation of PV and PCa incidence.

Previous studies have supported the combination of PPAT volume with PV for PCa assessment [40,41]. However, its application was extremely restricted by the difficulty in measuring PPAT volume. To overcome this disadvantage, we replaced PPAT volume with thickness to create nPPAT thickness. A threshold of nPPAT thickness > 14 was shown to serve as a complementary risk stratification tool to the traditional PSAD > 0.15. Our study demonstrated this new marker was a promising imaging tool to improve diagnostic accuracy in ‘double gray zone’ patients.

Our study had some limitations. First, the sample size was limited by the low clinical prevalence of ‘double gray zone’ patients, who account for only 6.6% (219/3295) of all patients who underwent biopsy. The relatively small number of csPCa events (n = 29) may limit the stability of logistic regression analyses. Second, though the predictive value of nPPAT thickness was validated in the current study, the proposed threshold was derived and tested within the same single-centre cohort, which carried a risk of overfitting. External validation in a large, multicenter independent cohort is still required before this threshold can be widely adopted in clinical practice. Third, our findings may not be applicable to other biopsy techniques, particularly MRI-guided targeted biopsy, and require further validation in cohorts undergoing this procedure. Fourth, the retrospective nature of this study carries inherent selection bias, which may not accurately reflect the real-world clinical scenarios. Further, some important clinical data, such as rectal examination and family history could not be retrieved.

In conclusion, we focused on patients within the diagnostically challenging ‘double gray zone’ of PI-RADS 3 lesions and PSA 4–10 ng/mL, and proposed nPPAT thickness > 14 as a new decision criterion. This new marker outperforms existing PCa predictors and may optimize prostate biopsy decision-making in this indeterminate population.

Funding Statement

This work was supported by National Key Clinical Specialty Project [grant numbers: 20250829] and Beijing Key Clinical Specialty Project [grant numbers: 20240930].

Disclosure statement

No potential conflict of interest was reported by the author(s).

Availability of data and materials

The data used to support the findings of this study are restricted by the Institutional Review Board of Beijing Friendship Hospital, in order to protect the patient privacy. Data are available from Yinong Niu (Email: niuyinong@mail.ccmu.edu.cn) for researchers who meet the criteria for access to confidential data.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/21623945.2026.2683754