A novel composite index of PSA and periprostatic adipose tissue quantification for enhancing high-grade prostate cancer prediction

Abstract

Background

To explore the efficacy of combining MRI-derived quantitative data on Periprostatic adipose tissue (PPAT) with clinical biomarkers, including prostate-specific antigen (PSA), to enhance the high-grade (PCa) screening.

Methods

In a retrospective analysis, we reviewed clinical and pathological records of patients who had undergone prostate MRI between January 2020 and January 2023. Two radiologists measured PPAT metrics - subcutaneous fat thickness (SFT), periprostatic fat thickness (PPFT), periprostatic fat area (PPFA), and periprostatic fat volume (PPFV) - on T1-weighted axial images. Ratios of PPFA to prostate area (PA) (PPFA/PA) and PPFV to prostate volume (PV) (PPFV/PV) were calculated, collinearity testing was performed, and differences between groups for PPAT metrics and PSA levels were analyzed. Selected variables underwent multivariate binary logistic regression to identify independent predictors of high-grade PCa. Model performance was assessed using ROC curves and AUC.

Results

The study included 215 patients. Significant differences between high- and low-grade PCa groups were observed for PPFA, PPFA/PA, PSA, Prostate specific antigen density (PSAD) and the combined index PSA×PPFA/PA (P ≤ 0.001). Multivariate analysis identified PPFA/PA and PSA levels as independent predictors of high-grade PCa, with odds ratios (OR) of 1.011 (95% CI 1.002–1.021, P = 0.018) and 1.044 (95% CI 1.006–1.082, P = 0.022), respectively. The PSA, PSAD, PSA × PPFA/PA, and composite indicator models demonstrated strong predictive performance, with AUC values of 0.771, 0.796, 0.818, and 0.814, respectively. Among these, the PSA × PPFA/PA model showed superior performance, with an optimal cutoff value of 42.135.

Conclusions

The PSA×PPFA/PA index promises enhanced prediction of high-grade PCa, demonstrating that incorporating PPAT measurements alongside PSA improves screening efficacy and supports more informed clinical decision-making in the management of PCa.

Trial registration

Not applicable.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12894-025-01884-7.

Background

Obesity is recognized as a significant risk factor for numerous malignant diseases [1]. Increasing evidence suggests an association between obesity and adverse clinical outcomes in prostate cancer (PCa), including elevated mortality rates and diminished treatment response [2]. Obesity is generally categorized into visceral fat obesity (abdominal fat obesity) and subcutaneous fat obesity, with visceral fat exhibiting a closer relationship to metabolic disruptions associated with obesity [3]. Visceral fat tissue serves as a metabolically active endocrine organ, playing a crucial role in the tumor microenvironment [4]. Periprostatic adipose tissue (PPAT) surrounding the prostate has paracrine effects that influence tumor progression and dissemination. Studies [5, 6] indicate that adipocytes in periprostatic fat can secrete various adipokines, including leptin, tumor necrosis factor-α, and CCL-7, contributing to cancer cell proliferation, migration, and invasion. Conversely, the extracapsular extension of PCa leads to changes in the phenotype and characteristics of fat cells in the periprostatic region, such as a reduction in lipid content and adipocyte differentiation markers, an increase in inflammatory factors, and the promotion of free fatty acid release—a primary energy source for PCa. The reciprocal interaction between tumor cells and fat cells further fuels the progression of prostate cancer. Therefore, investigating PPAT is of paramount importance in comprehending the mechanistic underpinnings of tumorigenesis.

With the advancement of imaging technology and in-depth research, ultrasound, computed tomography (CT), and magnetic resonance imaging (MRI) have been employed to assess PPAT, with MRI being one of the most commonly utilized techniques [7–9]. Researchers have also explored various measurement methods to evaluate the role of PPAT in the onset and progression of PCa, ranging from one-dimensional periprostatic fat thickness (PPFT) to two-dimensional periprostatic fat area (PPFA), and progressing to three-dimensional periprostatic fat volume (PPFV). These indices have been validated as independent predictors of aggressive PCa [10–12]. Moreover, these quantitative parameters have gradually demonstrated predictive value in terms of lymph node metastasis, progression-free survival, and castration-resistant PCa [13–16]. However, the optimal integration of this quantitative information with existing clinical biomarkers to maximize utility is an area that still requires thorough exploration.

Prostate-specific antigen (PSA), a widely employed biomarker, remains a subject of significant attention and controversy. Thanks to the widespread application of PSA testing, the majority of PCa cases can be detected earlier and receive appropriate management. However, the lack of specificity in PSA often leads to a substantial number of false-positive results and the overdiagnosis of low-grade PCa. To address this issue, PSA density (PSAD), defined as the ratio of PSA levels to prostate volume (PV), was initially proposed by Benson et al. [17]. PSAD has been demonstrated as a more reliable predictor of PCa than PSA, enhancing the cancer-specific resolution of PSA [18, 19]. Consequently, we speculated that incorporating critical factors, such as quantitative data on PPAT, has the potential to further optimize the utility of PSA in PCa diagnosis and prognosis.

Therefore, this study aimed to explore the predictive value of various markers for high-grade PCa based on clinical biomarkers (including PSA, among others) and quantitative imaging parameters associated with PPAT. Building on these finding, a novel marker integrating PSA and PPAT was proposed, aiming to delve deeper into whether incorporating visceral adipose tissue enhances the clinical utility of PSA in screening for high-grade PCa.

Materials and methods

Patients screening

The Institutional Ethics Committee of The Second Affiliated of Chongqing Medical University approved this retrospective study and granted an exemption from the requirement for informed consent (Decision No. 289, 2019). Patient who underwent prostate MRI scans at our institution from January 2020 to January 2023 were screened through the Picture Archiving and Communication System (PACS). Inclusion criteria comprised: (1) Patients underwent preoperative prostate MRI examination, and (2) Pathological confirmation of PCa within 3 months post-MRI. Exclusion criteria included: (1) Incomplete clinical data, (2) Patients with prior prostate treatment history, and (3) Images with poor quality or artifacts affecting assessment. PCa patients were stratified into high-grade (GS ≥ 4 + 3) and low-grade (GS ≤ 3 + 4) groups based on GS. The patient screening process was depicted in Fig. 1 and a total of 215 patients were ultimately enrolled in this study.

Fig. 1.

Flow chart of the patient selection process

Clinical data acquisition

Clinical data, including age, height, weight, Body Mass Index (BMI), initial serum PSA levels, Gleason score (GS), and other relevant pathological information, were retrieved from the Radiology Information System (RIS). PSAD was calculated using the standard formula:

MRI image acquisition

The MRI images were acquired using 3.0 Tesla MRI scanners (SIEMENS Prisma, Philips Ingenia CX) with an 8-channel phased-array coil. The imaging modalities included axial T1-weighted imaging (T1WI), fat-suppressed T2-weighted imaging (T2WI), diffusion-weighted imaging (DWI), and dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI). The detailed MRI acquisition parameters are presented in Table 1.

Table 1.

MRI protocol

	Sequence	TE (ms)	TR (ms)	Thickness (mm)	FOV (mm)	Acquisition matrix	Avgs
SIEMENS Prisma	T2WI T1WI	77 12	3290 889	3.0 3.0	200 200	320 × 240 320 × 240	2 1
	DWI (b = 0/1400)	84	3800	3.0	200	118 × 118	2
	DCE	1.77	5.08	3.0	260	192 × 154	1
Philips Ingenia CX	T2WI	100	3544	3.0	240	268 × 268	1
	T1WI	9.0	459	3.0	240	320 × 240	1
	DWI(b = 0/1400/2000)	94	5000	3.0	300	120 × 126	1
	DCE	1.42	4.6	3.0	240	180 × 165	2

TE, echo time; TR, repetition time; FOV, field of view; DCE, dynamic contrast enhanced; Avgs, averages

The measurement of PPAT

PPAT is defined as the adipose tissue surrounding the prostate and located anterior to the rectum. The anatomical boundaries for its measurement are as follows: the superior boundary is the bladder, the inferior boundary is the urethral sphincter, the anterior boundary is the pubic symphysis, the posterior boundary is the Denonvilliers’ fascia, and the lateral boundary is the obturator internus muscle [5].This study employed three measurement methods to assess PPAT: thickness, area, and volume. Measurements and segmentation were performed on T1WI axial images using Syngo.via, version 4.1 (Siemens Healthcare GmbH, Frankfurt, Germany). All the measurements were independently conducted by two radiologists (XJ: with 3 years of experience in radiological diagnostics; QXF: with 8 years of experience in radiological diagnostics). The specific measurement methods are described as follows:

Thickness measurement

Subcutaneous fat thickness (SFT) and PPFT were measured at the level of the pubic symphysis. SFT represents the anterior-posterior distance from the skin to the superior border of the pubic symphysis(as illustrated in Fig. 2A) [5], whereas PPFT is defined as the shortest anterior-posterior distance from the posterior edge of the pubic symphysis to the anterior border of the prostate (as depicted in Fig. 2B).

Fig. 2.

Schematic diagram of PPAT measurements: Case presentation of a 66-Year-Old Patient with High-Grade Prostate Cancer (GS 4 + 5 = 9). A The solid green line depicts SFT. B The solid red line illustrates PPFT. C The yellow area signifies PPFA, while the purple area denotes PA. D The yellow areas indicate PPAT. PS, Pubic symphysis; P, Prostate; BL, Bladder; SFT, Subcutaneous fat thickness; PPFT, Periprostatic fat thickness; PPFA, Periprostatic fat tissue area; PA, Prostate area; PPAT, Periprostatic adipose tissue

Area measurement

The PPFA and prostate area (PA) were measured at the level of the pubic symphysis, specifically at the axial slice where the bladder and prostate intersect. The anatomical boundaries for PPFA measurement were defined as follows: the ventral boundary was the pubic symphysis, the dorsal boundary was Denonvilliers’ fascia, and the lateral boundary was the obturator internus muscle. The delineation encompassed the pre-venous plexus and adipose tissue posterior to the pubic symphysis, while excluding the prostate and seminal vesicle regions at this level [12], as illustrated in Fig. 2C and Supplementary Fig. S1. To mitigate the confounding influence of inter-patient variability in prostate gland dimensions, we introduced the PPFA/PA ratio as a two-dimensional normalization metric.

Volume measurement

Using the delineation method outlined in Sect."Area Measurement", PA and PPFA were manually segmented on each axial T1WI slice spanning from the base to the apex of the prostate [5]. Three-dimensional reconstruction was generated using Syngo.via software, which automatically calculated the peri-prostatic fat volume (PPFV) and prostate volume (PV) (Fig. 2D and Supplementary Fig. S2). The PPFV/PV ratio was introduced as a three-dimensional normalization index, thereby allowing for standardized assessment of periprostatic adiposity regardless of prostate size heterogeneity.

Statistical analysis

We conducted statistical analysis on patient characteristics using SPSS software (version 26; IBM Corporation, Armonk, NY, USA). Agreement between the two observers’ measurements was evaluated using the intraclass correlation coefficient (ICC), with a value exceeding 0.75 indicating substantial consistency. Normality of continuous variables was assessed via the Shapiro-Wilk test. Normally distributed variables are presented as mean ± standard deviation (SD), while non-normally distributed variables were analyzed using the Kruskal–Wallis test and expressed as median (interquartile range, 25% and 75%). Variance inflation factor (VIF) was computed to detect significant collinearity among variables, with non-collinear variables subjected to inter-group difference analysis. For normally distributed data, the two-sample t-test was utilized, whereas the Mann–Whitney U test was employed for non-normally distributed parameters. This study employed Spearman’s rank correlation coefficient for correlation analysis and visualized the correlation matrix using a heatmap. Variables exhibiting significant inter-group differences were included in binary logistic regression analysis. To evaluate the predictive performance of the model, Receiver Operating Characteristic (ROC) curves and the area under the ROC curve (AUC) were calculated. The optimal cutoff value was determined by maximizing Youden’s index (J = sensitivity + specificity − 1). The model’s goodness of fit to observed indicators was assessed using the Hosmer-Lemeshow goodness-of-fit test, where a P-value > 0.05 indicates a satisfactory fit. DeLong’s test was employed to compare AUC values among different models, with a P-value < 0.05 considered statistically significant.

Results

Clinical, PPAT, and composite indicator characteristics of patients

The study ultimately included a total of 215 patients, with 63 cases of low-grade PCa and 152 cases of high-grade PCa. The median age was 72 years, ranging from 49 to 90 years. Among them, 66 individuals (30.7%) were overweight (BMI ≥ 25 kg/m²), and 3 individuals (1.4%) were obese (BMI ≥ 30 kg/m²). The median preoperative serum PSA level was 25.90 (12.00–84.50) ng/mL. The baseline characteristics are presented in Table 2. The measurements of SFT, PPFT, PPFA, PA, PPFV and PV by the two observers demonstrated excellent consistency (ICC values all ≥ 0.85, P < 0.001), presented in Supplementary, Table S1.

Table 2.

Baseline characteristics

Characteristics	Values
Age, M (Q1, Q3), y	72 (68, 79)
Height, M (Q1, Q3), m	1.66 (1.61, 1.70)
Weight, Mean ± SD, kg	64.94 ± 8.77
BMI, Mean ± SD, kg/m2	23.71 ± 2.76
PSA, M (Q1, Q3), ng/ml	25.90 (12.00, 84.50)
SFT, M (Q1, Q3), mm	29.85 (22.24, 38.23)
PPFT, M (Q1, Q3), mm	6.90 (3.80, 9.67)
PPFA, M (Q1, Q3), cm2	17.78 (13.10, 25.21)
PA, M (Q1, Q3), cm2	13.52 (10.76, 17.36)
PPFV, M (Q1, Q3), cm3	48.76 (36.36, 63.52)
PV, M (Q1, Q3), ml	49.73 (36.11, 67.25)
PPFA/PA, M (Q1, Q3), %	133.18 (83.24, 191.67)
PPFV/PV, M (Q1, Q3), %	99.43 (73.83, 116.49)
PSAD, M (Q1, Q3), ng/ml2, %	58.53 (24.04, 153.14)
PSA×PPFA/PA, M (Q1, Q3), ng/ml	36.27 (13.69, 112.88)
PI-RADS, n (%)
2	3 (1.4)
3	28 (13.0)
4	35 (16.3)
5	149 (69.3)
Gleason score, n (%)
≤ 6	31(14.4)
3 + 4	32(14.9)
4 + 3	39(18.1)
8	57(26.5)
9–10	56(26.1)

SD, Mean ± Standard Deviation; M (Q1, Q3), Median (Interquartile Range, 25% and 75%); SFT, Subcutaneous fat thickness; PPFT, Periprostatic fat thickness; PPFA, Periprostatic fat tissue area; PA, Prostate area; PPFV, Periprostatic fat tissue volume; PV, Prostate volume; PSA, Prostate specific antigen; PSAD, Prostate specific antigen density; BMI, Body mass index

Differential analysis between high and low-grade PCa groups

Among all the baseline characteristics, collinearity analysis excluded Height, Weight, BMI, PA, and PV. Among the remaining indicators, statistically significant inter-group differences were observed for PPFA, PPFA/PA, PSA, PSAD, and PSA × PPFA/PA, as illustrated in Fig. 3.

Fig. 3.

Box plots of inter-group differences between high and low-grade PCa. SFT, Subcutaneous fat thickness; PPFT, Periprostatic fat thickness; PPFA: Periprostatic fat tissue area; PA: Prostate area; PV, Prostate volume; PPFV, Periprostatic fat tissue volume; PSA, Prostate specific antigen; PSAD, Prostate specific antigen density

Correlation analysis of fat quantification metrics

In this study, fat-related parameters included the clinical indicator BMI and imaging-based metrics SFT, PPFT, PPFA, PPFV, PPFA/PA, PPFV/PV, and PSA×PPFA/PA. To explore the relationships among these fat-related indices, a correlation analysis was conducted (Fig. 4). The results revealed significant correlations (|r| ≥ 0.2) between PPFV and PPFA, PPFA and PPFV/PV, PPFT and PPFA/PA, PPFV and PPFV/PV, PPFT and PPFV, PPFT and PPFV/PV, BMI and SFT, PPFA and PPFA/PA, PPFA and PSA×PPFA/PA, PPFA/PA and PPFV/PV, as well as PPFA/PA and PSA×PPFA/PA. Notably, BMI and SFT, as well as PPFA and PPFA/PA, exhibited a strong correlation(|r| ≥ 0.6).

Fig. 4.

Spearman correlation analysis of fat quantification metrics. *Indicates statistical significance at the P < 0.05 level (two-tailed); **Indicates statistical significance at the P < 0.01 level (two-tailed). Correlation coefficients are interpreted as follows:|r| >0.6: Strong correlation; 0.4 ≤|r| ≤ 0.6: Moderate correlation; 0.2 ≤|r| < 0.4: Weak correlation; 0 ≤|r| < 0.2: No correlation. A positive correlation is indicated when r > 0, and a negative correlation when r < 0. BMI, body mass index; SFT: subcutaneous fat thickness; PPFT: periprostatic fat thickness; PPFA: periprostatic fat tissue area; PPFV: periprostatic fat tissue volume; PA: prostate area; PV: prostate volume; PSA: prostate specific antigen

Analysis of predictive factors for high-grade PCa

The indicators demonstrating statistically significant inter-group differences were included in a multivariate binary logistic regression analysis. The findings revealed that the PPFA/PA ratio and PSA levels were independent factors predicting high-grade PCa, with odds ratios of 1.011 (95% CI 1.002–1.021) and 1.044 (95% CI 1.006–1.082), respectively. The corresponding P-values were 0.018 and 0.022, as depicted in Fig. 5.

Fig. 5.

Analysis of predictive factors for high-grade PCa. OR, odds ratios; CI, confidence interval; PPFA, periprostatic fat tissue area; PA, prostate area; PSA, prostate specific antigen; PSAD, prostate specific antigen density

Predictive model construction for high-grade PCa

Predictive models for high-grade PCa were constructed using statistically significant indicators and their composite. The analysis of the ROC curves have been depicted in Fig. 6.The AUC values of the multivariate binary logistic regression model for predicting high-grade PCa were presented in Table 3, in which PSA, PSAD, PSA × PPFA/PA, and the merged models demonstrated excellent predictive performance, with AUC values of 0.771, 0.796, 0.818, and 0.814, respectively. Notably, PSA × PPFA/PA exhibited the highest predictive efficacy. The optimal cutoff value for PSA × PPFA/PA was 42.135, with a Youden index of 0.5166.

Fig. 6.

ROC curves for high-grade PCa prediction. Blue represents PPFA, red represents PPFA/PA, green represents PSA, orange represents PSAD, yellow represents PSA × PPFA/PA, and teal represents Merge (composite indicator of PPFA + PPFA/PA + PSA + PSAD + PSA×PPFA/PA)

Table 3.

AUC values of the multivariate binary logistic regression model for predicting high-grade PCa

Variable	AUC (95%CI)	P value
Merge	0.814 (0.755–0.873)	< 0.001
PPFA/PA	0.652 (0.570–0.734)	< 0.001
PPFA	0.648 (0.567–0.729)	< 0.001
PSA	0.771 (0.707–0.835)	< 0.001
PSAD	0.796 (0.734–0.858)	< 0.001
PSA×PPFA/PA	0.818 (0.759–0.877)	< 0.001

PPFA, periprostatic fat tissue area; PA, prostate area; PSA, prostate specific antigen; PSAD, prostate specific antigen density. Merge: composite indicator of PPFA + PPFA/PA + PSA + PSAD + PSA×PPFA/PA

The DeLong test indicated notable variances in AUC among several models. However, no statistically significant differences were found in the AUC values the PSA×PPFA/PA model versus Merge. Showed that both models could predict high-grade prostate cancer, with the relatively slightly higher AUC values of the PSA PPFA/PA model model.Refer to Table 4 for comprehensive details.

Table 4.

DeLong test results for ROC curves

Variable	Z	P value
Merge - PPFA/PA	4.704	< 0.001
Merge - PPFA	4.433	< 0.001
Merge - PSA	1.634	0.102
Merge - PSAD	0.896	0.370
Merge - PSA×PPFA/PA	−0.282	0.778
PPFA/PA - PPFA	0.143	0.886
PPFA/PA - PSA	−2.206	0.027
PPFA/PA - PSAD	−3.142	0.002
PPFA/PA - PSA×PPFA/PA	−4.386	< 0.001
PPFA - PSA	−2.337	0.019
PPFA - PSAD	−2.593	0.003
PPFA - PSA×PPFA/PA	−4.294	< 0.001
PSA - PSAD	−1.406	0.160
PSA - PSA×PPFA/PA	−2.280	0.023
PSAD - PSA×PPFA/PA	−1.418	0.156

PPFA, periprostatic fat tissue area; PA, prostate area; PSA, prostate specific antigen; PSAD, prostate specific antigen density. Merge: composite indicator of PPFA + PPFA/PA + PSA + PSAD + PSA×PPFA/PA

Discussion

The effective detection of high-grade PCa is paramount in guiding patient treatment and predicting outcomes. Given the crucial role of adipose tissue in secreting bioactive substances, its intricate connection with high-grade PCa, alongside traditional clinical biomarkers, deserves deeper investigation. Our study aimed to assess the predictive power of standard clinical biomarkers (including PSA, among others) versus quantitative imaging measures related to PPAT for identifying high-grade PCa. We discovered that a novel biomarker, the product of PSA and PPAT area (PSA×PPFA/PA), outperformed others in predicting high-grade PCa. This finding underscores the value of integrating quantitative PPAT data to refine the screening effectiveness of PSA for high-grade PCa. Such an approach promises substantial clinical benefits and advocates for its broader implementation.

Obesity is widely acknowledged as a contributing factor to various cancers. BMI and SFT serve as prevalent metrics for gauging obesity, with their relevance to high-grade PCa screening being a subject of considerable debate in the scientific community. Research conducted by Dahran [5], Woo [10], Tan [11], and others has consistently demonstrated a lack of significant association between BMI and GS. This study corroborates these findings, indicating no significant differences in BMI between patients with high-grade and low-grade PCa. However, it’s important to note that certain studies have identified BMI as a predictive marker for high-grade PCa [7, 20]. This discrepancy may arise from various factors: firstly, BMI represents overall obesity but may not accurately depict fat distribution. Indeed, individuals with a normal BMI can still possess abdominal obesity. A substantial prospective study found that abdominal obesity indicators, such as waist circumference or waist-to-hip ratio, offer better predictive value for disease mortality than BMI does [21]. Pischon and colleagues [22] also discovered that a higher waist circumference, even in those with lower BMI, is linked with an elevated risk of high-grade PCa. Secondly, the influence of BMI might be modulated by regional and racial differences, as well as the size of the study cohort: obesity distribution varies markedly between Chinese populations and those from the US or Europe. In this research, the prevalence of obesity (BMI ≥ 30 kg/m²) was a mere 1.4%, contrasting with rates up to 12.6% reported in studies by van Roermund and others [20]. SFT, another crucial obesity measure, has been deemed by most research [5, 10, 14, 20, 23], including our own, as an ineffective predictor for high-grade PCa. This is likely because SFT only quantifies subcutaneous fat at the pubic symphysis level, offering a limited view that may not accurately reflect total abdominal fat. Furthermore, evidence suggests that plasma high-density lipoprotein cholesterol and triglyceride levels have distinct associations with subcutaneous versus retroperitoneal Fat, the latter of which correlates more closely with waist circumference, thus providing a more reliable measure for assessing abdominal obesity [24].

The findings of this study demonstrated no significant correlation between BMI or SFT and parameters associated with PPAT, further reinforcing the notion that BMI and SFT are unreliable surrogates for assessing visceral fat content or distribution. In fact, previous studies have shown that PPAT volume is independent of BMI and does not increase even in individuals with obesity who exhibit excessive accumulation of other white adipose tissue depots, including visceral fat [25]. The regulatory mechanisms underlying PPAT volume and function remain incompletely understood. Emerging evidence suggests that certain pharmacologic agents, such as 5α-reductase inhibitors, may reduce PPAT volume, while sex hormone levels appear to influence its metabolic, endocrine, and angiogenic characteristics [26, 27].

As a significant source of pro-inflammatory mediators, PPAT has been increasingly implicated in the pathogenesis and progression of PCa [26]. Our findings indicate that, beyond the scope of BMI and SFT, PPAT emerges as a more relevant factor closely associated with high-grade PCa. PPAT, enriched with highly active fat cells, plays a pivotal role in the progression of PCa through paracrine signaling pathways, especially the CCR3/CCL7 chemotactic route [6]. This interaction significantly promotes the aggressive encroachment of PCa into adjacent adipose regions. Such insights underscore the necessity for detailed exploration of PPAT, prioritizing quantitative assessments like thickness, area, and volume. Our analysis, focusing on these dimensions, revealed that both PPFA and PPFA/PA ratio display significant variances between high-grade and low-grade PCa groups, with a P-value ≤ 0.001. Notably, the PPFA/PA ratio stands as an independent prognosticator for high-grade PCa, presenting an odds ratio of 1.011 (95% CI 1.002–1.021). However, parameters such as thickness (PPFT) and volume (PPFV) did not showcase significant distinctions between the PCa categories. This aligns with findings from researchers like Zhai, Zhang, and others, who have validated the correlation between PPFA, especially the PPFA/PA ratio, and the aggressiveness of prostate cancer, affirming their predictive accuracy for tumor staging or grading [12, 13, 23]. Contrarily, studies by Tan et al. found noteworthy correlations between PPFT, PPFV/PV, and the PCa GS [5, 10, 11], yet these metrics lacked substantial discriminatory capacity in our study. This variation may stem from the intrinsic limitations of PPAT measurement techniques. While PPFT, as a one-dimensional metric, is simple and highly replicable, it might not fully capture the comprehensive distribution of PPAT. Theoretically, volumetric assessments could offer more precise quantification of adipose tissue. Nevertheless, the complexity and potential for error in collecting volumetric data (PPFV) dampen its practical utility. In comparison, the PPFA/PA ratio emerges as a more accessible and objective metric, providing a powerful tool for evaluating periprostatic adipose involvement in prostate cancer. This highlights the PPFA/PA ratio significant clinical promise for the refined differentiation between high-grade and low-grade PCa, advocating for its wider application in clinical settings. Particularly, the introduction of the PSA×PPFA/PA index represents a noteworthy advancement, achieving the highest predictive accuracy for high-grade PCa in this study with an AUC of 0.818.This aligns with the findings of Zhai et al. [12], who illustrated that the inclusion of PPFA/PA into a clinical model—encompassing PSA, age, digital rectal examination (DRE), family history of PCa, and PIRADS score—markedly enhances the model’s capacity to detect PCa, boosting the ROC AUC from 0.75 to 0.93 over PSA alone.These results underscore the potential of integrating periprostatic adipose tissue data to substantially improve the predictive capability of clinical parameters such as PSA in pinpointing high-grade PCa. Corroborating this view, Tianyu Xiong et al. [28] have demonstrated that models enriched with fat characteristics significantly elevate AUC values for PCa prediction (AUC = 0.850 vs. 0.819) and for identifying clinically significant PCa (csPCa) (AUC = 0.827 vs. 0.798), compared to models without these fat characteristics.Additionally, the literature suggests an inherent connection between PPAT and PSA levels. The team led by Jeong Won Lee, employing FDG PET/CT for PPAT measurement, identified a significant positive correlation between CT attenuation, FDG uptake, tumor staging, and serum PSA levels [29]. In a similar vein, Saglam K et al. uncovered a linkage between leptin released by PPAT, circulating PSA levels, and the Gleason score of prostate biopsies [30]. This indicates that the diverse factors secreted by PPAT cells may encourage cancer cell proliferation, migration, and invasion, thereby exerting an influence on PSA secretion. Furthermore, the measurement techniques employed in this study, particularly MRI-based PPAT quantification, are technically feasible and can be integrated into routine clinical practice with minimal additional time or resource requirements. Given the growing availability of high-resolution MRI across various clinical settings, incorporating these measurements into patient management could offer both practical and clinical benefits.

This study has several limitations. First, as a single-center investigation with a relatively modest sample size, the generalizability of our findings may be constrained. Future multicenter, large-scale prospective studies are warranted to validate the robustness and broader applicability of our results. Second, the assessment of PPAT-related parameters in this study was conducted through visual evaluation and manual delineation, which may introduce potential selection bias and limit our ability to capture more nuanced imaging features, such as grayscale variations, texture, and other intrinsic characteristics. To overcome these limitations, future research could incorporate radiomics and artificial intelligence techniques to enable automated segmentation and the extraction of higher-order features, such as texture, which are often undetectable to the human eye. This approach could enhance prostate cancer risk prediction models and provide deeper insights into the complex interplay between PPAT and high-grade prostate cancer. Furthermore, the median PSA level in our cohort was relatively elevated (25.90 ng/mL), which may reflect variations in prostate cancer screening and diagnostic protocols across different countries and healthcare systems. Larger, more diverse patient populations in future studies will be crucial to validate the model and assess its generalizability across varied demographic and clinical contexts.

Conclusion

To encapsulate, this study investigated the prognostic significance of clinical biomarkers, such as PSA and others, combined with quantitative evaluations of PPAT, for detecting high-grade PCa. We proposed a novel composite index, PSA×PPFA/PA, demonstrating significant effectiveness in predicting high-grade PCa. Integrating PPAT metrics offers a promising approach to enhance the precision of high-grade PCa screening when combined with established clinical markers such as PSA, thus improving the accuracy of clinical decision-making. Given the complex role of PPAT, future research should further explore the interactions among PPAT, PCa, and clinical biomarkers, with the aim of creating refined predictive models suitable for clinical use.

Abbreviations

PPAT

Periprostatic adipose tissue

PCa

Prostate cancer

PSA

Prostate-specific antigen

SFT

Subcutaneous fat thickness

PPFT

Periprostatic fat thickness

PPFA

Periprostatic fat area

PA

Prostate area

PPFA/PA

Peri-prostatic fat area to prostate area ratio

PPFV

Periprostatic fat volume

PV

Prostate volume

PPFV/PV

Peri-prostatic fat volume to prostate volume ratio

CT

Computed tomography

MRI

Magnetic resonance imaging

PSAD

Prostate-specific antigen density

PACS

Picture Archiving and Communication System

RIS

Radiology Information System

BMI

Body Mass Index

GS

Gleason score

DRE

Digital rectal examination

Author contributions

Conceptualization, J.X.;Data curation, J.X.,X.F.Q.and G.Y.A.; Formal analysis, Y.F.L.; Funding acquisition, X.J.H.;Investigation, J.X.;Methodology, Y.F.L.;Project administration, X.J.H.; Resources, X.J.H.; Supervision, X.J.H.;Validation, X.J.H.and Y.F.L.;Visualization, J.X.,Y.F.L.and J.Q.M.;Writing-original draft, J.X.andY.F.L.; Writing-review & editing, X.J.H.All authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Funding

This study was supported by the Science and Health Joint Medical Research Project of Chongqing (2024ZDXM004);Senior Medical Talents program of Chongqing for Young and Middle aged; Kuanren Talents Program of the second affiliated hospital of Chongqing Medical University; and Program for Youth Innovation in Future Medicine, Chongqing Medical University(W0140).

Data availability

The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethical approval

The Institutional Ethics Committee of The Second Affiliated of Chongqing Medical University approved this retrospective study and granted an exemption from the requirement for informed consent (Decision No. 289, 2019).

Competing interests

The authors declare no competing interests.

Consent for publication

Not applicable.