According to the Global Cancer Statistics 2020 record, there was predicted to be 1.4 million new cases of and 375,000 deaths from prostate cancer (PCa) worldwide in 2020. PCa ranks second in the incidence of male malignant tumors and fifth in mortality (1, 2). In recent years, with the acceleration of demographic aging, changes in dietary structure and improvements in PCa diagnostic technology, the incidence of PCa in China has shown a significant upward trend. The morbidity and mortality of PCa rank second and fifth for Chinese male malignant tumor incidence and mortality, respectively, and it is gradually becoming an important disease affecting the health of elderly men in China (1). There are various treatment options for PCa, including watchful waiting and active surveillance, radical prostatectomy (RP), and radical radiotherapy. RP is the preferred therapy for clinically confined PCa, including complete removal of the prostate and seminal vesicles. However, approximately 11%-40% of patients have been reported to have positive surgical margins (PSM) after undergoing RP, which is the highest PSM rate among male malignant tumors (3–5).

PSM is defined as the presence of tumor cells in postoperative pathological specimens that are in contact with pigment markers on the specimen surface. There are actually two types of PSM in clinical practice: one is when the cancer tissue has infiltrated outside the capsule, the tumor cannot be completely removed by surgery, and some cancerous tissue is left outside the prostate on the ink-stained margins; the other is when the cancer tissue is confined to the capsule, the fascia or capsule around the prostate is incised for various reasons, and enters the prostate gland by mistake, resulting in the disappearance of part of the fascia and capsule in the specimen, and the prostate remains on the ink-stained incision edge (6). The apex is the most frequent site of PSM due to the broad range of the dorsal venous complex, its location under the pubic bone approaching the neurovascular bundles, and the lack of capsule coverage. Surgeons preserve the urethral sphincter as much as possible to protect postoperative urinary control function and for many other reasons (7). PSM is considered to be an important factor leading to an elevated risk of biochemical recurrence and recurrence of local tumors, which occasionally require secondary treatment (4, 6). During 7-13 years of follow-up, 27-44% of patients diagnosed with PSM developed biochemical recurrence, 6.8-24.3% exhibited systemic disease progression, and 0.8-3.7% experienced cancer-related death (8–11). A meta-analysis of 32 studies involving 140,000 PCa cases showed that the overall survival rate of patients determined to have negative margins was 1.11 times that of patients determined to have PSM (12). Moreover, patients with PSM may experience severe depression due to excessive psychological stress (13). Although laparoscopic radical prostatectomy (LRP) and robot-assisted laparoscopic radical prostatectomy (RALRP) improve functional recovery after surgery compared to open radical prostatectomy, there is no remarkable efficacy in terms of PSM reduction. The occurrence of PSM after RP has become a common and difficult-to-ignore problem. Correctly diagnosing and evaluating PSM, reducing the incidence of PSM, and improving the therapeutic effect of radical prostatectomy are challenges that urgently need to be solved.

Significant patient-specific predictors such as age, body mass index, and prostate-specific antigen (PSA) have been identified associated with PSM (14). Treatment-specific factors correlating with the PSM have also been confirmed, proficiency in anatomical hierarchy and the experience of surgeons may be major determinants of the PSM (15). Preoperative neoadjuvant therapy also contribute to reduce the incidence of PSM, nevertheless the effectiveness is still debatable (16, 17). Tumor-specific characteristics including prostate volume, T stage and Gleason score exposed significant notable association with PSM (18, 19). As the evolving progress of reliable prognostic biomarkers in urological cancer, increasing number of biomarkers for predicting PSM have been discovered (2): immunohistochemical staining for Ki-67 may indicate high risk of biochemical recurrence in patients with PSM after radical prostatectomy (20). Meanwhile, moderate or strong heterogeneous nuclear ribonucleoprotein A1 (hnRNPA1) immunostaining was relevant to positive surgical margin and adverse tumor features (21). Nevertheless, low or negative expression of carcinoembryonic antigen-related cell adhesion molecule 1 (CEACAM1) have linked to PSM in patients with ERG fusion-positive PCa (22). In addition to the abnormal expression of genes or proteins which included BRCA1-associated protein-1 (BAP1), RNA-binding motif protein 3 (RBP) and Fas-associated death domain-containing protein phosphorylation (FADD) by immunohistochemical analysis, changes of mutational level are also strongly associated with PSM by fluorescence in situ hybridization analysis (2, 23–25). Kluth M et al. reported that Chromosome 16q deletion was linked to PSM whether ERG fusion in PCa patients was positive or negative (26).

Multiparameter MRI (mpMRI) has become the preferred method for the noninvasive diagnosis of PCa due to its high resolution for soft tissue and multifaceted and multiparametric imaging. MRI has played an important guiding role in clinical staging, postoperative efficacy evaluation and recurrence (27–30). Although the measurement of various parameters related to the prostate has been achieved in two-dimensional (2D) planes based on mpMRI, the recognition of the three-dimensional (3D) structure of the prostate and surrounding tissue and the spatial distance measurements are still far from satisfactory. To achieve a better understanding of anatomy, the 3D printing technique rose to the occasion, while the prohibitive cost has limited large-scale application so far. With the in-depth promotion of precision medicine and digital medicine, clinical medicine has entered the digital age. One of the important technologies in digital medicine is to use human medical image datasets to achieve 3D visualization through software. Compared with 2D images, which provide less complete information on the complexity of the pelvic anatomy, 3D visualization technique may enable better multiangle observation of the anatomical structure and the variations in blood vessels and surrounding tissues. Therefore, we aimed to determine the value of 3D visualization of preoperative prostatic MRI based on measurements of anatomical structures in predicting PSM and APSM after RP.

Among the 149 patients, 41 patients were determined to have PSM, and 22 of the 41 patients had multiple PSM sites. The PSM locations are presented in Figure 9 . The most frequent location of PSM was the apex, accounting for 47.1% of the total PSM.

PSM is a complex issue for urologists since it is strongly associated with worse clinical outcomes (6, 39). The cutoff PSM rate in our study was 27.5%, which is high in view of the global PSM rate of 6.5-32% (6), but in the Asian region, the cutoff rate has been approximately 21.9%-38.3%, making our rate low to moderate there (19, 40–43). This phenomenon may be due to the presence of objective factors such as screening technology and the prevalence of physical examination in Asian studies, which lead to the late detection of PCa, resulting in a lower proportion of patients with an early clinical stage and thus indirectly increasing the PSM rate. Furthermore, some studies have reported that the diverse outcomes of PCa between Western and Eastern countries may be due to different pelvic anatomical structures (44–48). The most common site of PSM is the prostatic apex, with an incidence of 40-63%. Our results are in line with these, at 47.1%, due to the inclusion of multiple positive margins (19, 40, 49, 50). One reason why the apical part of the prostate is the most likely site for positive margins may be the lack of a fibrous envelope in the apical part of the prostate. The other is that freeing the distal prostate is the most challenging and thought-provoking part of the procedure in RALRP. Since the prostatic apex is located below the pubic bone, the dorsal venous complex in this area is extensive and close to the neurovascular bundle. Complete removal of the tumor by more extensive excision of the neurovascular bundle may lead to postoperative incontinence, impotence, and other lower urinary tract sequelae (51). Preservation of neurovascular bundles becomes incompatible with complete resection of the PCa. The high incidence of APSM may also be related to the surgeon’s excessive intraoperative stretching of the prostate, resulting in fissures in the prostate peritoneum and the attempt to preserve additional urethral length (52).

Since MRI is one of the most important examination methods for patients with PCa, we explored the anthropometric measurements measured on MRI. The seminal study by Permpongkosol S et al. measured the pubic angle, work space AP, work space transverse and bladder protrusion (35). These imaging parameters were analyzed, and some statistically significant anthropometric measurements were found to be associated with PSM. The reason why pubic angle 2 was strongly correlated with PSM is that the pubic angle influences the access of a laparoscopic instrument or robotic arm to the prostatic apex and thus might indirectly affect surgical outcomes. For the same reason, the longer the work space AP is, the wider the surgical space and field of view, thus lowering the overall incidence of PSM. However, it was found that ISD, BFW, pelvic brim distance, and AD in this study were not statistically significant predictors. These conclusions were probably due to the proportion of patients who underwent RALRP in our study, in whom the flexible robotic arm was able to perform more precise manipulations in a relatively narrow space than laparoscopy would have achieved.

In our study, the variation in prostate apical shape may have been strongly associated with PSM, which indicated that different prostate apical shapes would affect the difficulty of stripping the membranous urethra during the operation. Preserving the MUL as much as possible with complete tumor resection may be one of the crucial steps to avoid delayed recovery of urinary control, especially for tumors located in the apical region of the prostate. Since we uncovered a strong correlation between apical type 3 and PSM, 3D visualization for adequate evaluation was necessary before the operation, which may remind clinicians perform careful stripping of the prostatic apical stump. Owing to the fact that apical type 4 did not enclose a membranous urethra at apex, this gives the urologist a maximum likelihood of completely removing the prostate and preserving adequate MUL, which may contribute to explaining why patients with apical type 4 had a better prognosis, a finding consistent with previous studies (36). In addition, finding effective surgical techniques or methods for stripping the apex of the prostate may reduce the high incidence of PSM and APSM, such as adjusting the robotic lens to face 30° upward and pulling the prostate cephalad, dissecting from the posterior inferior surface of the gland to the junction of the acinus and the membranous urethra, and early dissection of the dorsal venous complex without ligation by adding pneumoperitoneum (53, 54).

Although many meaningful parameters were identified in the MRI in this study, due to the existence of objective factors such as the limitation of viewing angle in conventional MRI, our study innovatively focused on 3D visualization techniques to more finely measure the diameters and volumes that cannot be measured or measured accurately in 2D images. For the accurate measurement of prostate volume, for instance, most medical institutions today use the elliptical formula for estimation to substitute actual data. However, this method is fundamentally flawed because the formula only obtains the appropriate data for estimation from three dimensions (sagittal, coronal and axial images). Additionally, due to the presence of tumors in PCa patients, the irregular shape of the prostate caused by the interaction of various stresses in the prostate, the diversity of tumor locations and the great individual differences in the prostate, the deviation in measurements of the prostate in patients with PCa using the conventional method will be even greater, reducing the clinical relevance. It should be noted that the interpretation of this result is somewhat limited by the fact that some patients contained multiple prostate tumors, and there were particular restrictions on tumor selection. The above results are consistent with the findings of Song C et al., who also mentioned limitations in tumor volume measurements and utilized calculated volumes rather than actual volumes. Our study has overcome the above limitation with 3D technology, making the results more convincing (55). The reason that no significance was found in the multivariate analysis might be due to the larger tumor having closer proximity to the prostate apex, which is another important finding of ours: patients with a longer UD had a lower incidence of PSM and APSM. Quentin M et al. similarly also found the above result (38). Considering the recovery of early urinary control in patients, the surgeon will try to preserve the urethral length when releasing the prostate apex, which may indirectly contribute to the higher likelihood of PSM and APSM in patients with a shorter UD. We recommend preoperative 3D visualization for patients with a potentially short UD identified by MRI, which, by looking at the relative position of the tumor to the membranous urethra from more planes, enables a highly individualized preoperative surgical plan and reduces the likelihood of PSM by widening the extent of the membranous urethra to be resected.

Only a few previous studies have performed nomograms for the preoperative prediction of PSM, and only correlations regarding clinical information and conventional MRI have been explored (14, 56). In this study, 3D visualization technology was combined with significant clinical data and imaging parameters based on MRI to improve the predictive capability and accuracy of the nomograms. The AUC and C-index values both demonstrated a great capability of prediction. Therefore, urologists can predict the likelihood a patient will suffer from PSM and APSM preoperatively, thus reducing the occurrence of events during surgery by applying neoadjuvant therapy or expanding the scope of surgical resection.

This study had a few limitations that should be considered. First, it was conducted in a single center, resulting in a limited sample size as well as a lack of external validation. Second, due to the most of the included patients lived in northeastern China, selection bias and potential confounding factors may have influenced our conclusions. Furthermore, we conducted the retrospective clinical study from December 2016 to April 2022. During this time period, all the patients in our center were scanned with a 3.0-mm-thick layer, which may affect the quality of the 3D visualization. As the layer thickness of the scans may affect the observation of detailed anatomy, thin-slice MRI would improve the quality of 3D visualization more accurate. In order to verify the conclusions of this study and the power of the independent risk factors and nomograms for predicting PSM and APSM, large-scale, multicenter, prospective studies are warranted. Moreover, an external validation cohort is needed to verify our findings. By utilizing 3D visualization techniques based on thin-slice imaging, more potentially instructive factors can be measured in more accurate images, providing clinical guidance for the subsequent development of imaging guidelines for prostate PSM prediction. 3D visualization can help observe tumor and surrounding environment changes more intuitively and assess treatment efficacy more effectively after endocrine therapy or chemotherapy. We further make adjustment for therapeutic regimen timely based on the precise monitoring the changes of the degree of tumor volumetric or the size of pelvic lymph nodes. Although our study did not conclude that MUL can act as an independent influencing factor for PSM, we are the first to study MUL using 3D visualization. In other aspects of digital medicine, we have embarked on deep learning of 2D and 3D imaging images using artificial intelligence with the aim of quickly predicting the likelihood of PSM preoperatively. Through more in-depth research on this topic, we will explore ways to control PSM as well as APSM from various dimensions to help reduce PSM and APSM rates, improve patient prognoses, and enhance the quality of life of PCa survivors in China and around the world.

In conclusion, our findings show that UD, work space AP, prostate apical shape and public angle 2 are significant predictors of PSM and that UD and pubic angle 1 are independent factors for APSM. The constructed nomograms performed well at predicting these events. The routine use of 3D visualization in clinical diagnosis can help to tailor individual treatment regimens to benefit PCa patients. We will validate the current findings and reveal the mystery in future prospective large-sample multicenter studies by utilizing higher-quality 3D technology and investigating the risk factors for PSM and APSM in greater depth to provide clinical guidance for the development of imaging guidelines for PSM and APSM prediction.

The original contributions presented in the study are included in the article/supplementary material. Further inquiries can be directed to the corresponding authors.

The studies involving humans were approved by the Ethics review committee of the Second Affiliated Hospital of Dalian Medical University (ethical approval number: 202285). The patients/participants provided their written informed consent to participate in the study. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.

BF, ZL, LW and JL contributed to the study concept and design, undertook project leadership and guaranteed this work. LZ, YCW and were responsible for the acquisition of data. YTW, HW, ZX and XD analyzed data and interpreted the data. LZ and YCW controlled quality of data and algorithms. BF, LZ and YCW wrote the first draft of the manuscript. BF, ZL, LW and ZD reviewed and revised the manuscript. HP and JX provided technical support. All authors contributed to the article and approved the submitted version.