This study aimed to assess the performance of MRI in determining the lymph node stage of PCa patients based solely on morphological characteristics derived exclusively from T1W and T2W sequences, as current imaging protocols for the lymph node evaluation do not include DWI. To our knowledge, no recent studies have evaluated MRI without functional sequences in PCa staging, except for the Hovels meta-analysis (1980–2003) [5]. Our findings demonstrate a pooled sensitivity of 24.5% and a pooled specificity of 95%. Notably, all the readers failed to detect patients with high-volume metastatic disease.

Hovels reported a higher sensitivity (39%), which is likely attributable to suboptimal histological reference standards. In many cases, only limited PLND was performed, potentially leading to an underestimation of true N1 prevalence. Conversely, their lower specificity (82%) may reflect technological advancements in MRI over the years. Given that 1.5T MRI was introduced in the mid-1980s and that 3T MRI emerged between 1999 and 2002, Hovels findings may now be outdated. Additionally, studies included in the meta-analysis lacked standardized patient selection criteria, particularly regarding PCa risk stratification, PSA levels, clinical T stage, and biopsy-based Gleason scores.

In accordance with the Hovel findings, MRI demonstrated sensitivity comparable to that of CT (42%) while maintaining similar specificity. Consequently, our results can be compared with those of a recent nationwide Swedish registry study, which reported an 18% N1 detection rate for CT in high-risk PCa patients. This closely aligns with the MRI detection rates in our study (7%, 14%, and 18% for the three readers, respectively), yielding a pooled detection rate of 13% [15]. As the data were derived from the National Prostate Cancer Register of Sweden (NPCR) between 2014 and 2019, they reflect real-world clinical practice and support the generalizability of our findings.

Conventional MRI relies on size and morphological criteria for assessment. Given that over 80% of metastatic lymph nodes in PCa are less than 8 mm in size [16], detection rates remain low. The addition of functional imaging, such as DWI, has demonstrated superior performance [6]. A study utilizing 3T MRI with T1W, T2W, and DWI/ADC achieved a sensitivity of 55% and a specificity of 90% [17]. Similarly, Vallini et al. (2016) reported 84.6% sensitivity in intermediate- and high-risk PCa patients, using an ADC threshold value of 0.91 × 10–3 mm2/s, with ePLND as the reference standard [18]. However, the authors acknowledged the need for validation in larger studies and highlighted potential limitations, including DWI motion artifacts and the inability to detect micrometastases in lymph nodes smaller than 5 mm. Differentiating malignant from benign nodes via DWI and ADC values was superior to the use of size criteria alone, as demonstrated by Eiber et al. (2010) [7].

Conversely, a 2020 meta-review suggested that DWI provides only a modest improvement in diagnostic performance [19]. The review highlighted several limitations, including false-positive results due to necrotic or reactive nodes, partial volume effects in normal-sized lymph nodes, and artifacts from motion or bowel gas. Additionally, ADC values vary across MRI vendors, further complicating standardization.

The 2024 European guidelines on metastatic staging highlight that choline and 68Gallium PSMA PET-CT, along with whole-body MRI, outperform traditional bone scans and CT in detecting metastatic disease [2]. However, this conclusion is largely based on the findings of Hofman´s proPSMA randomized multicenter trial [20], which included a mix of regional and distant metastases. A systematic review and meta-analysis by Perera et al. included only five studies with PLND as the reference for lymph node staging, likely overestimating PSMA PET‒CT sensitivity at 77% [21]. A recent meta-analysis suggested that PSMA PET-CT may outperform MRI, with pooled sensitivities of 65% and 41%, respectively, while maintaining comparable specificity [8]. However, the included studies were predominantly retrospective and varied in methodology. Interestingly, the reported MRI sensitivity in this meta-analysis closely mirrors that of the Hovels earlier study (39%) [5], which likely did not include DWI sequences. In contrast, a study on 18F-PSMA-PET/CT using strict methodology and ePLND as the histological reference showed a sensitivity of only 26.9% [22].

Despite some promising findings, high-level evidence supporting treatment decisions based on PSMA PET-CT remains limited, and its global availability is restricted [9]. Consequently, the routine use of PET-CT for initial staging is not yet recommended [23]. Given the widespread use of MRI for PCa detection prior to guided biopsies [24, 25], optimizing MRI for metastatic screening could be a cost-effective approach. This study emphasizes the need for standardized guidelines concerning the use of DWI and more advanced criteria for N1 evaluation. Future research should focus on incorporating functional imaging, potentially enhanced by Artificial Intelligence algorithms, to improve the differentiation of metastatic from benign lymph nodes. The authors conclude that combining DWI - without necessarily relying on ADC values - with morphological criteria from T1W and T2W sequences, might provide superior diagnostic performance compared to morphological assessment alone.

The strengths of this study include the high proportion of high-risk PCa patients (93%) and the robust histological reference standard provided by ePLND. The histological reference is critical for validating radiological modalities, and local data indicate that approximately 40% of high-risk PCa patients undergoing ePLND have N1 disease, an observation confirmed by this study (Table 3). Additionally, the median number of lymph nodes harvested (17 per patient) reinforces the reliability of the histopathological assessment. Another strength is the study´s focus on primary PCa staging, avoiding the confounding effects of patients’ biochemical recurrence after curative intent, which is a limitation of other studies [26]. Furthermore, the interobserver agreement analysis demonstrated that MRI interpretation based solely on morphological criteria was relatively reproducible (Table 4).

However, the time interval between MRI and ePLND varied considerably, particularly among patients enrolled in earlier years, introducing a potential bias related to disease progression. Additionally, owing to the absence of standardized N1 criteria in clinical guidelines, readers rely on subjective judgment to classify lymph nodes, reflecting real-world clinical practice. Another limitation is the absence of data on the size of lymph node metastases in the histopathological reference, which could have provided further insights into detection challenges.