Exosome diagnostics beyond the hype: why study design still matters

Dear Editor, We read the review by Li et al, “Exosome-Based Liquid Biopsy for Rare and Complex Diseases,” with great interest[1]. The authors provide a broad and timely overview of exosome-based biomarkers across a wide spectrum of conditions and highlight important technological advances. From a urology and liquid biopsy perspective, however, several methodological and interpretive issues merit clarification to support a more balanced and clinically useful view of the current evidence. Transparency In the reporting of Artificial Intelligence – The TITAN guideline[2]. Although the manuscript synthesizes findings from a wide range of studies, it does not describe how the literature was identified, selected, or appraised. In a field where small exploratory cohorts, heterogeneous isolation pipelines, and single-center experiences are common, the absence of transparent selection criteria makes it difficult for readers to judge how representative the included evidence is. This issue becomes particularly important in areas directly relevant to urology and nephrology, where diagnostic decisions – such as whether to perform a kidney biopsy or interpret early signs of rejection – require biomarkers supported by clearly traceable evidence[3]. The manuscript frequently highlights high diagnostic accuracy, often reporting AUC values above 0.90. However, these performance metrics are rarely contextualized with respect to study size, validation level, or methodological rigor. A visualization of reported AUCs extracted from the review (Fig. 1) illustrates how the highest values cluster among small studies, a pattern consistent with early-phase overoptimism. Without a clearer discussion of sample sizes, external validation, and potential bias, readers may inadvertently overestimate the maturity of several promising exosomal signatures. Figure 1.: Reported diagnostic performance of exosome-based biomarkers across non-oncologic diseases. Lollipop chart illustrating AUCs extracted from Li et al, grouped by disease and biomarker type. Each point represents a reported diagnostic panel, ordered from lowest to highest performance. The dashed line at AUC = 0.90 highlights the concentration of near-perfect values among relatively small exploratory studies. This pattern supports the concern that early exosome biomarker reports may reflect small-study overoptimism rather than mature, externally validated performance. A second concern relates to the manuscript’s positioning of exosome-based assays as noninvasive alternatives to tissue biopsy. While this is an important long-term goal, the review rarely connects these biomarkers to histopathologic gold standards[4]. For renal diseases such as diabetic kidney disease, lupus nephritis, Fabry nephropathy, tuberous sclerosis, or antibody-mediated rejection, it would be valuable to understand how exosomal markers correspond to chronicity indices, podocyte injury, microvascular inflammation, or specific Banff lesions. Without such anchoring, the enthusiasm for “biopsy replacement” risks appearing premature. The discussion of terminological precision is commendable, particularly the alignment with MISEV 2023 recommendations. Yet many cited studies identify vesicles using ultracentrifugation and tetraspanin markers alone, making them more accurately classified as small extracellular vesicles rather than true exosomes. A simple categorization of included studies based on the rigor of EV characterization would help align the review’s terminology with its evidence base. Finally, the sections addressing artificial intelligence and machine learning present attractive directions but remain general[5]. In practice, EV-omics datasets are strongly affected by batch effects, p≫n imbalance, and domain shifts between centers. These issues directly influence the reproducibility of exosome-based models in renal oncology, transplant monitoring, and metabolic disease. A brief acknowledgment of such challenges would provide a more realistic and clinically grounded perspective on the trajectory of AI-enabled biomarker development. Overall, this review provides an informative and forward-looking overview of exosome-based diagnostics beyond oncology. Clarifying literature selection, contextualizing reported performance metrics, strengthening links to histopathology, and more fully addressing methodological constraints would enhance the manuscript’s usefulness for clinicians and researchers. Future work integrating standardized EV pipelines, multicenter validation, and domain-specific clinical endpoints will be essential for translating the promise of exosome-based liquid biopsy into routine practice.