Xinhua Hospital Launches Large-Scale Retrospective Study of 2,000 Pediatric Solid Tumor Cases for Pathological Diagnosis Using AI
Study Design and Current Status
Xinhua Hospital, affiliated with Shanghai Jiao Tong University School of Medicine, has registered a large-scale AI model study (NCT06822842) to analyze pathological images and diagnostic results of pediatric solid tumors. This is a non-interventional, diagnostic observational study, categorized as a non-interventional clinical development stage rather than a Phase 1/2/3 trial, with a target enrollment of 2,000 cases. According to the ClinicalTrials.gov entry updated on February 12, 2025, the study is in a pre-recruitment state, with a study start date of February 1, 2025, and an expected completion date of April 30, 2025. Subsequent status updates are a key monitoring point.
Indications and Model Strategy
The study focuses on neuroblastoma, medulloblastoma, Wilms tumor, hepatoblastoma, and rhabdomyosarcoma. The core strategy is to develop a versatile model that can handle heterogeneous rare pediatric cancers, rather than an algorithm optimized for a single disease. The model integrates high-resolution pathological images and diagnostic text, incorporating tumor molecular characteristics, pathological grade, and diagnostic rules to align image and language representation spaces. The primary tasks are tumor region segmentation, cancer detection, and subtype identification. The study focuses on building and validating a diagnostic model, rather than evaluating treatment efficacy.
Clinical Significance Compared to Standard Diagnosis
Currently, the standard approach involves hematoxylin and eosin (H&E) staining interpretation by a pathologist, combined with immunohistochemistry (IHC) and molecular/genetic testing, with specialist center re-review performed when necessary. This study aims to develop an AI-powered tool to enhance the consistency of rare subtype identification and grade assessment, rather than replacing expert judgment. In particular, the 2,000-case dataset will become clinically valuable when sufficient sample sizes per tumor type, external validation, and scanner generalization performance are achieved.
Competitive Landscape and Market Potential
The competitive landscape includes digital pathology and pathology foundation models such as Paige's Paige Prostate, Philips' IntelliSite Pathology Solution, and the RuiPath platform developed by Ruijin Hospital and Huawei. However, Paige Prostate is a prostate biopsy aid, and RuiPath is a pan-cancer model. Xinhua Hospital's differentiated asset is its clinical data specifically focused on multiple subtypes of pediatric solid tumors. The global digital pathology market is projected to grow from USD 1.3 billion in 2025 to USD 5.75 billion in 2034, but this study does not involve any licensing, sales, or investment transactions.
Regulatory Pathway and Commercialization Hurdles
The U.S. FDA approved Philips IntelliSite as the first whole-slide imaging (WSI) system for primary diagnosis on April 12, 2017, and Paige Prostate as a Class II adjunct diagnostic software via the De Novo pathway on September 21, 2021. NCT06822842 is not registered as a device study subject to FDA regulation and does not include FDA, EMA, or PMDA approval or an advisory committee (AdComm) review. Therefore, short-term outcomes will focus on model performance and external validation data. Commercialization will require multi-center prospective validation, a quality system, and regional software medical device approvals.
From an investment perspective, the target of 2,000 cases represents a scale that can establish a barrier to entry in the rare pediatric solid tumor data space, but the pre-recruitment status and external validation plan are key variables in value assessment. For researchers, the design of a single non-interventional diagnostic study covering neuroblastoma, medulloblastoma, Wilms tumor, hepatoblastoma, and rhabdomyosarcoma provides a basis for validating inter-subtype transfer learning and fine-grained grade assessment. For the industry, it presents a pathway for specialized product development in pediatric cancers, in contrast to the general and adult cancer areas dominated by Paige Prostate and RuiPath in the USD 1.3 billion digital pathology market in 2025. The long-term value will be determined by factors such as accuracy metrics, multi-center reproducibility, reduction in pathologist reading time compared to experts, and achievement of FDA Class II-level clinical validation and medical device approval.
Source: ClinicalTrials.gov (api_ct)