{
  "abstract": "Background Accurate preoperative staging is fundamental in gastric cancer because it determines eligibility for endoscopic or surgical treatment and the appropriate use of neoadjuvant therapy. Because management decisions hinge on depth of invasion (T stage) and nodal status (N stage), staging errors can lead to over- or under-treatment. Endoscopic ultrasound (EUS) is widely used for locoregional assessment, but reported diagnostic performance varies across centres, operators and technical approaches. We aimed to estimate pooled EUS accuracy for overall T and N staging, accuracy by T substage (T1 to T4), and rates of T over- and under-stagingMethods We conducted a systematic review and meta-analysis according to the PRISMA guidelines. On 2 April 2025, a search of PubMed, the Cochrane Library, and Scopus was carried out for studies published from between January 2010 and 2 April 2025 in English (including ≥ 15 patients) that reported EUS staging accuracy (defined as the proportion correctly staged against histopathology) in histologically confirmed gastric cancer, without neoadjuvant treatment. Study selection was managed using Rayyan, and methodological quality and risk of bias were assessed with the QUADAS-2 tool. Data were extracted into Excel and synthesized in R (using the Metafor package) with a random-effects model to calculate pooled proportions with 95% CIs and to quantify heterogeneity (I2, Tau2). The results included overall and tier-specific T-stage accuracy (T1–T4), overall N-stage accuracy, and T over- and under-staging rates.Results Out of an initial 5,432 records, 20 studies were included. The total pooled accuracy of T stage assessment by EUS was 65.4% (95% CI 0.59–0.70), with consistently high heterogeneity (I2 94.4%). The tier-specific accuracies were: T1, 66.6%; T2, 52.6%; T3, 69.4%; and T4, 41.2%. The total accuracy of N-stage assessment was 73.9% (95% CI 0.67 to 0.79; I2 85.3%). The pooled proportions for both T over-staging and under-staging were 16.5% and 17.2% respectively. Analysis of study-level profiles indicated moderate sensitivity and higher specificity (60.3% and 88.6%) respectively, reflecting a pattern of centre-dependent variation in over or under-calling.Conclusions EUS demonstrates fair to good overall T-staging accuracy in gastric cancer, with highest for T1 and T3, and lowest for T4 followed by T2. The accuracy of N-staging is modest (73%) and appears to improve when EUS-guided tissue acquisition (FNA/FNB) complements morphological assessment. The substantial heterogeneity observed highlights the need for standardised EUS performance and reporting metrics such as blinding, threshold criteria, and probe specifications to improve reproducibility. These findings support the current consensus that EUS should be integrated with complementary modalities (e.g., CT, diagnostic laparoscopy with cytology) within comprehensive, multi-modal staging protocols.",
  "authors": [
    {
      "affiliations": [
        "University of Liverpool, Liverpool, United Kingdom"
      ],
      "name": "Mario Saucedo Padilla"
    },
    {
      "affiliations": [
        "University of Liverpool, Liverpool, United Kingdom"
      ],
      "name": "Mushfique Alam"
    },
    {
      "affiliations": [
        "University of Liverpool, Liverpool, United Kingdom"
      ],
      "name": "Nathan Stephens"
    },
    {
      "affiliations": [
        "Royal Liverpool University Hospital, Liverpool, United Kingdom"
      ],
      "name": "Thomas Skouras"
    },
    {
      "affiliations": [
        "University of Liverpool, Liverpool, United Kingdom"
      ],
      "name": "D Mark Pritchard"
    }
  ],
  "title": "O18 Utility and limitations of endoscopic ultrasound for preoperative T and N staging of gastric cancer: a systematic review and meta-analysis",
  "uid": "f56d79bf-d38b-59e9-bea8-3d624999066d"
}