Minimum Number of Removed and Examined Lymph Nodes is Essential in Gastric Cancer Patients: Reply to Letter Corrado Pedrazzani • Daniele Marrelli • Armands Sivins • Franco Roviello Published online: 18 March 2010 Ó Socie´te´ Internationale de Chirurgie 2010 We thank Dr. Biondi and colleagues for their comments and questions, which allow us further clarification about this controversial issue [1]. The lymph node ratio has been extensively studied in gastric cancer, and several articles reported it as a more accurate prognostic factor with respect to the current UICC and JGCA classifications. However, most of these studies were performed at institutions with fair or good experience with gastric cancer surgery and pathologic assessment. In a recent report from the Italian Research Group for Gastric Cancer [2], we confirmed these results in a large cohort of gastric cancer patients operated on in six specialized Italian centers, where the quality of lymph node dissection, nodal retrieval, and pathologic assessment had been standardized for years. Indeed, patients treated by ‘‘limited’’ or ‘‘extended’’ lymphadenectomy had a median of 11 and 30 examined lymph nodes, respectively, and rarely fewer than 10. In patients with such numbers of removed lymph nodes, the lymph node ratio acts very well. The present study aimed to evaluate if the lymph node ratio is applicable and accurate similarly to the data from the Latvian Oncology Center, which resembles what may happen in nondedicated institutions around the world. In this series of 526 consecutive patients operated on between January 1999 and December 2005, a median of five lymph nodes were examined. We emphasize that ‘‘ratio’’ is a mathematic concept that is used to describe a clinical phenomenon; it includes a numerator (metastatic lymph nodes) and a denominator (examined lymph nodes). As the denominator decreases, the accuracy of the nominator in describing the clinical phenomenon (prognosis) progressively decreases. On the other hand, a prognostic factor may be clinically useful when it is able to allocate most patients into groups with either a good or a poor prognosis, thus classifying fewer patients than possible in groups with intermediate out- comes [3]. As we underlined in the discussion of our article, the key to the problem is not the small number of pNR1 cases but the large group of patients classified in the pNR2 group. Indeed, with the commonly adopted cutoff levels of the lymph node ratio that we used in our study (0, 1–25%, [25%), when 10 or fewer lymph nodes are examined, patients with 3 to 10 involved nodes are paradoxically classified in the same prognostic group (pNR2). Accord- ingly, this included an extremely heterogeneous group of patients (82% of the total number of node-positive cases). In contrast, the pNR1 group identified only a few patients (18%) with a better prognosis. On this basis, we said that the clinical utility of the lymph node ratio for identifying patients with a good or a poor prognosis is greatly pre- cluded. One alternative may be to adapt the cutoff levels of the lymph node ratio to the number of removed lymph nodes, but this may result in a complicated and confusing staging system. We also clearly stated in our study that the current TNM staging system (6th edition) had poor prognostic value when only a few lymph nodes are removed. Indeed, this classification requires that the pathologic assessment of the C. Pedrazzani (&) Á D. Marrelli Á F. Roviello Department of Human Pathology and Oncology, Unit of Surgical Oncology, University of Siena, Siena, Italy e-mail: pedrazzani@unisi.it A. Sivins Riga Eastern University Hospital, Latvia Oncology Center, Riga, Latvia A. Sivins University of Latvia, Riga, Latvia 123 World J Surg (2010) 34:1138–1139 DOI 10.1007/s00268-010-0524-2