Safety of endoscopy in cancer patients with thrombocytopenia and neutropenia

Background and Aims: Cancer patients are prone to thrombocytopenia and neutropenia, which increase the risk of bleeding and infection. We assessed the safety of endoscopic procedures in cancer patients with thrombocytopenia and/or neutropenia. Methods: We studied consecutive cancer patients with thrombocytopenia and/or neutropenia who underwent endoscopic procedures from 2010 through 2015. Neutropenia was defined as an absolute neutrophil count (ANC) < 1000 cells/mu L, and thrombocytopenia as a platelet count < 100 x 10(3)/mu L. Univariate and multivariate generalized estimating equation models were used to assess factors associated with risk of adverse events (AEs) or death. Results: We identified 588 patients who underwent 783 procedures; 608 procedures were performed in the setting of thrombocytopenia and 675 procedures in the setting of neutropenia. Concurrent neutropenia and thrombocytopenia were recorded in 500 endoscopies. Twenty-four patients (4.1%) experienced infectious AEs, whereas 29 (4.9%) experienced bleeding AEs within 1 week of the procedure. On multivariate analysis, platelet count <= 50 x 10(3)/mu L was associated with risk of bleeding AEs. In contrast, poor performance status was associated with increased risk of infection AEs (P <.01). No association was observed between low ANC and infectious AEs. Poor performance status (P <.01) and platelet count <= 100 x 10(3)/mu L (P <.05) were associated with increased risk of 30-day mortality. A persistent platelet count < 20 x 10(3)/mu L after the procedure, with a baseline platelet count of <= 20 x 10(3)/mu L before the procedure, was associated with significant risk of bleeding AEs compared with a platelet count > 20 x 10(3)/mu L after the procedure (P <.01); furthermore, if the platelet count increased to > 50 x 10(3)/mu L after the procedure, the bleeding risk after the procedure was greatly reduced (P <.01). Conclusions: Endoscopic procedures are relatively safe in cancer patients with platelet count > 50 x 10(3)/mu L. Nevertheless, a platelet count of >= 20 x 10(3)/mu L could be an appropriate threshold for platelet transfusion if 50 x 10(3)/mu L is difficult to achieve. The functional status of the patient, in the absence of the need for urgent or necessary endoscopic interventions, should be considered when deciding whether to perform endoscopy. The risk of procedure and the ANC did not seem to affect the outcomes.