Does immunostaining effectively upstage colorectal cancer by identifying micrometastatic nodal disease?

Purpose: Measure the association between the incidence of primary tumor staining and the identification of mediastinal lymph node (MLN) using cytokeratins, NM23, DCC-positive tumors, and vascular endothelial growth factor (VEGF) expression in T₂ and T₃/N₀ colorectal cancers. The impact of MLN on both recurrence and survival was assessed. Materials and methods: There were 153 CORC patients (T₂, T₃/N₀) selected from a prospectively accrued database. All patients had been staged by routine histopathology after a curative resection and no patients received adjuvant chemotherapy. The primary tumors (PT) were assessed with a panel of immunohistochemical stains (cytokeratin, DCC, Nm23, and VEGF). If the PT was positive, the regional nodes were assessed with that marker(s). For any positive tumor marker, all lymph nodes (LNs, mean of 12.6 ± 4.2) were stained for this marker. Results: Patient age ranged from 38 to 86 years with amean age of 61.56 ± 25.56 years. Mean follow-up was 72.1 ± 32.4 months. Recurrence rate of the whole group was 19/153 (12.4%) and the mean time to recurrence was 37.6 ± 23.6 months (15 to 77 months). Crude mortality was 39.9%, while the cancer specific mortality was 11.2% after the whole follow-up period. The relationship between PT staining and MLNs was: cytokeratin-PT 143 (93.5%)/MLN 9 (6.3%); NM23-PT 51 (33.3%)/ MLN 3 (5.9%); DCC-PT 79 (53%)/MLN 3 (3.8%); and VEGF-PT 72 (47%)/MLN 4 (5.6%). Nineteen (12.4%) patients experienced tumor recurrence. No correlation exist between PT and/or MLN staining and either recurrence or survival. No patient with MLN with any stain experienced a recurrence. There was no advantage to using an individual stain or all four stains. Conclusion: Immunohistochemical stains for PT and focused analysis of regional nodes did not improve prediction of survival or recurrence. Sentinel LN evaluation and the provision of adjuvant chemotherapy in node-negative patients should be questioned and not be utilized outside of a research protocol.