BACKGROUND The number or ratio of lymph node metastases detected by hematoxylin & eosin (H&E) staining is the most important predictor of survival in esophageal cancer. The survival effect of lymph node metastases detected on immunohistochemistry (IHC) is controversial. My colleagues and I hypothesized that the extent of nodal disease determined by both H&E and IHC examination would more accurately predict survival than either technique alone. METHODS The study population consisted of 37 patients who underwent en bloc esophagectomy as primary therapy for esophageal adenocarcinoma 5 or more years ago. All had mediastinal and upper abdominal lymphadenectomy. No patient received neoadjuvant or adjuvant therapy. Tissue blocks were sectioned for H&E staining to confirm the initial histology, and a second slide was stained with monoclonal antibodies AE1 and CAM 5.2, which are directed at a number of cytokeratin antigens. The slides were reviewed by an investigator blinded to clinical outcome. The effect of IHC staining on prognosis was assessed by comparing 5-year survival based on H&E and IHC findings. RESULTS A total of 1,970 nodes were examined in the 37 patients. Routine H&E staining detected metastases in 29 patients (78%); the remaining 8 with N0 disease all survived at least 5 years after operation (median not reached). In the 29 patients with N1 disease, survival was 41% at 5 years. In 20 of the 29 N1 patients, metastases were detected by H&E in less than 10% of the nodes removed; 55% of the patients survived 5 years, and 39% survived 8 years. Nine of the 29 patients had metastases detected in more than 10% of the nodes removed, and all died at a median of 17 months. IHC staining was performed on the nodes from the 8 N0 patients and the 20 patients with less than 10% nodal involvement (a total of 28 patients). Additional nodal metastases, not identified on H&E examination, were found in 51 nodes from 17 patients (60.7%). Of the 8 patients who were node negative on H&E examination, 3 had metastases detected by IHC, and all survived 5 years or more free of disease. Of the 20 patients with less than 10% nodal metastases on H&E, 14 (70%) had additional metastases detected by IHC (median, 2 nodes per patient). When combined with the results of H&E staining, the node ratio remained less than 10% in 13 patients and exceeded 10% in 7. Survival in patients whose ratio remained less than 10% was significantly better than in those whose ratio exceeded 10% (actual 5-year survival, 77% vs 14%; chi2 = 4.662; p = 0.03). CONCLUSIONS IHC staining techniques can identify nodal metastases missed by routine H&E examination in a large number of patients. The combination of H&E and IHC examination is useful in patients with less than 10% nodal involvement by H&E examination in that IHC detection of micrometastases allows classification into low-risk (> 75% survival) and high-risk (< 15% survival) groups. IHC-detected micrometastases are not of prognostic importance in N0 patients or those with greater than 10% nodal metastases on H&E.