Medical records of patients with locally advanced esophageal cancer treated at Roswell Park Cancer Institute (RPCI) between January 1, 1996 and December 31, 2006 with neo-CRT followed by esophagectomy were reviewed and analyzed retrospectively. All who attained CPR-HE after neo-CRT (n=18) were studied further with CK staining to detect any microscopic residual disease to determine a CPR-CK for the purpose of our pilot study. Patients were followed till the last available follow up in the medical records or till they progressed, whichever was earlier.

Standard HE staining was performed on pathology specimens (which included the entire gastro-esophageal junction) from these 18 patients and reviewed by a surgical pathologist with special interest in gastro-intestinal malignancies (C.L.). One tissue block per patient with representative tissue was selected for immunohistochemical staining after review of all available blocks. In this study, the same block as examined by HE was used for immunohistochemical stains. To perform immunohistochemical stains, deeper sections were taken. The formalin fixed, paraffin wax embedded sections were dewaxed. To enhance immunostaining, these sections were subjected to enzyme digestion for epitope retrieval using protease 1 (Ventana Medical Systems, Tucson, Arizona, USA). Endogenous peroxidase and non-specific binding were blocked before addition of the primary antibody. Sections were then immunostained using the DAKO autostainer (Carpentaria California) immunohistochemistry system; using antibodies to CK5/6 for SCC (DAKO, 1:100 for 30 minutes) and AE1/3 for AC (DAKO, 1:100 for 30 minutes). Monoclonal mouse (IgG) antihuman antibodies to specific cytokeratins were used. The appropriate biotinylated secondary antibodies were detected using an iVIEW DAB (3, 3-diaminobenzidine tetrahydrochloride) conjugate. The complex was visualised with DAKO Mouse Envision Kit.

Patients with no evidence of CK positive residual cancer cells were defined as CPR-CK. ‘True CPR' was defined as HE negative and CK negative for evidence of residual disease (HE- and CK-).

To study the statistical relationship between pairs of nominal variables, Fisher's exact test was used. The exact Wilcoxon test was used to study the statistical relationship between nominal and ordinal variables. A 0.05 nominal significance level was used in all testing. Survival analysis was done for purely exploratory purpose. Overall survival (OS) and progression free survival (PFS) distributions were calculated by the Kaplan-Meier method [₁₅]. Due to small sample sizes, a permutation test based on the log-rank statistic was used to test the significance with respect to the survival distributions [₁₆]. All statistical analysis were done as a secondary goal and on a purely exploratory basis.

Eighteen patients were found to have CPR-HE after neo-CRT followed by Ivor-Lewis esophagectomy for locally advanced EC between January 1, 1996 and December 31, 2006 at RPCI. Sixteen patients had AC and two had SCC. All patients were males. Median age at diagnosis was 58.5 years (range- 50 to 74 years) while mean age was 59.4 years. Patients received protracted infusion of 5-fluorouracil with oxaliplatin or cisplatin along with a total of 50.4 Gy of radiation in the neoadjuvant setting during this time period. One patient received 3 cycles of MIMIC (mitomycin, ifosamide and cisplatin) at an outside facility prior to transfer of care at RPCI. Two other patients received capecitabine along with oxaliplatin while another two received cisplatin along with irinotecan as neoadjuvant therapy. Table 1 summarizes baseline patient demographics.

Figure 1

Table 1: Patient Demographics and Treatment Received

Asterisk * denotes patients with signet ring cell feature on pathology. Abbreviations used: AC- adenocarcinoma; Poor AC- poorly differentiated AC; mod AC- moderately differentiated AC.

Results of CK staining of the pathological slides of esophagectomy specimens are shown in Table 2.

Figure 2

Table 2: Staining pattern and survival data (based on last follow up at RPCI).

All patients had CPR-HE. Abbreviations used- NA- not applicable, NED- no evidence of disease, AWD- advanced.

Seven patients (39 %) had CK+ staining. On second look HE examination of these cases, 3 out of the seven (43 %) were found to have malignant cells. CK staining aided in detection of subtle residual disease missed on original HE in these patients (Figure 1).

Figure 3

Figure 1: Resection specimens deemed to have a complete pathological response were stained with either cytokeratin AE 1/3 (for cases with a biopsy diagnosis of adenocarcinoma) or cytokeratin 5/6 (for cases with a biopsy diagnosis of squamous cell carcinoma). A 100X magnification of a hematoxylin and eosin stained section with arrows pointing to groups of atypical cells. These cells are not diagnostic of residual carcinoma by H&E stain alone. A 100X magnification of a cytokeratin AE 1/3 stain of the same area shown in panel A, which demonstrates residual adenocarcinoma.

After a median follow up of 21.7 months (range 7.8 to 107.2 months), there were 5 recurrences (patients # 6, 7, 9, 12, 13). Four had distant metastasis and one had nodal recurrence (#7). Two patients died of advanced disease. The remaining 16 patients were still alive with follow-up ranging from 7.8 to 107.2 months. The estimated PFS for all 18 patients was 69.7% at 12 months.

Seven patients had residual disease detected on CK staining (# 3, 5, 6, 9, 10, 14 and 18). All had an initial diagnosis of AC. Median age of this group was 58 years (Mean = 59.7 years, range- 53 to 74 years). Of the seven CK+ patients, three (43%) had systemic relapse. This represented 60 % of the patients with systemic relapse (# 6, 9 and 13). The other four CK+ patients remained NED at 37, 51, 46, and 23 months, respectively. Median PFS was at least 11.7 months in the CK+ group.

The remaining 11 patients had negative CK staining. Second look HE did not show any evidence of residual disease in this subset. Median age of this population was 59 years (Mean= 59.2 years, range-50 to 71 years). All but two patients had NED status at last follow-up. One patient progressed at 10.4 months with nodal recurrence (# 7) and the other had systemic disease (# 11). Median PFS was at least 10 months in the CK- group (sample size inadequate for statistical analysis).

There was no significant difference in the PFS and OS distributions between the CK+ and CK- groups. There was no statistically significant association between CK staining and clinical stage, age, histology, or lymph node status. On further analyses, there was no difference between patients with CK-/second look HE + (residual disease on repeat HE staining), CK-/HE- (true CPR) and CK+/HE- (patients with residual disease on IHC but not on repeat HE). Again, the study is limited by small sample size.

We would like to acknowledge Paul Lin, MD for his assistance in clinical data collection and Ms Deborah Driscoll for her valued comments and review of the manuscript.