Multi-colour FISH in oesophageal adenocarcinoma-predictors of prognosis independent of stage and grade

Abstract

Background: Oesophageal adenocarcinoma or Barrett's adenocarcinoma (EAC) is increasing in incidence and stratification of prognosis might improve disease management. Multi-colour fluorescence in situ hybridisation (FISH) investigating ERBB2, MYC, CDKN2A and ZNF217 has recently shown promising results for the diagnosis of dysplasia and cancer using cytological samples.

Methods: To identify markers of prognosis we targeted four selected gene loci using multi-colour FISH applied to a tissue microarray containing 130 EAC samples. Prognostic predictors (P1, P2, P3) based on genomic copy numbers of the four loci were statistically assessed to stratify patients according to overall survival in combination with clinical data.

Results: The best stratification into favourable and unfavourable prognoses was shown by P1, percentage of cells with less than two ZNF217 signals; P2, percentage of cells with fewer ERBB2- than ZNF217 signals; and P3, overall ratio of ERBB2-/ZNF217 signals. Median survival times for P1 were 32 vs 73 months, 28 vs 73 months for P2; and 27 vs 65 months for P3. Regarding each tumour grade P2 subdivided patients into distinct prognostic groups independently within each grade, with different median survival times of at least 35 months.

Conclusions: Cell signal number of the ERBB2 and ZNF217 loci showed independence from tumour stage and differentiation grade. The prognostic value of multi-colour FISH-assays is applicable to EAC and is superior to single markers.

Figure 1

(A Oesophageal adenocarcinoma with HE staining (TMA): magnification 400 × , for verifying carcinoma cells against epithelium, low- or high-grade dysplasia and normal soft tissue in between tumour cells. (B) Oesophageal adenocarcinoma with HE staining (TMA): magnification 500 × , for counting and verifying carcinoma tissue and epithelium, low- or high-grade dysplasia and normal soft tissue. (C) Representative sample of EAC (TMA): FISH showing gain in ERBB2 and ZNF217, loss of CDKN2A and unspecific signals for MYC. Fluorescence in situ hybridisation–LSI probes for ERBB2 (green), MYC (blue), CDKN2A (red) and ZNF217 (gold). Note the gain in strong ERBB2 spots in comparison to the low number of ZNF217, indicating an unfavourable EAC with regard to our predictors P1–P3. (D) Tissue from normal cardiac mucosa with the FISH–LSI probes for ERBB2 (green), MYC (blue), CDKN2A (red) and ZNF217 (gold) as the control group and standard values.

Figure 2

Predictor 1 (P1): P1 discriminates patient survival the best in low-stage groups. Kaplan–Meier survival analysis of P1: percentage of cells with less than two signals for ZNF217. Best dichotomization of data for favourable outcome below a cutoff of 16%. Results grouped by clinical stage. Median survival in months for each category is shown in parentheses (log-rank test, P<0.0001).

Figure 3

Predictor 2 (P2): P2 is able to separate patients with low-stage tumours into groups with good and poor prognosis. Kaplan–Meier survival analysis of P2: percentage of cells with fewer ERBB2 than ZNF217 signals (ERBB2/ZNF217<1), with favourable outcome observed at or above a cutoff of 28%, grouped by clinical stage. Median survival in months for each category is shown in parentheses (log-rank test, P<0.0001).

Figure 4

Predictor 3 (P3): P3 is able to separate patients with low-stage tumours into groups with good and poor prognosis. Kaplan–Meier survival analysis of P3: the overall ratio of ERBB2 to ZNF217 signals, with favourable outcome observed below a cutoff ratio of 1.5, grouped by clinical stage. The late crossing of the curves in high-stage groups is owing to the small number of patients with these tumour stages. Median survival in months for each category is shown in parentheses (log-rank test, P<0.0001).

Figure 5

A combination of predictors is able to identify individuals with a poorer prognosis in low- and high-tumour stages (A), and by grade in the standard tripartite grading system (B). Prognosis grouped by histological grade or stage and FISH results, favourable if one of the three FISH predictors P1–P3=P is favourable in Kaplan–Meier survival analysis. Median survival in months for P for each category is shown in tables under the plots. Regarding stage (A), FISH results separate low stages IA and IIA significantly (log-rank test P<0.0001) but not higher stages (log-rank test, P=0.8060). Regarding grade (B), FISH results tend to separate G1 EAC (log-rank test, P=0.0586), G2 EAC not significantly (log-rank test, P=0.5596), but especially G3 EAC (log-rank test P=0.0007).

Figure 6

ZNF217/cell serves as a predictor and subdivides individuals into stage groups and within each grade according to prognosis. Kaplan–Meier survival analysis of the FISH parameter ZNF217/cell; definition and cutoff: ZNF217/cell with a cutoff of 2.8 and a favourable outcome above the cutoff, grouped by clinical stage (A), and histological grade (B). ZNF217/cell and stages I and IIA, III and IV (log-rank test, P<0.0001); ZNF217/ cell and grades 1–3 (log-rank test, P=0.0009). Median survival in months for each category is shown in tables under the plots.