GATA3 as an Adjunct Prognostic Factor in Breast Cancer Patients with Less Aggressive Disease: A Study with a Review of the Literature

Background: GATA binding protein 3 (GATA3) expression is positively correlated with estrogen receptor (ER) expression, but its prognostic value as an independent factor remains unclear. Thus, we undertook the current study to evaluate the expression of GATA3 and its prognostic value in a large series of breast carcinomas (BCs) with long-term follow-up. Methods: A total of 702 consecutive primary invasive BCs resected between 1989 and 1993 in our institution were arranged in tissue microarrays, immunostained for ER, progesterone receptor (PR), ki-67, HER2, p53, and GATA3, and scored. Clinico-pathological data were retrospectively collected. Results: GATA3 was evaluable in 608 (87%) of the 702 cases; it was positive in 413 (68%) cases and negative in 195 (32%) cases. GATA3 positivity was significantly associated with lower grade (p < 0.0001), size (p = 0.0463), stage (p = 0.0049), ER+ (p < 0.0001), PR+ (p < 0.0001), HER2− (p = 0.0175), and p53 wild-type pattern (p < 0.0001). The median follow-up was 183 months, GATA3 positivity was associated with better overall survival (HR 0.70, p = 0.001), and its prognostic value was retained in a multivariate analysis. The association with better overall survival was stronger in patients with grade 1–2, pT1–2, pN0, stage I–II, ER+, PR+, ki-67 < 20%, HER2−, a wild-type p53 immunohistochemical pattern, and in luminal B BC. Conclusions: Our findings indicate that GATA3 is a positive prognostic marker in BC patients, especially in patients with biologically less aggressive BC. Incorporating GATA3 immunohistochemistry into routine practice could help further stratify BC patients for their risk.

ranged from 0 to 300 for each core. Except for p53, scoring results were dichotomized into either negative or positive, using the pre-defined threshold values dictated by the 12th St. Gallen International Breast Cancer Conference and reported in Table S1 (1). The membrane staining HER2 was considered either negative or positive according to conventional guidelines (2,3). Differently, p53 was considered to have either a mutated pattern, when it was completely negative (null pattern) or with at least 60% of BC cell nuclei showing intense positivity (missense pattern), or a wild-type pattern when the tumor showed a variable weak-moderate positivity in 1%-59% of cells (4).

Statistical analysis
The features found to be significant in univariate analysis were assessed for the multivariate analysis using enter logistic regression model, to evaluate which features were independent. For multivariate analysis, we compared the log-log survival curves and the curves predicted by the Cox model with the observed ones according to the Kaplan-Meier method to check graphically the proportional hazards assumption for all variables. The study time endpoint was evaluated starting at total overall survival followup (28 years) with a progressive 5-years reduction until 5-years follow-up, then 1-year by 1-year time interval. We used Cox proportional hazards modeling and the likelihood ratio to evaluate survival differences between the different groups (backward parametric statistical Wald method). A setup procedure was used and variables were added to the model if the two-sided significance level was <0.1 in univariate analysis. To control for potential confounding factors, we adjusted HR estimates per age. To evaluate the effect of single variables on patient outcome, the endpoint for overall survival was considered any death irrespective of cause. Patients without an adverse event were censored at the time of the last follow-up.
A total of 31 TMA blocks were built for this study. Overall, of the 702 BC cores arranged in the TMA, an average of 655 (93.3%) cores per antibody were scorable, whereas on average 47 (7.7%) cores per antibody were unscorable, due to tissue loss, unrepresentative tissue, excessive tissue folding, or non-specific staining per IHC staining. Altogether, a total of 3930 TMA cores were suitable for IHC evaluation in this study.

Overall, ER and PR were expressed in a higher percentage of BC cells compared with
Ki-67 and p53 (85%, 57% vs 7%, 13%, respectively). Moreover, the majority of BC cases were positive (≥1%) for ER (81.5%) and PR       Figure S1. Pie chart representing the distribution by and molecular subtypes of our series of breast carcinomas (left panel). Box-plot (middle panel) and cumulative relative frequency charts (right panel) show that GATA3 histological score was significantly higher in luminal intrinsic BC subtypes (p<0.0001) when compared to HER2-positive (+) and triple-negative (TNBC) subtypes.

Figure S2
Figure S2. Kaplan-Meier overall survival curves of subgroups of breast cancer patients according to IHC expression of GATA3. After adjusting for the patients' age, GATA3 IHC positivity is associated with a significant better overall survival in breast carcinoma patients with histological grade 1 and 2, pT1-T2, pN0 and Stage I and II. The age-adjusted hazard ratio (HR) for death and the 95% confidence interval (CI) estimated with Cox regression analyses are reported.