Jiang et al provided an interesting overview of the prevalence of GATA3 (GATA-binding protein 3) mutations in a cohort of Chinese patients with breast cancer and a comparison of their clinical features with GATA3-mutant tumors collected by The Cancer Genome Atlas (TCGA). Given the prevalence of GATA3 mutations among patients with estrogen receptor (ER)-positive breast cancer (approximately 15% in prior studies), retrospective studies such as this will be very important in informing our understanding of the contribution of these mutations to breast cancer pathobiology and patient outcomes.
We have identified 2 significant areas of concern in the article by Jiang et al that would benefit from further clarification from these authors. First, the mutation profile identified in their cohort is strikingly, perhaps even implausibly, different from that observed in larger well-controlled studies. Second, their claims regarding the relationship between GATA3 mutation status and patient survival in the TCGA study are simply not correct.
Prior analyses of GATA3 mutations in > 1000 cases from several large cohorts[2-5] have indicated that the vast majority of mutations (> 98%) affect exons 5 and 6, which encode a region including the second of 2 zinc fingers. Given the overwhelming bias toward mutations in this region as observed in previous studies, we were astonished to note that only one-half of the mutations observed by Jiang et al affected this region of the protein. In addition, these prior studies have identified hotspot mutations resulting in disruption of a splice site at codon 308 and causing a frameshift at codon 409fs. In the latest data from TCGA, 28 of 85 GATA3 mutations (32.9%) affect 1 of these 2 hotspot loci. It is striking that not a single patient in the Fudan University Shanghai Cancer Center cohort had a mutation at these loci. The lack of information in the report concerning the sequencing primers used makes it impossible to determine whether the sequencing strategy was sufficient to detect these mutations, if indeed they existed in the studied Chinese population.
Another interesting disparity between the study by Jiang et al and prior investigations is the high prevalence of point mutations in the GATA3 coding sequence in the Fudan University Shanghai Cancer Center cohort, including a recurrent L190M mutation in 10 of the 46 GATA3-mutant cases, an unprecedented frequency of 21.7%. Prior studies have generally found frameshift mutations to be predominant and, when they occur, point mutations tend not to accumulate at hotspots (with the possible exception of mutations at M294 that have now been noted in 3 cases). It is remarkable that from such a relatively small number of cases, these investigators managed to identify an apparently new hotspot at L190M.
The authors correctly note that prior sequencing studies were performed primarily in Western populations. Although this is true, 57 Asian females were included in the most recent breast cancer data from TCGA. Of these, 36 had ER-positive breast cancer, including 8 patients with GATA3 mutations. It is important to note that these 8 GATA3-mutant tumors included 2 cases with the 308 splice site and 2 with 407fs hotspot mutations, whereas the remaining 4 GATA3 mutations affected codons 327 to 433 (encoded by exons 5 and 6), a pattern that is not dissimilar to that found in other ethnic groups. These data suggest that that invoking putative fundamental biological differences between breast tumors of Asian and Western women is not a plausible explanation for the GATA3 mutation profiles reported by Jiang et al.
The authors claim, without providing data, that GATA3 mutations were associated with significantly better outcomes among the patients in the TCGA study with ER-positive disease (P = .041; data not shown). We have tried and failed to replicate this using data from the cBioPortal, and downloaded directly from TCGA. In both cases, we found no statistically significant difference in overall survival between patients with GATA3-mutant and GATA3 wild-type ER-positive tumors with the available length of follow-up. Thus, contrary to the claim in the article by Jiang et al, we do not believe that, as of this date, there is any evidence from TCGA data to suggest that GATA3-mutant tumors are associated with a better outcome among patients with ER-positive breast cancer.
In conclusion, given the significant disparity between the mutational spectra reported by these investigators and the strong consensus in mutations from TCGA and other studies,[3-5] we believe that the study by Jiang et al should be interpreted with caution until such time as more detailed information is provided by the authors and/or confirmatory data are available from other independent cohorts. In particular, the claim by the authors that GATA3 mutations may identify a subgroup of patients for whom limited therapy may be appropriate should be viewed with great caution.