The treatment of TNBC, therefore, remains a Nutlin-3 difficult challenge in clinical practice. TNBC comprises approximately 10–16% of breast cancer cases. The main characteristic of TNBC is that it frequently affects younger patients, occurring predominantly in premenopausal women. The molecular mechanisms of TNBC still remain unclear, although their association with poor prognosis is thought to be due to aggressive biology and resistance to presently available endocrine therapies, agents targeting HER2 pathways, and standard cytotoxic chemotherapies. Recent evidence supports the idea that the epidemiology and prognosis of breast cancer differs between races, most likely due to different genetic compositions. Discrepancies in the prognosis of TNBC between Western and Asian populations in Taiwan were specifically noted. However, few studies have investigated the genetic differences between breast cancers from Caucasian and Asian populations, let alone lower incidence of TNBC than other subtypes. Since the advent of microarray chips, the mechanisms of breast cancer have been studied intensively, such that the subtype of breast cancer can be identified by its gene expression profile. Breast cancer gene expression profiles have been identified across different microarray platforms by different research groups. Several prediction models have been proposed, such as the 70- gene profile, two-gene ratio, or singular value decomposition, to predict lymph node metastasis. The 70-gene profile for disease outcome prediction was a pioneering study, and has since been verified in several other studies. It not only predicts outcomes effectively, but also outperforms other methods based on clinical parameters. Other studies using microarray gene expression profiles for clinical outcome prediction also provide satisfactory results. The use of microarray chips has proven to be a useful strategy to detect candidate genes and pathways involved in tumor progression. The aforementioned success, however, has not been observed for TNBC patients, especially in the prediction of recurrence. Difficulties have arisen because of its relatively complex etiology, and because of the deficiency of TNBC samples. Yet, knowledge of the genes associated with recurrence of TNBC is desperately needed for designing prediction models and treatment strategies, possibly including targeted therapy. This study used a Cox proportional hazards model to predict TNBC recurrence in a Taiwanese population. We compared the expression profiles of breast cancers from 185 Taiwanese patients to profiles from a Caucasian population. The microarray results revealed differences in TNBC gene expression profiles in different ethnic groups. Pathway analysis showed that several canonical pathways, such as cAMP-mediated signaling and ephrin receptor signaling, are activated in association with recurrence in TNBC. Furthermore, six prognostic genes were identified for predicting the risk of recurrence of TNBC in Taiwanese patients.