Background

With an estimated 1.8 million new patients each year, breast cancer is the most common cancer in women worldwide [1]. In Italy, age-standardized (European standard) incidence and mortality rates in 2012 were 118/100,000 and 23/100,000, respectively, i.e., higher than those from other southern European countries [2].

Since 2005, triple-negative breast cancer (TNBC) identifies a specific subtype of breast cancer, characterized by the lack of expression of estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) [3]. TNBCs include a heterogeneous group of diseases which account for about 10–20% of all breast cancers and are more frequent among African American and Hispanic women, and in women with younger age, higher premenopausal body mass index, earlier age at menarche, and higher parity [3,4,5]. Moreover, they have higher expression of the Ki-67 antigen, higher mitotic index, and more frequent BRCA1 mutations. TNBCs are generally more aggressive than other breast neoplasms and have limited therapeutic options; therefore, they have usually a high risk of recurrence or death within 5 years since diagnosis [6].

Data on the clinical and pathological prognostic determinants for TNBC tumors are scanty and inconsistent and they generally derive from small hospital cancer registries including around a few hundreds of patients. TNM stage – including in particular the number of axillary lymph nodes involved – is one of the best-established prognostic factors for breast cancer, but its prognostic value in TNBCs, as in other intrinsic subtypes of breast cancer, is less clear [7, 8]. The ratio of positive lymph nodes on the total number of lymph nodes removed has been proposed as an additional and more accurate prognostic factor than the number of lymph nodes involved, although only a few studies have specifically evaluated its role in TNBC survival [9]. Although histologic grade has been shown to be a good predictor of survival for breast cancer, its prognostic role in TNBCs may be more limited given that most of these tumors are of high grade [10]. Furthermore, the findings on the prognostic value of the proliferation marker Ki-67 in TNBCs have been inconsistent [11]. Scantier data exist on tumor histotypes, tumor infiltrating lymphocytes (TIL), necrosis, and lymphovascular invasion (LVI) and survival from TNBC [12,13,14,12]. Accordingly, we did not find any difference in terms of recurrence and mortality between lobular and invasive ductal carcinomas, whereas other TNBC subtypes (mostly medullary and apocrine carcinomas) showed significantly better outcomes than invasive ductal carcinomas.

Among the best-established prognostic factors for breast cancer there is histologic grade [10]. This notwithstanding, in our large cohort of TNBC patients – as reported in a few other smaller studies [8, 26] – grade had no role in survival outcomes. This may be at least in part due to the high histological grade of TNBC patients [3], which might make it difficult to disentangle the role of grade on TNBC prognosis. Indeed, in our cohort only 1.5% of patients were G1 and over 3 out of 4 patients were G3.

Various large studies recently found that tumor TILs (mainly stromal) – a surrogate marker of adaptive immune response – is associated with a favorable prognosis in TNBC patients [39,40,41,42], although the use of TILs as an additional prognostic factors in TNBCs is not yet recommend giving the lack of standardization and clinical validation of this marker [13]. In our cohort, although death rates were lower among TNBC patients with TILs, no significant association was found between TILs and cancer progression or mortality. However, we had no information on the number/proportion of TILs and we did not specifically measured stromal TILs.

LVI – which refers to the invasion of lymphatic spaces and blood vessels – has long been considered a relevant prognostic marker of breast cancer, although it has not been incorporated in most internationally recognized staging system as the AJCC/TMN one [14, 21]. A few small studies which have investigated the relationship between LVI and DFS or OS in patients with TNBC showed that LVI is an independent predictor of poor outcome [14, 26, 43]. In our large cohort we found that LVI presence has a negative impact on both tumor recurrence and mortality when taking into account only study center and age at diagnosis. However, after allowance for other clinicopathological characteristics, the association between LVI and mortality was no more significant, thus do not supporting a relevant prognostic role of this marker.

Scanty data are available on the role of necrosis on prognostic outcomes in TNBC patients. In a study on 154 TBNCs from China, tumor necrosis was found to be a significant prognostic factor, although only results from univariate analyses were provided [15]. In our large cohort, necrosis at baseline was significantly associated to survival outcomes, even after allowance for other clinicopathological factors.

The results of this study should be interpreted after taking into consideration various limitations, mainly inherent to its retrospective design. Thus, we could not retrieve information on vital status at follow-up for 311 out of 1152 TNBC patients (about 27%) and we had to exclude them from the present analyses. Moreover, for some patients important clinical and pathological data were missing because not originally included in the medical records, and those missing information may have to some extent influenced the associations evaluated. Furthermore, some misclassification of patients may have resulted from the classification of tumors in different laboratories across hospital centers, where clinical and pathological testing practices can vary. However, pathology materials were reviewed centrally by three pathologists following the same national/international breast cancer guidelines in order to uniformely classify TNBCs across hospital centers and standardize ER, PgR and HER2 immunohistochemical results for TNBC samples, according to the ASCO/CAP recommendations [17].

The strengths of our study include its uniquely large sample size – including from one third up to half of all new Sardinian TNBC patients over the study period, the comprehensive and standardized nature of the registry database with patients’ characteristics, pathological tumor features, cancer treatments, and the complete ascertainment of patient status at regular follow-up intervals. This also allowed us to derive multivariate HR estimates for OS and DFS after allowance for a number of potential confounders.

Conclusions

In this uniquely large cohort, we provide further evidence that, besides tumor stage at diagnosis, lymph node ratio among lymph node positive tumors is an additional relevant predictor of mortality and recurrence in TNBCs, while Ki-67 seems to be predictive of mortality, but not of recurrence.