Nottingham Prognostic Index in Triple-Negative Breast Cancer: a reliable prognostic tool?

Breast cancer comprises a complex and heterogeneous group of diseases at clinical, morphological and molecular levels [1–3]. It is clear that breast tumours of the same histological type show remarkably different clinical behaviour, which is probably a reflex of their distinct pattern of molecular aberrations [1, 4]. Microarray technology has changed the way we understand breast cancer classification by looking towards a molecular-based approach instead the traditional morphology and histopathological-based system [5, 6]. Pioneered by the Stanford group [6–9] and lately explored by several other groups, a new taxonomy for breast cancer based on expression profile has claimed that the morphological heterogeneity of breast cancer can be recapitulated and systematically classified at the transcriptomic level and into clinically meaningful groups [10–12]. Such studies have shown that the molecular profile of breast cancer present a systematic variation which allowed its differential identification into two distinct branches [11], the ER-positive branch, comprising the luminal A and B subtypes, and three ER-negative branch, which comprises at least, two reproducible subtypes: the HER2-overexpressing group and the basal-like group [6, 8, 9, 13–15]. An additional group of tumours displaying molecular features of normal breast tissue, and therefore named as "normal-like", has been also included in this ER-negative branch. However, it has been suggested that this group represents an artefact with high contamination from normal breast tissue rather than a distinct molecular subtype [14]. Among the molecular subtypes of breast cancer identified through gene expression profiling studies, none has generated as much interest or controversy as the basal-like breast cancer group (BBC) [1]. Recently, a review published by a large group of renown breast pathologists and clinicians, advocated that there is still no internationally accepted definition for BBC and discussed how best to define these tumours [1]. Nevertheless, it is commonly accepted that the term "basal-like" reflects the similarity of the protein expression profile of these tumours with the one of basal epithelial cells of the normal mammary gland [16–19], including high-molecular-weight cytokeratins (CK) 5/6, 14 and CK17, vimentin, P-cadherin, caveolins-1 and 2, αB-crystalin and fascin [8, 20–29]. BBC, the only group consistently defined by gene expression arrays [30], account for up to 15% of all breast cancers [1, 11]. These tumours frequently lack or show low levels of ER and PR, lack HER2 overexpression and amplification [21, 31, 32] and in approximately 85% of the cases display p53 expression by immunohistochemistry or TP53 mutations [8, 33]. Additionally, BBCs show exceedingly high levels of proliferation-related genes [6, 8, 9, 13] and express EGFR in a significant number of cases [21, 34]. Defined by microarray-based expression profiling or by panels of immunohistochemical markers as surrogates, BBCs are known by their clinically aggressive behaviour [8, 10, 21, 32].

Specimens that display BBC features (hormone receptors and HER-2 lack of expression), are called, in routine practice, as "triple-negative" breast cancer (TNBC). Controversial and provocative data has been recently published questioning whether TNBC and BBC are synonymous. Because a majority of BBCs are also TNBCs and approximately 80% of TNBCs are also BBCs [21, 35], it has been claimed that the TNBC and BBC are effectively synonymous [36, 37]. However, clinical, microarray and immunohistochemical data have shown that equating TNBC with BBC is misleading [15, 33].

TNBC patients lack the benefit of routinely available target therapy, which explains the undeniable growing attention of both pathologists and oncologists as an easily recognisable group of breast cancer with aggressive behaviour and poor therapeutic options [2, 38]. The prognosis of women with TNBC is significantly poor, compared to women with other subtypes of breast cancer. The higher recurrence and mortality rates of TNBC patients may be in part explained by different routes of metastatic spread [39]. There are conflicting results on the prevalence of lymph node metastasis at the time of diagnosis in TNBC patients [11]. Some studies described a higher prevalence of lymph node metastasis in TNBC [40], while others have found no statistical differences [3] or even an inverse association between TNBC and lymph node metastasis [41]. The currently accepted theory is that TNBCs seems to disseminate to axillary nodes and bones less frequently than the non-triple-negative cancers, presenting a preferential haematogenous route [32, 42–44] with a proclivity to develop metastatic deposits in the brain and lungs [11].

The Nottingham Prognostic Index (NPI) combines nodal status, tumour size and histological grade [45], reflecting metastatic behaviour, growth rate and genetic instability of breast cancers [46, 47]. Most importantly, as a continuous variable, NPI offers a responsive and sensitive means of modelling a continuum of clinical aggressiveness [46], indexing the outcome likelihood of invasive breast cancer patients [48]. NPI can define 3 subsets of patients with different probabilities of dying from breast cancer; good (≤3.4), moderate (3.41 - 5.4), and poor (> 5.4) prognosis groups [48]. Three factors, found to be independently associated with survival on multivariate analysis, were combined to give the NPI algorithm. One of these factors is the lymph node stage, which has traditionally been regarded as the most powerful prognostic factor in breast cancer. The greater the number of nodes involved, the worse the prognosis [45].

As above mentioned, TNBCs are believed to infrequently disseminate to axillary lymph nodes in favour of distant and visceral metastatic spread [32, 39, 42], an assumption that theoretically jeopardize the reliability of using NPI as a tool for TNBCs prognostication, since lymph nodal status is a major component for NPI calculation.

Herein, we investigated a large series of breast tumours and also a second cohort only composed by TNBC phenotype. These cohorts were used to test the clinical utility of NPI in predicting breast cancer patient outcome. Comparative analyses were performed within the TNBC subgroup of patients in order to evaluate the contribution of each Nottingham Prognostic components to the risk of worse survival and prognosis in TNBCs.

The prognosis of women with TNBC is significantly poor, compared to women with other subtypes of breast cancer. The underlying difference in recurrence and patient mortality rates may be explained in part by different routes of metastatic spread [51]. The current theory point out to the suggestion that TNBCs metastasize to axillary nodes and bones less frequently than the non-triple-negative subset of breast tumours, favouring a haematogenous spread [32, 42–44].

In terms of survival, it has been described that the survival curve shape for TNBC or BBC differs from that of patients with other types of breast cancer: there is a sharp decrease in survival during the first 3 to 5 years after diagnosis, but distant relapse after this time is much less common [31, 40, 41, 52]. In a study published by Dent and colleagues, the median time to death was 3.5 years for TNBC compared to 5.7 years for patients with other cancers [39]. In fact, as we can infer by the survival functions, TNBC experienced a severe decrease in their outcome before 48 months, a curve shape that overlaps with the one draw by the NPI in those patients. These findings reinforce the reliability of NPI as a tool to be reproducibly used in TNBC tumours.

In our cohort of TNBC we have found that tumour size is considerably higher compared with other subtypes of breast cancer. This difference in tumours size was strongly significant in relation to luminal, but not so marked when compared with HER2 tumours. Additionally, we found that, histologically, most TNBCs were high grade tumours. These results concerning tumour size and grade are largely in accordance with previous studies where these prognostic factors were studied within breast cancer subtypes, with special emphasis on TNBC versus non-TNBC patients [40, 50, 53, 54]. One of these studies used a notably large series of TNBC (6.370 patients) and non-TNBC (44.704 patients), and similar findings concerning histological grade and tumour size were found [53]. Interestingly, we found no difference regarding lymph node metastization between luminal, HER2-overexpressing tumours and TNBC. Thus, although it has been suggested that TNBC tend to disseminate in a lower frequency to lymph nodes, we found that 51.4% of TNBC developed metastasis to lymph nodes. This percentage shows that 1) the lymph node involvement in TNBC is as frequent as in other subtypes of breast cancer and, 2) the extension of this involvement do not differ between breast tumour subtypes. In the last 4 years, interesting studies have reported data concerning positive lymph node status in TNBC compared to non-TNBC, describing percentages of positive lymph nodes in TNBC ranging from 42.5% [55] to 54.4% [40], therefore, corroborating the results presented herein. Based on that, and even considering some putatively less prominent lymph node involvement in TNBC, which accordingly to our results was not observed, tumour size and grade variables are someway playing a compensatory score augmentation to NPI algorithm. Additionally, survival curves concerning tumour size and lymph node status demonstrated that these prognostic factors are equally important to stratify survival outcomes in non-TNBC as in TNBC. Taken together, and considering that the majority of TNBC lesions are grade III, these largely studied prognostic factors are reliable to be used in the assessment of NPI in TNBC. Interestingly, we also found a slight association between tumour size and lymph node status in TNBC. This result contributes to some controversy concerning the existence or not of a relationship between size and lymph node status, since some authors already argued a lack of association between these two prognostic factors in TNBC [40]. In our TNBC series, we have a proportion of 74% of BBC. The discrepancies about the association between tumour size and lymph node status could be explained if the TNBC cohort used by Dent et al. were enriched in basal-like breast tumours compared with the cohort we studied here. These hypothesis lay on a robust study recently published by Ellis group, where although a trend (non-significant) to display lymph node involvement with increasing tumour size was seen (especially in tumours larger than 4 cm), basal-like tumours do not seem to obey the "size-node" rule [56].

While basal-like tumours is a designation only revealed by gene profiling signatures [6], TNBC is a clinico-pathological classification which identifies a group of breast cancer with aggressive behaviour [2, 38], and where improvements on therapy development and prognostication are compulsory. In the study presented herein, we demonstrated that besides the fact that TNBC disseminate to axillary lymph nodes as frequently as luminal or HER2 tumours, we also showed that TNBC are larger in size compared with other subtypes and almost all grade III, therefore making truthful, at a clinical standpoint, the applicability of NPI as a prognostic tool in TNBC.