Neutrophil-lymphocyte ratio and platelet-lymphocyte ratio as prognostic factors in non-metastatic breast cancer patients from a Hispanic population

Introduction

Breast cancer is the most frequent malignancy among women worldwide. According to the GLOBOCAN Registry, this tumor ranks as the second cause of cancer death in women from developed countries and the first in less developed regions [1]. For this reason, the identification of patients with poor prognosis through easy-available and inexpensive methods is warranted in order to choose an appropriate therapy and follow up.

It has been acknowledged that immunological response is associated with clinical outcomes in cancer patients [2]. Indeed, inflammatory conditions in the tumor microenvironment can enhance the proliferation and survival of malignant cells, favor angiogenesis and promote tumor invasiveness [3]. For instance, tumor-infiltrating lymphocytes and tumor-associated neutrophils have been recognized as prognostic variables in patients with solid tumors [4]. Similarly, circulating neutrophils, lymphocytes, and platelets have been associated with poor outcomes in patients with cancer [5].

The neutrophil:lymphocyte ratio (NLR) and the platelet:lymphocyte ratio (PLR) have been established as independent prognostic factors in a large list of solid tumors, with scarce data regarding breast cancer patients [6,7]. Furthermore, the prognostic role of these variables from Hispanic population is lacking. It has been demonstrated that Hispanic and Black patients show different blood counts of neutrophils and lymphocytes in comparison with White counterparts [8,9]. In addition, recent trials have determined that the relationship between overall survival (OS), disease-free survival (DFS), NLR, and PLR partly depends on ethnicity [10]. Therefore, we aimed to determine the prognostic value of the NLR and PLR to predict DFS and OS among breast cancer patients from a Hispanic background (Costa Rica) and to establish the best cutoff to properly classify these patients according to their risk or poor outcomes.

Material and Methods

This was an observational retrospective study, carried out in 172 consecutive patients diagnosed with invasive, non-metastatic breast cancer from January 1^st to December 31^st 2010 that were managed in the San Juan De Dios Hospital at San José, Costa Rica. Patients were treated and followed-up in the Breast Cancer Unit according to clinical guidelines [11].

Clinical records were reviewed to collect the following variables: age, comorbidities, clinical stage, tumor grade, expression of estrogen receptor (ER), progesterone receptor (PgR), and human epidermal growth factor receptor 2 (HER2). The assessment of these receptors was done by immunohistochemistry. Fluorescent in situ hybridization for HER2 amplification was performed on cases showing 2+ of HER2 staining. Tumor stage was on the basis of criteria of the Seventh American Joint Committee on Cancer [12]. Histologic type and grading followed the 2003 World Health Organization classification [13].

Patients with hematological disorders, corticosteroid use or any acute or chronic inflammatory disease were excluded. The NLR (defined as the absolute neutrophil count divided by the absolute lymphocyte count) and PLR (defined as the absolute platelet count divided by the absolute lymphocyte count) were calculated from full blood test performed prior surgery or neoadjuvant chemotherapy. The study was approved by the Institutional Review Board.

Results

Table 1 shows the main characteristics of the studied population. The median pre-treatment NLR and PLR were 1.62 (range: 0.2–11.7) and 130.21 (range: 43–964), respectively. After statistical analysis, the optimal cutoff point was set at 3 for NLR (corresponding to a log-likelihood of 19.51) and 250 for PLR (corresponding to a log-likelihood of 21.89).

Patients with NLR greater than or equal to 3 (n = 26) and PLR greater than or equal to 250 (n = 13) had more triple negative tumors (ER-/PgR-/HER2-) than their counterparts. Similarly, patients with PLR less than 250 had more ER+ tumors than patients with PLR greater than or equal to 250.

The median follow-up time was 79.3 months (range: 2–90 months). During this period there were 39 events of disease recurrence and 37 deaths. Overall, the 5-year overall survival probability was 88.4%, 82.5%, and 63.0% for clinical stage I, II and III, respectively.

The probabilities of DFS and OS according to the NLR and PLR cutoff values are shown in Figs 1–4. Both NLR ≥ 3 and PLR ≥ 250 were associated with worse DFS and OS. Among patients with NLR ≥ 3 the 5-year OS probability was 47.8% versus 84.5% in patients with NLR < 3. Similarly, among patients with PLR ≥ 250 the 5-year OS probability was 38.5% versus 83.1% in patients with PLR < 250.

The multivariate analyses did not show evidence of interaction between NLR and PLR. After adjusting for potential confounders (breast cancer tumor subtype and clinical stage), a PLR ≥ 250 was independently associated with DFS and OS (Table 2). Similarly, clinical stage and triple negative tumors (defined as the absence of estrogen receptor, progesterone receptor and HER2) were associated with poor outcomes.

Discussion

The role of inflammatory cells in the pathogenesis of cancer has been widely explored by several authors. For example, Coffelt and colleagues revealed that neutrophils enhance the malignancy development from tumor initiation to metastatic spread [15]. Similarly, both tumor infiltrating lymphocytes and circulating lymphocytes have been considered prognostic factors in breast cancer patients [4]. Through recognition of tumor-specific antigens or tumor-associated antigens adaptive and innate immune cells play an important role in the development of cancer [16]. Platelets have also been implicated in cancer progression by promoting angiogenesis through the release of vascular endothelial growth factor (VEGF) or by adhering to tumors cells and facilitating metastasis from bloodstream [17].

Although the previous observations can explain the role of inflammatory cells in cancer pathogenesis, the exact mechanism between high NLR and PLR, and poor outcomes remains unclear. It is believed that these indexes are a reflection of a systemic inflammatory response induced by several cytokines such as IL-6, IL-7, IL-9 among others [18]. Despite the absence of a clear mechanism, it has been widely demonstrated that these two parameters are associated with poor outcomes. Our findings are in accordance with previous studies that have correlated high NLR and PLR with poor outcomes in patients with other solid tumors [4,5]. According to a recent meta-analysis this association is greater for metastatic than nonmetastatic diseases and it is independent of primary tumor location and clinical stage [5].

To our knowledge this is the first study to show a significant association between high NLR and PLR and worse outcomes in Hispanic patients with non-metastatic breast cancer. Although previous reports have shown this association in several solid tumors [4,5], there is relatively scarce data regarding the prognostic value of these variables in breast cancer patients with conflicting results [19]. The majority of the aforementioned studies have been carried out in different ethnicities, especially Asians and White patients with little research done in Black and Hispanic patients who are known to have lower NLR and PLR than the former ethnicities. Besides, the survival of Asian and White patients with breast cancer is longer than that of Hispanic women [1]. Therefore, our findings add valuable evidence to consider these biomarkers as prognostic variables in our population.

Through this study we have defined cutoff values for our population. Previous authors have acknowledged that commonly reported cutoff points for NLR and PLR must be interpreted cautiously in non-white populations since these values can be much worse prognostic indicators on different ethnicities [8]. Due to these racial disparities, we considered that our findings establish a valid cutoff point to discriminate breast cancer patients according to their risk of poor outcomes.

The statistical method used to discretize the NLR and PLR was different from the employed by other authors (i.e. quartiles, ROC curve analysis). Since NLR and PLR were considered continuous prognostic variables for time-to-event outcomes (DFS and OS), a different statistical method was applied to adequately dichotomize these covariates. The sequential approach suggested by Contal and O’Quigley was used in this study due to the good performance of this test to correctly identify the best cutoff for continuous variables [14]. With this approach, the described cut points for NLR (3) and PLR (250) are different to the reported by other authors. Although racial differences and statistical methods can explain this variability, a large amount of diseases and conditions can alter the blood cell counts in each report. We tried to limit this bias through exclusion of patients with acute and chronic inflammatory disorders or previous use of steroids. Despite these criteria we cannot exclude the presence of other confounding variables due to the retrospective design of this study.

Patients with an increased NLR and PLR were more likely to have triple negative tumors and less expression of estrogen or progesterone receptors. However, after adjustment for these possible confounders the multivariate analysis showed that the PLR but not the NLR was independently associated with worse DFS and OS. Although the sample size was relatively low and a type II error could explain the lack of independently association between NLR and the aforementioned outcomes, it must be highlighted that several studies have found that PLR, not NLR, was a predictor of mortality in patients with pancreatic, ovarian, colorectal, and breast cancer patients [4].

In conclusion, our findings contribute to establish a novel an inexpensive biomarker in breast cancer patients from Hispanic background. Further studies must validate this association in our region. The addition of the NLR and PLR to prognostic variables in non-metastatic breast cancer should be warranted in order to obtain a more precise risk profile for poor outcomes among these patients.

Conflicts of interest

None.