Matrix metalloproteases and TIMPs as prognostic biomarkers in breast cancer patients treated with radiotherapy: A pilot study

Abstract Breast cancer (BC) is the most common tumour in women and one of the most important causes of cancer death worldwide. Radiation therapy (RT) is widely used for BC treatment. Some proteins have been identified as prognostic factors for BC (Ki67, p53, E‐cadherin, HER2). In the last years, it has been shown that variations in the expression of MMPs and TIMPs may contribute to the development of BC. The aim of this pilot work was to study the effects of RT on different MMPs (‐1, ‐2, ‐3, ‐7, ‐8, ‐9, ‐10, ‐12 and ‐13) and TIMPs (‐1 to ‐4), as well as their relationship with other variables related to patient characteristics and tumour biology. A group of 20 BC patients treated with RT were recruited. MMP and TIMP serum levels were analysed by immunoassay before, during and after RT. Our pilot study showed a slight increase in the levels of most MMP and TIMP with RT. However, RT produced a significantly decrease in TIMP‐1 and TIMP‐3 levels. Significant correlations were found between MMP‐3 and TIMP‐4 levels, and some of the variables studied related to patient characteristics and tumour biology. Moreover, MMP‐9 and TIMP‐3 levels could be predictive of RT toxicity. For this reason, MMP‐3, MMP‐9, TIMP‐3 and TIMP‐4 could be used as potential prognostic and predictive biomarkers for BC patients treated with RT.


| INTRODUC TI ON
Breast cancer (BC) is the most common tumour in women and the fifth cause of cancer death worldwide. 1 BC is a heterogeneous disease at the inter-and intra-tumour level, which is relevant to the prognosis and therapy of the disease. 2 The type of BC, its location and other factors (differentiation grade, size, presence of different proteins-Ki67, p53, E-cadherin-sentinel lymph node, patient age and response to treatment) are also relevant to the prognosis. 3 Breast cancer can be classified according to location (in situ and invasive or infiltrating), 3 histology (ductal, lobular, nipple and not otherwise specified), 4 and presence or absence of oestrogen receptors (ER), progesterone receptors (PR) and human epidermal growth factor receptor-2 (HER2). According to these receptors, BC is classified as luminal A or B (ER and PR positive), basal or triple negative (all negative) and HER2-enriched (only HER2 positive) 5 ; being the triple negative, the subtype associated with worse outcome. 6 Differentiation grade indicates the rate of tumour growth and dissemination, and it is determined based on how similar tumour cells compared with normal cells in breast tissue. Tumours can be differentiated into three grades: grade I or well differentiated; grade II or moderately differentiated; and grade III or poorly differentiated. 3 Different proteins are considered markers of prognosis. There is evidence that Ki67 is involved in cell division, 7 and its immunohistochemical (IHC) detection is used to evaluate tumour proliferation. 8 The p53 protein is involved in cancer development. 9 Several studies in mice have shown that p53 mutations can result in a more aggressive tumour behaviour and metastasis. 10,11 E-cadherin mediates cell-cell adhesion and is expressed by epithelial cells. 12 Damage in E-cadherin structure or alteration in its expression are related to tumour progression and metastasis. 13 Involvement of the sentinel lymph node is not only important for disease prognosis but also for tumour invasion and metastasis into the lymph nodes. 14 Sentinel lymph node biopsy is now essential to evaluate the local-regional extension of the disease. Different studies have shown that patients with a negative sentinel lymph node biopsy do not require axillary lymph node dissection. 15 Age and menopausal status are another factor to consider in BC development. Breast lobules undergo age-related lobular involution (ARLI) which begins around the age of 40 and accelerates after menopause.
ARLI is associated with a lower risk of BC. 16,17 Radiation therapy (RT) is used in the treatment of most tumours.
RT regimen is different depending on the patient characteristics, threshold dose and the tissues and organs where tumour is located.
In addition, RT can be combined with surgery and/or systemic therapy. 18 In BC, the most commonly used RT regimens are conventional RT (45)(46)(47)(48)(49)(50) Gy in fractions of 2 Gy) and hypofractionated RT (normally 42.5 Gy in fractions of 2.66 Gy). 19 The efficacy obtained with RT can vary according to the chosen RT regimen and to increased radioresistance in patients. RT toxicity may also appear and depends on the chosen RT regimen, total dose received, volume of irradiated breast and patient age. Skin toxicity is the most common adverse effect in BC patients and they can be acute (erythema, desquamation, ulceration and haemorrhage) or late/chronic (hyperpigmentation and telangiectasia). De Felice et al found that higher irradiated breast volume and conventional RT have a negative effect on acute skin toxicity. 18 Matrix metalloproteases (MMPs) are a family of enzymes that differ in their structure, substrate specificity, sequence homology, cellular localization and secretion. For this reason, they are divided into different subfamilies: collagenases, gelatinases, matrilysins, stromelysins, membrane type MMPs and others MMPs. 20,21 MMPs are mainly involved in extracellular matrix (ECM) remodelling, 20 but they are also capable of processing proteins unrelated to the ECM and activating other MMPs and proteases. 22 MMPs are key regulators of cell-cell interactions and perform different functions in a variety of normal biological 23 and carcinogenic processes (tumour growth, angiogenesis, degradation of collagen in basal membrane, changes in the epithelial-mesenchymal transition [EMT], invasion and metastasis). 24 These processes may be favored by the increase in MMP activity after RT. The balance between MMPs and theirs tissue inhibitors (TIMPs) playing a crucial role in cancer progression and metastasis. 25 Endogenous TIMPs are not only endogenous inhibitors of MMPs, but they also have biological activities that are independent of MMPs including cell growth and differentiation, angiogenesis, apoptosis and synaptic plasticity. 26 The four TIMPs described in humans (TIMP-1 to -4) have different inhibition spectrum and affinity for human MMPs. 27 Changes in MMP and TIMP expression may contribute to the development of BC, and these genes have been examined as potential prognostic serum biomarkers in BC. 24 Different studies have linked high serum levels of MMPs and TIMPs with a poor prognosis 26,28 ; specifically, they have been identified as predictors of adverse outcome and with poor survival. 29,30 Prognostic implications are related to cell type that expresses MMPs (stromal versus tumour cells). 24 Several studies showed that MMP-1, -7, -9, -11, -13 and -14 immunostaining of tumour cells, stromal fibroblasts and mononuclear inflammatory cells were associated with shorter relapse-free survival. 31,32 Considering that RT could influence MMP and TIMP gene expression levels, this therapy could be used to interfere with the different steps of the metastatic cascade. Metastasis is the main cause of death in patients with cancer, and it has been estimated that approximately 90% of BC deaths arise from the metastatic spread of primary tumours. 33 For this reason, the main objective of this work was to study the effects of RT on MMP and TIMP expression, as well K E Y W O R D S breast cancer patients, matrix metalloproteases, prognostic and predictive biomarkers, radiotherapy, tissue inhibitors as their relationship with other variables related to patient and tumour characteristics, and examine their role as prognostic and predictive factors in BC in relation to their key role in tumour invasion and metastasis.  Table 1 describes in more detail the variables studied, the population (n) and the percentage compared with total population. The 6-month BC recurrence is also shown.

| GEO database
Studies with a similar aim were searched in different databases (cBi-oPortal, GDC Data Portal, GEO DataSets). We have not found any study that correlated the serum levels of MMPs and TIMPs with RT.
Only one study that correlates MMPs and TIMPs with RT was found on the GEO database (GSE101920). 34 This study shows the gene expression profile in BC biopsies taken both prior to RT, and after RT and radical mastectomy. Due to the limited number of matched samples (n = 5) obtained, the study was continued by focusing on the analysis of pre-RT biopsies only.

| Statistical analysis
Our results were expressed as median ± standard deviation (SD).  These results demonstrate a slight increase in the levels of most MMP and TIMP with RT. However, only the levels of TIMP-1 and TIMP-3, which decreased with RT, were statistically significant. Regardless of treatment, it is worth noting that MMP-2, -9, TIMP-1 and -2 serum levels were much higher than in the rest of the proteins analysed. In general terms, MMPs and TIMPs showed a tendency to increase, with the exception of TIMP-1 and TIMP-3 where the opposite was observed.

| Time course of MMP and TIMP serum levels
Due to the small sample size of our study, we searched different databases to identify works that support our pilot results. No studies showed similar results regarding the correlation between MMP and TIMP serum levels and RT. Only the study from the GEO database (GSE101920) was similar to our own. However, in that study, only 5 patients had the pre-and post-RT biopsy samples matched. Despite this limitation, pairs of samples were analysed but no significant results were obtained ( Figure S1).

| MMP and TIMP serum level correlation
The correlation between serum levels of MMPs and TIMPs before, during and after RT was obtained using the Spearman correlation coefficient, ρ (rho). According to the rho values, when the association is positive, 0 < ρ < 1, the expression level of the two genes compared shows a similar trend, either an increase or decrease in their expression level. A negative association, -1 < ρ < 0, means that the expression levels of the two genes compared are opposite. Finally, there is no linear correlation between the genes studied when ρ = 0.

| MMP and TIMP serum levels by variables
The correlation between serum levels of all MMPs and TIMPs detected and the different variables was investigated, including patient-dependent variables (age and menopausal status), tumour biology-dependent variables (classification based on hormones, differentiation grade, positive or negative E-cadherin and p53, Ki67 percentage and sentinel lymph node involvement) and RT-related variables (RT regimen, lymph node RT and presence of radiotoxicity). Due to elevated number of graphs obtained after analysing all MMPs and TIMPs with these variables, only the statistically significant correlations are shown and discussed. The data corresponding to the protein levels according to all the variables can be seen in Tables S1-S11. Figure 3 shows the statistically significant correlations between serum levels of MMPs and TIMPs before, during and after RT and the patient-dependent variables ( Figure 3A) and tumour biology-dependent variables ( Figure 3B and 3). The results showed that only MMP-3 and TIMP-4 levels were statistically significant for some variables. Figure 3A shows the correlation between MMP-3 levels and menopausal status of the patient. In general, this protein levels were higher in post-menopausal patients, but it is worth noting that TA B L E 1 Description of the variables studied related to patient characteristics, tumour biology and RT (n = 20) for each group of patients (pre-menopausal and post-menopausal), the levels are higher at a different time of the treatment. Figure 3B shows the correlation between MMP-3 levels and tumour classification, differentiation grade and E-cadherin presence. MMP-3 levels increased in patients with hormone-positive tumours compared with those with hormone-negative tumours, having an increase in both groups throughout the treatment with RT. Considering tumour differentiation degree, MMP-3 levels were higher after RT in grade I and III tumours. However, in grade II tumours, protein levels were higher during treatment. Respect to E-cadherin, patients with E-cadherin positive tumours also showed higher MMP-3 levels than those with negative E-cadherin, having a progressive increase in the E-cadherin positive group throughout RT. Figure 3C shows the correlation be-  Regarding the RT-related variables, Figure 4 shows that only MMP-9, TIMP-1 and -3 levels were found to be statistically significant for some of the variables. Figure 4A and 4 show a statistically significant correlation between MMP-9 and TIMP-3 levels with the type of radiation toxicity. MMP-9 levels were much higher in patients with erythema, showing this group a slight increase throughout the treatment. Moreover, MMP-9 levels in patients with radiodermitis were much higher during RT than before or after RT. However, TIMP-3 levels were very similar in both groups of toxicity but levels decreased throughout treatment. Figure 4C shows a statistically significant correlation between TIMP-1 levels and lymph node RT.
TIMP-1 levels were very similar in patients who have received lymph node RT to those who have not received it, with decreased levels in both groups at the end of RT.

| Comparison of MMP and TIMP serum levels, according to tumour recurrence
Breast cancer recurrence was determined six months after the termination of RT. Overall survival was 100%, and disease-free survival was 85% (recurrence in 3 patients) ( Table 1). The data corresponding to the protein levels according to the tumour recurrence can be seen in Table S12.
The serum levels of all the MMPs and TIMPs detected were compared before and after RT, this time taking into account the recurrence variable ( Figure 5). Patients were grouped into healthy (no recurrent BC) and sick (recurrent BC). Statistically significant values were found only for TIMP-1 and -3 levels. TIMP-1 levels ( Figure 5F) were statistically significant when comparing the levels before and after RT in healthy patients. However, TIMP-3 levels ( Figure 5H) were statistically significant not only when comparing before and after RT in healthy patients, but also when comparing the levels between healthy and sick patients before RT. RT regimen. This work has evaluated the imbalance between MMP and TIMP levels induced by RT. Although our study is limited by the small sample size, these preliminary evidences aim to do additional studies for further confirmation of our results.

CO N FLI C T O F I NTE R E S T
The authors confirm that there are no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.