Retrospective assessment of cyclin‐dependent kinase 5 mRNA and protein expression and its association with patient survival in breast cancer

Abstract Cyclin‐dependent kinase 5 (Cdk5) is an atypical member of the cyclin‐dependent kinase family and functions as a serine/threonine kinase that can be activated by non‐cyclin binding activators p35 or p39. Cdk5 expression and activity has been linked with the development and progression of cancer; however, its expression in breast cancer has not been fully described. Protein expression of Cdk5 was determined in a large cohort of early‐stage invasive breast cancer tumours (n = 1110) with long‐term follow‐up data using immunohistochemistry. Expression of CDK5 mRNA was assessed in the METABRIC cohort (n = 1980). Low nuclear and cytoplasmic expression of Cdk5 expression was significantly associated with shorter breast cancer‐specific survival (P = .004 and P = .001, respectively). Importantly, low nuclear and cytoplasmic expression of Cdk5 remained associated with survival in multivariate analysis, including potentially confounding factors (hazard ratio (HR) = 0.612, 95% confidence interval (CI) = 0.418‐0.896, P = .011 and HR = 0.507, 95% CI = 0.318‐0.809, P = .004, respectively). In addition, low nuclear and cytoplasmic expression of Cdk5 was significantly associated with clinicopathological criteria associated with adverse patient prognosis. Low CDK5 mRNA expression was associated with shorter patient survival (P = .005) in the METABRIC cohort; no associations between copy gain or loss and survival were observed. These data suggest that low Cdk5 expression is associated with poor clinical outcome of breast cancer patients and may be of clinical relevance.

however, it is clear that Cdk5 can play a role in cell types other than neurons and in the pathogenesis of certain diseases, such as cancer and Alzheimer's disease. Cdk5 functions in the development and progression of cancers through modulation of key pathways such as DNA repair and cellular migration.
Cdk5 was first described in 1992, 1 and its function and numerous substrates have subsequently been described (reviewed in Ref. [2]). Monomeric Cdk5 is inactive and requires binding of p35 or p39 to become activated, and it also functions in a negative feedback pathway through phosphorylation of p35 leading to its ubiquitination and subsequent degradation. 3 In neurons, Cdk5 functions in numerous important cellular pathways; for example, phosphorylation of apurinic/apyrimidinic endonuclease 1 (Ape1) by Cdk5/p35 complexes reduces the AP endonuclease activity of Ape1, an enzyme critical in the base excision repair pathway, to allow DNA damage to accumulate. 4 Cdk5 also phosphorylates focal adhesion kinase (FAK) allowing regulation of a microtubule fork that results in neuronal migration. 5 In cancer, Cdk5 participates in numerous pathways associated with tumour progression. Cdk5 can influence cellular migration in cancer, such as phosphorylation of FAK, promoting the formation of F-actin bundles to facilitate epithelial to mesenchymal transition and motility. 6 In addition, phosphorylation of GIV by Cdk5 is to play a role in the migration-proliferation dichotomy, 7 and in breast cancer, Cdk5 phosphorylation of adducin-1 is important for epidermal growth factor-induced cell migration and invasion. 8 In cancer, high Cdk5 protein and mRNA expression is associated with clinicopathological features associated with a poor prognosis or adverse survival in a number of tumour types. In lung cancer, high CDK5 mRNA expression has been associated with survival, and protein expression is associated with numerous clinicopathological criteria associated with a poor prognosis. 9,10 Studies have assessed the association between Cdk5 and survival; however, this has been in tumour types other than breast.
In colorectal cancer, high Cdk5 expression is associated with poor prognosis, 11 and in gastric cancer, low Cdk5 expression has been associated with shorter patient survival. 12 In breast cancer, Cdk5 expression has been associated with prognostic factors associated with worse patient prognosis; however, no association with survival was reported. 6 We sought to determine the frequency of Cdk5 overexpression in a large cohort of early-stage invasive breast cancers to determine whether Cdk5 expression was associated with poor patient survival and test associations with clinicopathological criteria.

| Patient cohorts
A total of 1110 early-stage invasive, primary breast cancer patients were included in this study and were treated between 1998 and 2006 at Nottingham University Hospitals. All patients underwent a wide local excision or mastectomy, which was determined by disease characteristics or patient choice; this was then followed by radiotherapy, if indicated. Oestrogen receptor (ER) positivity/negativity, Nottingham Prognostic Index (NPI) value and menopausal status were used to determine whether a patient received systemic adjuvant treatment.
Patients with an NPI index 3.4 or greater were candidates for CMF combination chemotherapy (cyclophosphamide, methotrexate and 5-fluorouracil) if they were ER negative or premenopausal; and hormone therapy if they were ER positive. Breast cancer-specific survival was calculated as the time between death resultant from breast cancer and primary surgery, and the reverse Kaplan-Meier method was used to calculate the median patient follow-up as 148 months.
REMARK criteria were used to report this study. 13 The molecular taxonomy of breast cancer international consortium (METABRIC) data series was also used (n = 1980) with information about the data set published elsewhere. 14  platforms for genomic and transcriptional profiling.

| Immunohistochemistry
Tissue microarrays were used for immunohistochemistry and were composed of 0.6 mm tissue cores taken from representative tumour areas which were determined using Haematoxylin and eosin-stained tissue. Leica Novolink Polymer Detection kit was used according to the manufacturers' instructions and has been described previously 15 ; briefly, the slides were heated on a 60°C hotplate for 10 minutes prior to deparaffinization and rehydration in xylene, ethanol and then water. Tissue was microwaved in 0.01 mol L −1 sodium citrate buffer for 10 minutes at 750 W followed by 10 minutes at 450 W. Novolink Peroxidase Block was incubated on the tissue for five minutes prior to washing with Tris-buffered saline (TBS) and followed by incubation

| Statistical analyses
IBM SPSS (version 24) was used to conduct statistical analysis. The final H-score and CDK5 mRNA expression was stratified into low and high expression using X-tile software based on breast cancer-specific survival. 16 Double assessment was performed in over 30% of cases, with, with both assessors blinded to both each other's H-score and the clinical outcome of the patients. Intraclass correlation coefficients (single measure) were 0.958 and 0.912 for Cdk5 cytoplasmic and nuclear scores, respectively, both indicating good concordance between scorers. The Pearson chi-squared test of association enabled assessment of the connection between clinicopathological variables and categorized protein expression. The Kaplan-Meier method was used to plot survival curves, with significance determined using the log-rank test.
All differences were considered statistically significant at the level of P ≤ .05. DARPP-32 staining and scoring has been reported previously. 17

| Cdk5 expression in invasive breast cancer
Cytoplasmic and nuclear Cdk5 staining varied from strong to weak, and representative photomicrographs are shown in Figure 1A-C. Nuclear Cdk5 expression was significantly

| Cdk5 expression and breast cancerspecific survival
Both low nuclear and cytoplasmic expressions of Cdk5 were associated with adverse breast cancer-specific survival (P = .004 and P = .001, respectively; Figure 1D,E). Low nuclear Cdk5 ex-

| D ISCUSS I ON
In the current study, we demonstrate that, in a large cohort of breast cancer patients, low nuclear and cytoplasmic expression of Cdk5 is significantly associated with adverse disease-specific survival (P = .004 and P = .001, respectively). Importantly, low nuclear and cytoplasmic Cdk5 expression was significantly associated with survival in multivariate assessment, when potentially confounding factors were included (P = .011 and P = .004, respectively). In addition to protein expression, low CDK5 mRNA expression was associated with shorter patient survival in the METABRIC cohort (P = .005).
Studies have assessed the expression of Cdk5 and its association with survival in cancer; however, this has been in tumour types other than breast. In gastric cancer, low Cdk5 expression has been associated with adverse survival in 240 patient specimens. 12

ACK N OWLED G EM ENTS
Tissue samples were kindly provided through the Nottingham Health Science Biobank and Breast Cancer Now.

CO N FLI C T O F I NTE R E S T
There is no conflict of interest to declare.

AUTH O R CO NTR I B UTI O N S
BS and SS conducted the studies and collected the data; ER, IE and AG provided tissue and clinical information; SS conducted statistical analysis; SGM and SS conceived the study; SS wrote the manuscript; and all authors approved the manuscript for submission.

E TH I C A L A PPROVA L
This study was approved by the Nottingham Research Ethics