Aurora Kinase A Is an Independent Predictor of Invasive Recurrence in Breast Ductal Carcinoma in situ

Introduction: Aurora Kinase A (AURKA/STK15) has a role in centrosome duplication and is a regulator of mitotic cell proliferation. It is over-expressed in breast cancer and other cancers, however; its role in ductal carcinoma in situ (DCIS) remains to be defined. This study aims to characterize AURKA protein expression in DCIS and evaluate its prognostic significance. Methods: AURKA was assessed immunohistochemically in a large well-characterized cohort of DCIS (n = 776 pure DCIS and 239 DCIS associated with invasive breast cancer [DCIS-mixed]) with long-term follow-up data (median = 105 months) and basic molecular characterization. Results: High AURKA expression was observed in 15% of DCIS cases and was associated with features of aggressiveness including larger tumour size, high nuclear grade, hormone receptor negativity, HER2 positivity, and high Ki67 proliferation index. AURKA expression was higher in DCIS associated with invasive breast cancer than in pure DCIS (p < 0.0001). In the DCIS-mixed cohort, the invasive component showed higher AURKA expression than the DCIS component (p < 0.0001). Outcome analysis revealed that AURKA was a predictor of invasive recurrence (p = 0.002). Conclusion: High AURKA expression is associated with poor prognosis in DCIS and might be a potential marker to predict DCIS progression to invasive disease.


Introduction
Although ductal carcinoma in situ (DCIS) is not lifethreatening disease, it is a precursor of invasive breast cancer (IBC) with subsequent risks of distant metastasis and breast cancer-related mortality [1,2]. A plethora of studies have investigated different potential risk factors predicting local recurrence or DCIS progression [3][4][5], however; predicting progression to invasive disease remains a challenge. As a consequence, large proportion of patients with DCIS is treated with extensive surgery including mastectomy or breast conserving surgery (BCS) with often frequent reoperations and/or radiotherapy [6, Pathobiology 2022;89:382-392 DOI: 10.1159/000522244 7] which could be considered as over-treatment of precursor lesions. Therefore, identifying DCIS that might be fatal is of great importance to reduce over-or undertreatment. Clinicopathological characteristics of DCIS such as lesion size, margin status, nuclear grade and presence of necrosis are useful predictors of DCIS behaviour [8,9]. However, these do not discriminate between invasive and in situ recurrence, and the best combination of these variables and the improved performance using molecular markers remain to be defined [5]. Identification of novel markers that play a role in DCIS progression might aid our understanding of the disease biology and risk stratification.
Aurora kinase A (AURKA) belongs to the family of serine-threonine kinases that play an integral role in cell cycle regulation [10] by recruiting the cyclin B1/CDK1 complex and committing cells to mitosis [11]. It has a key role in centrosome duplication and is a critical regulator of mitotic cell proliferation playing essential roles in mitotic entry, centrosome maturation, mitotic spindle assembly, and chromosome segregation processes [12]. The AURKA gene is localized on chromosome segment 20q13, which is amplified in many human cancers [13][14][15]. Ectopic overexpression of the kinase induces chromosomal instability, centrosome anomalies, and tumorigenic transformation of human cells [16,17]. AURKA represents a unique proto-oncogenic mitotic kinase that is involved in the genetic pathways underlying the two most commonly observed phenotypic alterations in human cancer cells: aneuploidy and centrosome aberrations [18]. This property is critically relevant for breast cancer as a disease driven by chromosome copy number alterations [19].
AURKA is expressed at elevated levels in IBC [15,20], colorectal [18], ovary [21], and gastric carcinomas [22] and is associated with poor prognosis [23]. It is one of the proliferation genes included within the gene panel of the Oncotype DX prognostic assay for both DCIS and IBC [24][25][26][27]. Moreover, AURKA is differentially expressed between normal breast tissue and IBC [28]; however, the role of AURKA in DCIS has yet to be established. In this study, we aim to assess the pattern of AURKA protein expression and its prognostic significance in a large wellannotated DCIS series. In addition to further characterize the prognostic significance of AURKA at the transcriptomic level, a large cohort of invasive breast cancer (n = 1,980) with long-term follow-up was used as surrogate for DCIS.

Study Cohort
A well characterized annotated cohort of DCIS including pure DCIS (n = 776) and DCIS mixed with IBC (DCIS-Mixed) (n = 239) diagnosed between 1990 and 2012 at Nottingham City Hospital, Nottingham, United Kingdom was used as previously described [29]. Patients' demographic data, histopathological characteristics, management including postoperative radiotherapy and development of local recurrence were collected. Patients were presented as asymptomatic screen detected (52.2%) or symptomatic palpable lesions (47.8%). Along the study period, patients were managed either by mastectomy (51.9%) or BCS (48.1%) with or without radiotherapy (29.6% and 70.4%, respectively).
Local recurrence-free survival (LRFS) was defined as the time (in months) between 6 months after the first DCIS surgery and occurrence of ipsilateral local recurrence (either as DCIS or IBC). Cases undergoing re-excision within the first 6 months due to close surgical margins or presence of residual disease were not considered as recurrence. Patients who developed contralateral disease following DCIS diagnosis were censored at the time of development of the contralateral cancer. Within a median follow-up period of 105 months (range 6-240), 83 patients (11%) developed local ipsilateral recurrence including invasive (53/83; 64%) or DCIS (30/83; 36%). Eight recurrence events developed after management with BCS followed by adjuvant radiotherapy, while the majority of the recurrences occurred after BCS alone. Patients who developed contralateral disease following DCIS diagnosis were censored at the time of diagnosis of the contralateral cancer.
Additionally, data on ER, PR, HER2, and Ki67 [29] were available and included. Classification of DCIS was done according to the molecular classes defined according to St. Gallen International Expert Consensus [30]. These classes are (i) luminal A (ER and/or PR positive, HER2 negative and Ki67 <14%); (ii) luminal B/HER2− (ER and/or PR positive, HER2 negative and Ki67 ≥14%); (iii) Luminal B/HER2+ (ER and/or PR positive, HER2 positive); (iv) HER2+/ER− (non-luminal) (ER and PR negative and HER2 positive); and (v) triple negative (ER, PR, and HER2 negative). For ER and PR, a 1% cut-off value was used to dichotomise cases into positive and negative [31]. HER2 status was considered negative if the immunohistochemical score was 0 or 1+, equivocal if the score was 2+, and positive if the score was 3+ [32]. For HER2 2+ cases, HER2 gene amplification was detected using chromogenic in situ hybridization and confirmed gene amplification was defined as six or more signals per nucleus or when clusters (clumps of aggregated green signals) were identified in the cell nuclei in more than 50% of tumour cells [33]. Ki67 high proliferation index was considered when more than or equal to 14% positively stained tumour cell nuclei were detected [34].
Tissue Microarrays and Immunohistochemistry Tissue microarrays were prepared from representative DCIS lesions of the pure cases and from DCIS and invasive tumours from the mixed cases as previously described [29]. In addition, a set of whole tissue sections from 10 cases containing DCIS and invasive tumours were assessed to evaluate heterogeneity and the pattern of AURKA expression in malignant breast lesions and adjacent stroma and normal tissue.
Primary antibody specificity for rabbit polyclonal AURKA antibody (Abcam; Cambridge, UK) was validated using Western blotting on whole cell lysates of MCF-7, SKBr3 and MDA-MB-231 human breast cancer cell lines (obtained from the American Type Culture Collection; Rockville, MD, USA). AURKA antibody was used at a dilution of 1:500 which showed a single specific band at the predicted size of 44 kDa.
Expression of AURKA protein was assessed by immunohistochemistry using the Novocastra Novolink polymer detection system (Code: RE7280-K, Leica, Newcastle, UK). Four micrometer tissue microarray and full-face sections were stained with the AURKA antibody (1:200) incubated for 24 h. 3,30-Diaminobenzidine tetrahydrochloride (Novolink DAB substrate buffer) was used as a chromogenic substance. Sections were counterstained with haematoxylin. Positive staining controls (human tonsil) were included while a negative control was achieved by omitting the application of the primary antibody.

Assessment of AURKA Expression
The percentage of nuclear AURKA staining was estimated in pure DCIS and mixed cases. Cores containing <15% of tumour epithelial cells were excluded from assessment. All cases were scored blinded to clinicopathological and outcome data. For the mixed cohort, each component; DCIS and invasive, was scored separately. For dichotomization of protein expression, cut-off points were defined according to the calculated results from X-tile bioinformatics software (Yale University, version 3.6.1) [35,36] with corrected p value and relative risk against LRFS. High AURKA expression was considered when more than 60% of tumour cells showed staining.

Analysis of AURKA mRNA Expression in Invasive Breast Cancer
To confirm the prognostic significance of AURKA in breast cancer and given the deficiency of data on the transcriptomic profiles of DCIS, AURKA normalized mRNA expression was evaluated as a potential prognostic marker using the Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) dataset that comprises 1,980 tumours of invasive breast cancer with comprehensive molecular characterization and long-term follow-up [37].

Statistical Analysis
Statistical analyses were performed using SPSS v26 (Chicago, IL, USA) for Windows. Student's t test and analysis of variance (ANOVA) were used to correlate between AURKA mRNA level as a continuous variable and other clinicopathological parameters in METABRIC data. Association with AURKA mRNA expression and breast cancer specific survival was performed after dichotomization of expression into high and low groups based on the median value.
Association between AURKA expression and clinicopathological parameters in pure DCIS was performed using χ 2 for categorized data, and Mann-Whitney and Kruskal-Wallis tests for continuous variables. Wilcoxon signed-rank test was used to compare the expression of AURKA between DCIS component and the invasive component within the mixed cases. Survival rates were determined using the Kaplan-Meier method and compared by the log-rank test. Multivariate analysis using Cox proportional hazard regression model determined the influence of AURKA expression, when adjusted to other variables, for all local recurrences (either DCIS or invasive breast cancer) and invasive recurrences. All tests were 2-tailed, and a p value of less than 0.05 was considered as statistically significant.

Frequency and Localization Pattern of AURKA
Assessment of ten whole tissue sections revealed nuclear expression of AURKA in a homogenous distribution pattern confirming the validity of using tissue microarrays to assess its expression. Epithelial cells in the adjacent normal breast terminal duct-lobular units showed negative or very weak cytoplasmic staining of AURKA "shown in Figure 1." After exclusion of uninformative cases (i.e., lost cores, folded tissue during processing and cores containing scanty tumour cells), a total of 632 pure DCIS and 217 cases associated with mixed DCIS and invasive components were assessed. AURKA expression showed a unimodal distribution with a median percentage of 30% positive nuclei in pure DCIS, 45% in the DCIS component of mixed cases, and 65% in invasive component (range 0-100%). Within the pure DCIS cohort, high AURKA expression was observed in 15% of cases which was significantly lower than that in the DCIS component of the mixed cases (29% with high expression, p = 0.004) and the invasive component (55% with high expression, p < 0.0001). A statistically significant difference in AURKA expression was also observed between tumour cells of DCIS and corresponding invasive components of the mixed cases (p < 0.0001) "shown in Figure 2." Association of AURKA with Clinicopathological Parameters in Pure DCIS In BCS treated patients, high AURKA expression was significantly associated with poor prognostic factors including large tumour size, high nuclear grade, negative hormone receptor expression, HER2 positivity, and high Ki67 proliferation index (Table 1). On the other hand, low AURKA expression was significantly associated with factors of good prognosis with highly significant association with luminal A subtype according to the molecular classification of breast cancer. Analysis of continuous data of AURKA expression showed similar results (online suppl. Table 1; for all online suppl. material, see www. karger.com/doi/10.1159/000522244).

Association between AURKA Expression and Patient Outcome
In univariate analysis, with over 20-year follow-up "shown in Figure 3," high AURKA expression within DCIS treated with BCS only without adjuvant radiotherapy showed a significantly shorter ipsilateral LRFS for all recurrences including in situ and invasive disease (hazard ratio [HR] = 2.7, 95% confidence interval [CI] = 1.6-4.6  and p < 0.0001). However, this association was not maintained in patients who received adjuvant radiotherapy (p > 0.05). When the subtype of recurrence was considered, high AURKA expression was associated with recurrence as invasive disease (HR = 2.9, 95% CI = 1.5-5.8 and p = 0.002). It is also shown that the probability of recurrence in cases with high AURKA expression was significantly lower compared to cases with lower expression over 20and 10-years' time. However, this was not statistically significant over 3-and 5-years' time, "shown in online supplementary Figure 1 and online supplementary Table 2." Multivariate survival analysis showed that AURKA expression is a predictor of local recurrence in patients treated with BCS independent of other well-established prognostic factors including patient age, tumour size, tumour grade, comedo necrosis, surgical margin width, molecular classes, and adjuvant radiotherapy use (HR = 6.9, 95% CI = 2.7-10.0, p < 0.0001). AURKA was also shown to be an independent predictor of invasive recur-rence (HR = 3.9, 95% CI = 1.7-9.1, p = 0.001) (shown in Table 2; Fig. 4).
When AURKA expression was combined with the other determinants of DCIS risk described by the VNPI [38], expression of AURKA was associated with outcome in all VNPI risk groups. Inclusion of AURKA into VNPI affected the HR of prediction of outcome from 2.9 (95% CI: 1.7-4.9) to 3.3 (95% CI: 1.4-6.7).
To validate the prognostic importance of AURKA in IBC, the METABRIC cohort was used to assess the correlation between AURKA mRNA levels and clinicopathological variables and outcome. High AURKA mRNA level was associated with young patient age (p = 0.001), high histological grade (p = 0.034), negative hormone receptor expression and HER2-positive phenotype (p < 0.0001), in addition to worse outcome in terms of shorter breast cancer specific survival (HR = 1.4, 95% CI = 1.1-1.6, p < 0.0001) (shown in online suppl. Table 3; online suppl. Fig. 2).

Discussion
The treatment of DCIS remains a challenge, as the clinicopathological features of the disease do not reliably stratify patients into distinct risk groups to guide treatment decisions [39]. For this reason, some studies have attempted to risk stratify DCIS based on genetic and molecular factors including the Oncotype DX ® DCIS score [26,[40][41][42][43]. Although this score minimizes the proportion of patients undergoing radiotherapy, the assay is relatively expensive and showed some inconsistent results [44,45]. Therefore, there is a pressing need to identify robust and cost-effective biomarker(s) to predict outcome for DCIS patients. AURKA was a candidate to study and immunohistochemical detection of the protein is a robust and relatively inexpensive method that may be able to assist with DCIS prognostication. Indeed, the HR we observed for the combination of VNPI and AURKA for any local recurrence after BCS without postoperative radiotherapy (HR = 3.3) is strikingly similar to that obtained for the Oncotype DX score in a recent analysis (HR = 1.95 and 2.48 in two cohorts) [7].
This study showed that high expression of AURKA was associated with factors of poor prognosis including high nuclear grade, negative hormone receptor expression, HER2 positivity, and high proliferative activity of tumours expressed as high Ki67 LI. This goes in line with other studies [20,46,47]. This is explained by the proliferation driving properties of AURKA that regulates the transition of cells from the G2 to M phase [46].
The current study showed that high AURKA expression is a poor prognostic factor for DCIS which is independent of other clinicopathological factors. These findings were similar for all recurrence events (DCIS or invasive) and also when the analysis was restricted to invasive recurrences only. These findings suggest that AURKA is a promising marker for introduction of a new high-risk DCIS group in addition to identification of patients with low risk for whom radiotherapy could be avoided.
Increased copy number of AURKA is associated with progression from a colonic polyp to invasive malignancy and is one of the most common copy number alterations in cancer. This finding reveals that it has a role not only in tumour migration and invasion but also in tumour development [48,49]. Similarly, although 20q gain is common in DCIS (∼25%) [50], it has also been observed as an IBC-specific event in mixed DCIS, suggesting an association with invasiveness in breast cancer [51]. In contrast to the study of 37 breast cancer patients by Hoque et al. [52], who reported that loss of AURKA expression correlates with the transition from in situ to invasive carcinoma of the breast, AURKA is differentially expressed between normal breast tissue and invasive breast [28,53]. Herein, we show that AURKA expression is most prevalent in invasive tumour cells, in line with [25], followed by a lower level in DCIS coexisting with invasive carcinoma and much lower in pure DCIS. These findings corroborate the role of AURKA in DCIS progression. Moreover, using the METABRIC series, we have shown a significant association between aggressive behaviour of IBC and higher levels of AURKA mRNA. These observations support our hypothesis that AURKA is a promising candidate biomarker that requires further functional studies to decipher its role in DCIS behaviour.
There are a number of plausible biological mechanisms to explain the involvement of AURKA in breast carcinogenesis and cancer progression. AURKA over-expression in fibroblasts and breast epithelial cells results in centrosome amplification and aneuploidy, suggesting it plays an important role in malignant transformation [54]. It is expressed in multiple carcinomas including breast cancer [22,[55][56][57], and this over-expression is associated with centrosome amplification and DNA instability [58]. The non-mitotic function of AURKA is implicated in breast cancer progression and resistance to chemotherapy agents through epithelial to mesenchymal transition and the acquisition of stem cell-like characteristics [59,60]. Breast cancer cells with these stem cell-like properties have consequently been associated with tumour progression and onset of distant metastasis [61,62].
In addition, nuclear AURKA acts as a trans-activating factor to promote the expression of MYC. The activation of MYC by AURKA is dependent on the nuclear localization of AURKA rather than its kinase activity [63]. Nuclear AURKA may also promote the expansion of breast cancer stem cells [64]. There is emerging evidence to suggest that AURKA also promotes cancer development through mechanisms independently of its kinase activity [65].

Conclusion
AURKA might have a potential role in DCIS aggressiveness through its oncogenic activity and its regulatory role in cellular division and proliferation. Additional functional studies are highly recommended to unravel the role of AURKA in regulating DCIS behaviour. AUR-KA might also be a useful prognostic indicator especially as a predictor of invasive recurrence.