The current prediction of prognosis in OSCC is based primarily on the TNM staging system and does not consider the contribution of molecular biomarkers or underlying tumour biology when assigning stage or predicting outcomes. This may be attributed to the lack of robust and reliable techniques for measuring biomarker expression that will help validate their use in the clinic. Several studies have investigated whether changes in the expression levels of apoptosis-associated proteins are associated with survival outcomes in OSCC [14
]. However, a lack of consensus in these studies has precluded the use of apoptotic protein expression as a biomarker in the clinic. Automated quantitative IHC-based techniques, such as AQUA, can be used to eliminate observer bias in the measurement of protein expression. Furthermore, AQUA can measure protein expression within distinct sub-cellular and tissue compartments (e.g. defining expression in the nucleus/cytoplasm and tumour/stroma), possibly improving the clinical applicability of this technique. In the present study, we measured the expression of three apoptosis-regulating proteins using AQUA and found that Bax expression was an independent prognostic marker in OSCC.
Several reports suggest that Bax plays a tumour suppressor role in human malignancies and high Bax expression is associated with favourable prognosis in several cancer types [31
]. Our univariate analysis showed that high Bax expression is associated with improved prognosis. Furthermore, multivariate analysis using Cox proportional hazards regression to adjust for important clinical covariates, confirmed that Bax expression was independently associated with DSS in OSCC. Tumours with high Bax expression are likely to be more susceptible to apoptosis than their low Bax-expressing counterparts. Previous studies have reported that Bax expression is selectively induced in apoptosis-proficient cells [36
]; therefore, OSCC cells with elevated Bax expression could be those that are primed for apoptosis due to the inherent genetic instability associated with malignant transformation.
High Bax expression was associated with increased Ki67 expression, a marker of proliferation. This is consistent with previously reported associations between Bax expression and proliferation in several cancer sites [37
]. Bax over-expression increases cellular proliferation and accelerates G1 to S-phase transition [39
]; this increased cellular proliferation in genetically unstable cancer cells can lead to mitotic catastrophe, thereby augmenting cell death induced by apoptosis in response to radiotherapy [41
]. The increased proliferation associated with elevated Bax expression could confer an improved response to radiotherapy in patients with high Bax. Indeed, stratification by radiation treatment demonstrated that the association between Bax expression and DSS was most pronounced in patients that received adjuvant radiotherapy ( Additional file 3
: Figure S3); however, we were unable to appropriately assess this effect in our cohort. Although not statistically significant, a weak association was observed between high Ki67 expression and improved 5-year DSS ( Additional file 1
: Figure S1).
Since the increased expression of Bcl-2 and Bcl-XL proteins are integral to the survival of malignant cells, it was not surprising that we did not observe any association between the expression of these proteins and 5-year DSS. In effect most tumours rely on the increased expression of antiapoptotic proteins and therefore these proteins will not likely identify unique patient subsets with regard to outcome. We also investigated if the balance between pro-apoptotic and anti-apoptotic proteins was associated with DSS by testing the impact of Bcl-2/Bax and Bcl-XL/Bax ratios on survival outcome. However, we did not find any significant association between these ratios and 5-year DSS. A possible explanation might be the altered expression of additional apoptosis-regulating factors that determine cell fate [30
We evaluated the cellular localization of the three apoptosis-associated proteins in our study and found that Bax expression was predominantly nuclear in normal OCSE but primarily cytoplasmic in OSCC. Previous studies have also reported the presence of Bax and Bcl-2 in the nucleus [42
], however the function of nuclear Bax in our study remains unclear. The sub-cellular localization of Bcl-2 and Bcl-XL was similar in normal and tumour tissue. Bax was the only biomarker that exhibited a difference in sub-cellular localization between normal OCSE and OSCC and was also the only apoptotic protein significantly associated with prognosis. The translocation of Bax from the nucleus to the cytoplasm in OSCC is consistent with increased Bax function at the mitochondria, leading to improved sensitivity to radiotherapy-induced apoptosis in tumours with elevated Bax expression.
Bax and Bcl-XL were over-expressed in most tumour samples, while a majority of the tumour samples expressed Bcl-2 at levels equivalent to normal OCSE. These results suggest that Bcl-XL might be the major anti-apoptotic protein in oral carcinogenesis and the increased Bax expression may be a compensatory response to elevated Bcl-XL levels. Interestingly, we observed that patients treated with surgery alone were over-represented in the low Bcl-2 group, whereas patients additionally treated with post-surgery RT were more common in the high Bcl-2 group (Table ). However, since Bcl-2 did not correlate with treatment outcome, this observation may be attributed to chance rather than a true biologic phenomenon.
A major strength of our study is the use of automated quantitative immunohistochemistry to quantify biomarker expression. The high-throughput technique eliminates observer bias and provides reliable and reproducible estimates for biomarker expression. We used normal tissue samples to obtain physiologically relevant estimates of biomarker expression and define over-expression or down-regulation of apoptotic biomarker in our OSCC cohort. Our study included samples from a single head and neck sub-site (oral cavity) and a uniformly treated cohort from a single cancer center, thus minimizing treatment heterogeneity and increasing the reliability of our results.
Our study is limited by its relatively small sample size and retrospective nature. Despite the advantages of the AQUA technique for discovery-phase studies, it is not widely available in routine laboratory diagnostics at present, limiting its current clinical utility. However, the robustness and reliability of AQUA makes it amenable for adoption as a routine tool for prognostication in the imminent future.