The results from this validation study provide further evidence that overexpression of PODXL protein is a predictor of poor prognosis in CRC, independent of tumour stage or other clinicopathological characteristics. Together with the results from our previous study
], this correlation has now been confirmed in three independent cohorts representing more than 1000 patients in total. Moreover, while the results from this study further confirm the association of high PODXL expression with a reduced overall survival, its impact as a biomarker of reduced time to recurrence and disease free survival in curatively treated patients is also demonstrated.
While PODXL is expressed in the cytoplasm in a considerable proportion of CRC tumours, it is mainly the presence of distinct membranous staining, here denoted as high expression, that seems to confer a poor prognosis
]. The most aggressive tumours show a strong, membranous staining in a subset of tumour cells at the invasive tumour front, corresponding well to the morphological term “tumour budding”, which has been demonstrated to be of prognostic importance in CRC
], and biologically closely related to EMT
]. These observations are also well in line with the study by Cipollone et al, where PODXL expression on the cell surface but not in the cytoplasm was significantly associated with a shorter disease-free survival in patients with high grade serous ovarian carcinoma
]. In that study, it was also demonstrated that forced expression of PODXL in serous ovarian carcinoma-derived OVCAR-3 cells resulted in localization of PODXL to the cell surface, decreased cell adhesion to mesothelial monolayers and diminished levels of β1-integrin, leading the authors to conclude that PODXL may facilitate transperitoneal metastasis of high grade serous carcinoma
]. In light of the significant association of high PODXL expression and increased T-stage, in particular stage T4 tumours, observed in our previous study
], and in Cohort II in this study, it would be of interest to perform further studies to investigate whether PODXL may have a role in the initiation of serosal invasion also in CRC.
Since membranous staining of PODXL in most cases is seen in only a fraction of tumour cells, these tumours do not necessarily have the highest level of protein in total and overexpression of protein will not be reflected in high mRNA levels. Hence, evaluation of PODXL expression should be based on a qualitative rather than quantitative assessment and IHC has several advantages over other types of assays in terms of clinical applicability as it is a comparatively simple, fast and inexpensive method.
Considering the prognostic importance of protein localization, the lack of a significant correlation between PODXL mRNA and protein expression levels, clinicopathological characteristics and survival is not surprising. Moreover, being a CD34-related protein, PODXL is expressed on vascular surfaces
], and thus, present in various amounts in the tumour-associated stroma.
It should also be pointed out that previous studies attempting to determine a direct correlation between mRNA levels and protein expression in tumours have shown divergent results, and analyses indicate that protein concentrations correlate with the corresponding mRNA levels by only 20-40 %
]. In some cases, such as HER2/neu, expression levels show highly significant correlation
], but in other studies regarding molecular markers in adenocarcinoma of the lung
] and in prostate cancer
], a relationship between mRNA and protein was not observed. There are several possible explanations for these discrepancies, e.g. post-translational modifications, variations in protein half-life and actual biological differences between mRNA and protein abundance. In addition, there are potential experimental errors to be considered, including anatomical origin and quality of tissue. A more comprehensive analysis of the correlation between PODXL mRNA and protein expression might be provided by performing microdissection of strongly staining areas from frozen tumour sections. In the clinical setting, however, it is evident that immunohistochemical assessment of PODXL is the method of choice, whereby recognition of its location on the cellular surface should be quite straightforward.
Some limitations related to the TMA-technique must be considered, not least its ability to accurately reflect the expression of heterogenously expressed markers. One way to compensate for this is to, whenever possible, ensure that tumour cores are sampled from different tumour areas, i.e. the invasive front and centre, respectively. While this had been done for the majority of the here analyzed tumours, it should be pointed out that tumour areas denoted as having distinct membranous PODXL expression could not only be found at the invasive front, but also in scattered areas within the tumour. However, we have previously compared results from paired TMA-cores with full-face sections with excellent concordance
]. Moreover, assessment of full-face sections from prospectively collected clinical samples have reveled a similar proportion of CRC cases with high PODXL expression as reported here and in our previous study (unpublished observations).
In both cohorts in this study, the proportion of tumours denoted as having high PODXL expression was lower compared to our previous study, where 13.4% of the tumours displayed high PODXL expression
]. Given the observed association between PODXL expression and a more advanced disease stage, this can be explained by the fact that the two cohorts examined here had a lower percentage of patients with stage IV disease (9.6 vs 7.9%) compared to the population-based prospective cohort used in the previous study (18.3%).
Furthermore, in patients with stage III disease in cohort 2, a trend towards the previously demonstrated benefit from adjuvant chemotherapy could be observed for patients with high tumour-specific PODXL expression, who had a similar DFS and OS as patients with PODXL-low tumours
]. Although this did not reach statistical significance, most likely due to the smaller subgroup available for analysis, these findings further indicate that patients with high PODXL-expressing tumours might benefit from adjuvant treatment. As adjuvant chemotherapy is given to the majority of patients with stage III disease according to current treatment protocols, assessment of PODXL expression might be particularly relevant in order to identify high-risk patients with stage II disease. Similar to our previous study
] the number of patients with stage II disease in this study who received adjuvant treatment were too few for a meaningful statistical analysis. Hence, these associations should be confirmed in larger retrospective studies or within randomized treatment trials.