This study was conducted to examine the prognostic impact of SATB2 expression in incident CRC cases from a large prospective, population-based cohort study. SATB2 has previously been described as a promising novel diagnostic marker for CRC (Magnusson et al, 2011
) and, in a smaller CRC cohort, loss of SATB2 has been linked to poor prognosis (Wang et al, 2009
). We found that high expression of SATB2 was an independent factor of good prognosis in colon but not rectal cancer. Moreover, in curatively treated patients with stage III–IV disease, SATB2 expression was a predictor of response to adjuvant chemotherapy, irrespective of tumour location, and in patients with rectal cancer, a significant interaction between high SATB2 expression and response to neoadjuvant therapy was observed. The findings of an association between SATB2 and an improved response to chemotherapy and radiation therapy are of potential interest, but should be interpreted with caution, as treatment data was not available for all patients in this cohort, hence only allowing for rather small subgroup analyses. On the other hand, since the MDCS started as early as in the mid 90s, when adjuvant chemotherapy was not yet standard of care in Sweden, the comparatively high proportion of patients with stage III tumours not receiving adjuvant treatment is a relative strength in the use of this cohort for biomarker studies. Nevertheless, the putative treatment predictive role of SATB2 should preferably be validated in tumour specimens from randomised, controlled treatment trials and the molecular basis for how SATB2 might modulate the effects of chemotherapy and radiation also remains to be elucidated. To date, the role of SATB2 in chemotherapy response has only been investigated in one study on head and neck squamous cell carcinoma (HNSCC) cells, where SATB2 was demonstrated to promote chemo- and radiation resistance by modulation of ΔNp63 (Chung et al, 2010
). These findings are in contrast to ours but, notably, in the same study, immunohistochemical detection of SATB2 was reported in more than 50% of human HNSCC tumours, which is not consistent with the antibody-based screening in the Human Protein Atlas, where SATB2 could not be detected in HNSCC using different well-validated antibodies (www.proteinatlas.org
). Moreover, the role of SATB2 in transcriptional regulation and as a driver of epigenetic events may well differ between different cancer forms.
The reduced expression of SATB2 in MSI tumours is consistent with studies on other markers of colorectal lineage, for example, CDX2 and CK20 (Lugli et al, 2008
). Furthermore, in light of the findings from several studies suggesting that 5-FU negatively affects outcome for microsatellite unstable tumours (Barratt et al, 2002
; Ribic et al, 2003
; Kim et al, 2007
), the herein observed association between SATB2 expression and MSI status fits with the improved benefit from adjuvant treatment seen for patients with SATB2 high tumours.
The differential prognostic impact of SATB2 expression in colon and rectal cancer is noteworthy and further underlines the importance of preserving a distinction between the two disease entities, which should also be considered in future validatory studies. In our study, SATB2 expression was found to be higher in rectal cancer compared with colon cancer, although this difference did not reach statistical significance. Additional studies are warranted to clarify whether this finding is coincidental or actually mirrors different tumour biological properties of colon and rectal cancers. As all rectal tumour samples were taken from post-treatment surgical specimens, it could be speculated that SATB2 levels are modified by neoadjuvant RT or chemotherapy. This is however less likely, as SATB2 expression did not differ between treated and untreated tumours.
SATB2 is closely related to SATB1, another member of the SATB family of transcription factors (Kohwi-Shigematsu et al, 1997
; Yasui et al, 2002
; Cai et al, 2003
; Cai et al, 2006
). Although the role of SATB1 has been more extensively explored in the context of cancer, its impact on prognosis seems to be cancer –type-dependent. In breast cancer, a role for SATB1 as being a master switch towards a metastatic phenotype and a marker of poor prognosis has been demonstrated in a study including immunohistochemical analysis of >1000 human breast cancer specimens (Han et al, 2008
). In a recent study on rectal cancer (n
=93), SATB1 expression was found to correlate with a more advanced TNM stage; however, its impact on recurrence or survival was not evaluated (Meng et al, 2011
). Wang et al (2009
) also found an inverse association between SATB1 and SATB2 in CRC cells in vitro
. In lung cancer, a significant loss of SATB1 expression was found in squamous preinvasive lesions and in non-small cell lung cancers compared with matched normal bronchial epithelium, and loss of SATB1 was an independent predictor of poor survival in squamous cell carcinomas (Selinger et al, 2010
The frequency of SATB2-negative tumours reported here is lower than that in the study by Magnusson et al (2011)
, but in that study only ~10% of the tumours had metastatic disease compared with ~17% in this cohort, which might in part explain these differences, although SATB2 expression was not found to be significantly associated with M stage, only T and N stage, in this study. A significant association between low SATB2 expression and metastatic CRC was, however, demonstrated in the study by Wang et al (2009)
, where ~30% of the patients had M1 disease and the frequency of SATB2 low tumours was >50%, but as the proportion of tumours lacking SATB2 expression was not reported, comparisons are difficult to make. Possibly, the lower, although non-significant, frequency of M1 tumours in rectal compared with colon cancers found here could in part explain the observed higher SATB2 expression in rectal cancer. Optimal cutoffs for assessment of the prognostic and treatment predictive value of SATB2 expression will have to be confirmed in future studies. Notably, in this study, although any vs
negative SATB2 expression was also of prognostic value in univariable analysis, only high expression according to the the CRT-derived cutoff at NS>9, corresponding to the highest expression level, was an independent favourable prognostic as well as treatment predictive factor.
As the Malmö Diet and Cancer Study is a population-based cohort study, a potential selection bias compared with the general population must be taken into consideration (Berglund et al, 1993
). However, the distribution of clinical stages at diagnosis is in line with the expected, with no favour of less advanced stages. As data on disease recurrence was not available for this study, the impact of SATB2 on recurrence-free survival, not least local recurrence in rectal cancer, should be assessed in future studies, preferably in cohorts where this information has been recorded prospectively.
In conclusion, the findings from this large cohort study demonstrate that high SATB2 expression is an independent factor of good prognosis in colon cancer and imply a putative role for SATB2 in mediating increased sensitivity to chemotherapy and radiation therapy in CRC. The mechanistic basis for these observations should be addressed in future studies.