Introduction
For locally advanced rectal cancers (LARC), neoadjuvant radiochemotherapy (nRCT) followed by surgery according to the principles of total mesorectal excision is the standard-of-care approach that can achieve downstaging rates as high as 28–62% of the cases, improved local control, higher rates of resectability and sphincter-sparing procedures (Villemure et al.
2003). Local pathological response to radiochemotherapy impacts on outcome rates after nRCT, irrespective of tumor stage (Mohindra et al.
2002). However, complete response rates vary between only 5–25%, while around 50% of rectal cancer patients respond poorly or are non-responsive to nRCT (Park et al.
2012; Ferlay et al.
2015; Lorimer et al.
2017). Predictive markers are needed that allow an accurate selection of patients, who will most likely benefit from nRCT or those who could be spared therapy that is associated with high comorbidities.
Aberrant fibroblast growth factor (FGF) and FGF receptor (FGFR) signaling has been reported for colorectal cancer and supports tumor cell survival (Shimokawa et al.
2003; Sonvilla et al.
2008,
2010; Heinzle et al.
2012). Recently, FGFR4 has been described to inhibit nRCT response in rectal cancer by stimulation of homologous recombination repair of radiation-induced DNA damage (Ahmed et al.
2016). However, this mechanism only works for mismatch repair (MMR)-competent tumor cells, leaving the question whether additional survival mechanisms operate in MMR-deficient lesions.
FGF8, an essential factor during embryonic development (Tickle and Munsterberg
2001; Brewer et al.
2016), is overexpressed in several tumor types including prostate, ovarian, breast, hepatocellular and colorectal cancers (Tanaka et al.
1998; Mattila and Harkonen
2007; Gauglhofer et al.
2011; Liu et al.
2014). The factor stimulates anti-apoptotic pathways (Zhang et al.
2006; Mattila and Harkonen
2007) and prevents tumor cell death (Gauglhofer et al.
2011) mediated by the IIIc-splice variants of receptors FGFR1-3 and FGFR4 (Dammann et al.
2015b). Up-regulation of FGF8 and FGFR3 was observed in CRC cells and xenografts receiving irinotecan-based chemotherapy, which led to induction of chemotherapy resistance (Erdem et al.
2017). In hepatocellular carcinoma, FGF8 is involved in resistance to epidermal growth factor receptor targeting therapy (Pei et al.
2017).
To investigate whether FGF8-dependent survival signaling is involved in nRCT resistance in rectal cancer, expression levels of FGF8 and the anti-apoptotic protein Survivin were analyzed using immunohistochemistry (IHC) in human rectal cancer tissue obtained from pre-nRCT biopsies and surgical specimens. Radiation-resistant cell line models were used to assess the role of FGF8 anti-apoptotic response at a molecular level.
Materials and methods
Patients and clinical samples
Biopsy specimens were collected retrospectively from the pathology department from 43 patients with rectal cancer who received nRCT at the General Hospital of Vienna during the years 2012–2014. The patients gave their informed consent and biopsies were taken during colonoscopic examination before preoperative radiotherapy. Tumor specimens were collected at surgery. The study protocol was approved by the ethics committee of the Medical University of Vienna (EK-1350/2016). All patients received a neoadjuvant regimen of capecitabine plus a total of 50.4 Gy radiation dose. Response to radiotherapy was determined by histopathological examination of surgically resected specimens and classified according to the amount of viable tumor cells in the resected tissue (Dworak et al.
1997).
Immunohistochemistry and scoring
Immunohistochemistry (IHC) was performed on paraffin-embedded sections from human rectal cancer tissue as described previously (Dammann et al.
2015a). Details are given in supplemental materials (Supplementary Table 2).
Scoring was performed by two blinded investigators (F. Harpain, G. Jomrich) as previously described (Dammann et al.
2015b). Points were given for staining intensity from 0 to 3 and for the percentage of positively stained epithelial cells from 0 to 4, 0 (< 1%), 1 (1–10%), 2 (10–50%), 3 (51–80%), and 4 (> 80%) in four separate fields of view. An immunoreactivity score (IRS) was generated by multiplication of the mean intensity and percentage scores. Scores greater or equal to 6 were considered as “high”, scores smaller than 6 considered as “low”.
Cell lines
HCT116 and HT29 were obtained from the American Type Culture Collection. DLD1 was obtained from European Culture Collections. All cell lines were authenticated by Eurofins (Vienna, Austria). They were kept under standard culture conditions (5% CO2 at 37 °C) using minimal essential medium containing 10% fetal calf serum (FCS) (Sigma-Aldrich, St. Louis, USA).
Ionizing radiation and in vitro radiosensitivity assay
Cells were irradiated using a Co-60 radiotherapy unit (Theratron 760, Theratronics, Ottawa, Canada). The surviving fraction of cells was determined by the clonogenic assay and calculated relative to the non-irradiated control (Franken et al.
2006).
For the selection of surviving cell clones, irradiation was performed using one or two series of 5 × 2 Gy doses of γ-radiation over a week each. After the final radiation dose, cultures were maintained for 10 days to permit the growth of long-term-surviving clones. Clones were harvested and pooled for analysis.
RNA isolation and quantitative real-time PCR assay
Total RNA was isolated using Trifast (PeqLab, Germany) according to the manufacturer’s instructions, reversely transcribed and cDNA amplified using TaqMan-based assay performed on an ABI 7500 fast real-time PCR system (Applied Biosystems, Foster City, California, USA), as previously described (Sonvilla et al.
2010). The TaqMan kits used are listed in Supplementary Table S3.
Protein isolation and western blotting
Proteins were extracted using HEPES lysis buffer supplemented with protease inhibitor cocktail (Complete—Roche, Germany) and phosphatase inhibitors and analyzed by western blotting using antibodies listed in Supplementary Table S2. Detection was performed using ECL Western Blot Detection Reagents (GE Healthcare).
Statistical analysis
Statistical analysis was performed using SPSS (version 24.0). Immunoreactivity scores were analyzed using an independent sample two-tailed t test. Pearson’s χ2 test was used for analyzing the association between FGF8/Survivin expression and clinicopathologic parameters. P values < 0.05 were considered significant (*p < 0.05, **p < 0.01, ***p < 0.001). All data are expressed as mean ± standard deviation unless otherwise stated.
Discussion
nRCT followed by surgical resection remains the standard therapy regimen for locally advanced rectal cancer patients (Villemure et al.
2003). However, therapy response varies widely (Ramzan et al.
2014) and radiotherapy may cause severe adverse events in irradiated patients. Thus, identifying biomarkers with the potential to predict therapy response is of utmost importance to enable the selection of a patient population, where benefits of radiotherapy clearly outbalance disadvantages.
Ionizing radiation mainly aims to damage and kill tumor cells by inducing DNA double-strand breaks (Sada et al.
2018). Cell cycle arrest and DNA damage repair through the homologous recombination or non-homologous end-joining pathway are physiological responses to the inflicted damage (Ramzan et al.
2014; Dammann et al.
2015a). If repair is unsuccessful, cells undergo programed cell death. Consequently, deregulation in repair and/or survival signaling impacts on radiation response and key regulators of this machinery may predict therapy response. Specifically, Survivin has previously been suggested as a sensitivity marker based on cell line models (Rodel et al.
2003) and FGFR-dependent signaling plays a central role due to its physiological functions (Turner and Grose
2010; Heinzle et al.
2011), especially in CRC (Sonvilla et al.
2008,
2010; Koneczny et al.
2015; Erdem et al.
2017).
In this study, protein levels of FGF8 and Survivin were strongly correlated with therapy response in a cohort of 43 rectal cancer patients indicating that they are promising candidates as predictive markers for nRCT response. Both markers strongly correlated with each other suggesting a mechanistic connection. This is conceivable based on our observation that FGF8 induced Survivin expression in DLD1 cells. The mechanism may be direct stimulation by FGFR-dependent survival signaling or indirect through the Wnt pathway, as Survivin is a Wnt-target gene (Widelitz
2005; Xu et al.
2017) and FGF signaling enhances β-catenin-dependent
transcription activity (Koneczny et al.
2015).
In vitro, Survivin up-regulation was evident for DLD1 and HCT116 cells (MMR deficient), but not for HT29 cells (MMR competent), leaving the question whether this reaction may be specific for MMR-deficient tumors. However, a relatively large proportion of high-Survivin-expressing rectal tumors (28/43; 65%) was found in our cohort although in the literature MMR-deficient tumors in the rectum are described less frequently (de Rosa et al.
2016). Rather the difference may be due to the high degree of radiation resistance, inherent already in HT29 control cells, as they up-regulate FGFR4, which mediates efficient repair of radiation-induced DNA damage (Ahmed et al.
2016).
The impact of FGF survival signaling on radiation response was also demonstrated by analysis of irradiated cell line models. Specifically, DLD1 cells, whose baseline FGF signaling activity is low, showed strong induction of FGF8 and FGFR3 mRNAs immediately after exposure. For the induction of resistance mechanisms, a fractionated irradiation protocol was chosen to model therapeutic treatment schedules. Long-term survivors of this treatment were more resistant to a further radiation dose than non-irradiated control cells. In survivors of all three cell lines, FGFR3-IIIc, the receptor mediating survival signaling of FGF8 and FGF18 (Zhang et al.
2006; Sonvilla et al.
2010), was significantly increased on mRNA level as compared to non-irradiated control cells. In addition, FGFR3 protein was elevated and phosphorylated, indicating an active signaling state. Based on our results, up-regulation of FGF-dependent survival signaling has to be viewed as a general phenomenon in response to radiation damage in CRC cells, independent of MMR status. A similar effect was described in CRC cells exposed to irinotecan, which induced FGF8 and FGFR3 expression and thereby hampered therapy response (Erdem et al.
2017). Unfortunately, the relevant receptor (FGFR3-IIIc) cannot be used as a tissue marker, because its expression is generally too low (Sonvilla et al.
2010; Erdem et al.
2017). However, the receptor may provide a useful target for combination therapy. When applied after irinotecan treatment, an FGFR inhibitor had a synergistic effect (Erdem et al.
2017). FGFR inhibitors usually block FGFR1–3 and somehow less also FGFR4 (Heinzle et al.
2011,
2014). FGFR inhibition might be a useful strategy in the neoadjuvant setting for those patients that express high levels of FGF8.
In summary, the strong connection between FGF8/Survivin expression and therapy response in rectal tumor tissue underlines that they are predictive markers not restricted to mismatch repair-deficient tumors alone. Whereas the results of our previous study identified low levels of FGFR4 protein as a marker for good response to radiation (Ahmed et al.
2016), this study now demonstrates that high FGF8 and Survivin protein levels identify tumors that respond insufficiently or not at all.