Expression of Transketolase like gene 1 (TKTL1) predicts disease-free survival in patients with locally advanced rectal cancer receiving neoadjuvant chemoradiotherapy

The present analysis comprises patients with histologically confirmed, locally advanced non-metastatic rectal adenocarcinoma (endorectal ultrasound stage cT3-4, any N or cT2, N+ distal rectum). All patients participated in clinical trials of intensified neoadjuvant chemoradiotherapy including weekly irinotecan (40 - 50 mg/m²) and cetuximab (initial dose of 400 mg/m² then 250 mg/m²), and daily capecitabine (400 - 500 mg/m² b.i.d.) in combination with pelvic radiotherapy (45 Gy + 5.4 Gy) as previously described [19, 20]. Follow up of patients was carried out according to the German study group for gastrointestinal Oncology [21]. Patient characteristics are listed in Table 1.

Patients provided written informed consent for the participation in the clinical study as well as for the investigation of biopsy material. The clinical study protocol was reviewed and approved by the local institutional review board and the study was performed according to the Declaration of Helsinki.

Surgery was performed 4 - 6 weeks after the completion of chemoradiotherapy. The pathological routine work-up was described earlier [22].

Two classification systems were applied to describe response to chemoradiotherapy. The grade of histopathological regression has first been described using the Japanese Society for Cancer of the Colon and Rectum (JSCCR) grading system [23]. Tumours were classified as good-responders when assigned to tumour regression grades 2 or 3 (TRG 2 or TRG 3), and as bad-responders at regression grades 0 or 1 (TRG 0 or TRG 1).

Moreover, the histopathological downstaging after completion of preoperative chemoradiotherapy was used as surrogate parameter of tumour response, as previously described by Valentini et al [24]. A T-level downstaging of at least one T-level was considered to be a sign of response.

Tumour material was obtained during rectoscopy before the initiation of chemoradiotherapy and during surgery of the primary tumour. Tumour biopsies (n = 33) and matching healthy mucosa (n = 33) were stored in RNAlater (Quiagen, Hilden, Germany) in liquid nitrogen, and stored at -80°C until further extraction. RNA extraction and cDNA synthesis was performed according to standard protocols [25]. Total RNA was extracted after homogenisation of 15-30 mg tissue with the Ultra Turrax Tube Drive (Ika, Staufen, Germany) using TRIzol™ reagent (Invitrogen, Karlsruhe, Germany) according to the manufacturer's instructions. RNA was reversely transcribed using random hexamer priming and MMLV reverse transcriptase (Invitrogen).

Expression analysis of VEGFR-1, VEGFR-2 and TKTL1 was performed using the LightCycler instrument 1.5 (Roche Diagnostics, Mannheim, Germany). For sequences of amplification primers and hybridisation probes used see additional file 1. Each 20 μl reaction mix contained 4 μl LightCycler Faststart DNA Masterplus Hyb Probes Master Mix (Roche Diagnostics), 2 μl cDNA template or plasmid dilution, 0.5 μM forward primer and 0.5 μM reverse primer, 0.25 μM anchor probe and 0.25 μM sensor probe (TIB Molbiol, Berlin, Germany). Cycler conditions were the following: 10 min denaturation at 95°C, 50 cycles of 10 sec at 60°C (annealing VEGFR-1, -2, and TKTL1) and 26 sec at 72°C (elongation). A 5 log series of plasmid dilutions was amplified within the PCR runs for quantification of VEGFR-1, -2 and TKTL1. The LightCycler software prepares standard curves using linear regression analysis of the plasmid dilutions and calculates copy numbers of the unknown sample [26]. Values below the lowest standard dilution for TKTL1 (4 copies) and VEFGR-1/2 (40 copies) were assigned as negative. Beta-Glucuronidase (GUS) mRNA was quantified as internal control as previously described [27].

Cloning and transformation of the PCR products of TKTL1, VEGFR-1 and VEGFR-2 amplified from cell lines (SW480, K562 obtained from DSMZ, Braunschweig, Germany) was performed according to the manufacturer's instructions using the PCR2.1 TOPO vector (Invitrogen). Amplification reactions were undertaken for 32 cycles at 60°C annealing temperature. Cloning and transformation into Escherichia coli TOPO10F' was performed according to the manufacturer's instructions (Invitrogen). Plasmid DNA containing the desired construct was isolated using the Plasmid Mini Kit (Qiagen). Insertion sequences were confirmed by direct sequencing. The resulting plasmid was linearized by Not1 digestion (Roche Diagnostics). Dilutions of the linearized plasmid were prepared in 10 mM Tris-HCl (pH 8.0) and 1 mM ethylenediaminetetraacetic acid containing 20 mg/mL tRNA (Roche Diagnostics).

ß-Glucuronidase (GUS) mRNA transcripts were measured as an internal control using a standard plasmid (pME-2) containing BCR-ABL, ABL, and GUS sequences [27]

In order to minimize dilution error of different plasmids containing target and housekeeping gene a common plasmid containing TKTL1 and GUS was constructed for further use using a previously published method [27].

KRAS mutation analysis was performed for all samples by direct sequencing from DNA of microdissected tumor tissue samples as described [28]. PTEN status was determined by immunohistochemistry (IHC) using the PTEN antibody as described [28] (1:400, Cascade Bioscience, Winchester, MA, USA). All 33 patients were screened for PTEN mutations, 30 of which were evaluable. Three samples could not be taken into the analysis due to poor sample quality. Survivin expression analysis was performed on cDNA level of 30 patients from the studygroup using qPCR.

Differences in expression levels were compared using two-tailed Mann Whitney test. Differences in regression rates concerning the investigated tissue markers were evaluated by means of a two sided Fisher's Exact test. Disease-free survival was defined as the time between the start of chemoradiotherapy and tumour relapse (local failure and/or distant metastases) or death due to non-tumour related causes using the Kaplan-Meier method. Cut off level of median gene expression was used to divide low and high expressing groups. A p-value of less than 0.05 was considered statistically significant.

Patient and tumour characteristics are depicted in Table 1. Tumour tissue of a total of 33 patients (n = 23 male, n = 10 female, median age 61, range 33 - 76) was analysed. TKTL1 expression was determined in 33 patients, while VEGRFR-2 expression was analysed in 32 and VEGFR-1 expression in 26 patients only due to scantness of tumour tissue.

Normalised VEGFR-1/-2 expression of patients with LARC was significantly higher in pre-treatment tumour tissue in comparison to the corresponding normal mucosa (VEGFR-1; p = 0.0023 and VEGFR-2; p = 0.0005) but failed to be statistically significant for TKTL1 (p = 0.082). Accordingly, after completion of chemoradiotherapy higher VEGFR-1/-2 expression levels (p = 0.025; p = 0.063) were observed in tumour tissue as compared to healthy mucosa at time of surgery, whereas no differences in gene expression levels of VEGFR-1 and -2 were observed in tumour samples before and after neoadjuvant radiotherapy (Figure 1).

A total of 11 out of 33 patients showed a poor response (33%; defined as TRG of 0 or 1), while 22 patients had good response (67%; TRG 2 or 3, Table 1). Pathological T-downstaging was accomplished in 15 patients (45%, "T-responder"), 3 of whom (9%) achieved a pathological complete remission (ypT0 N0). No significant correlation between the two scores could be seen in our cohort (CI: 0,7-2,1, p-value:0,4).

Median pre-treatment expression levels of the three genes were used as a cut off value, dividing patients into low and high-expression groups (Figure 1) No significant correlation was observed neither for the JSCCR grading (TRG) nor the complete pathological response (pCR) nor for gene expression levels (Table 2).

All but one patient underwent curative surgery. The latter had an irresectable T4-tumour and revealed tumour progression with peritoneal spread during chemoradiotherapy. Median follow-up time was 33 months (range: 9-51). Local recurrence and distant metastases were recorded in two (12.5%) and 10 patients (65%), respectively.

A longer three-year DFS could be observed in patients with low TKTL1 expression (3-year DFS: 87% versus 39%, n = 33, p = 0.01, HR: 0.19, CI: 0,05-0,6), while three-year DFS was virtually identical for patients with high vs. low VEGFR-1 expression (3-year DFS: 69% versus 68%, n = 26, p = 0,9, HR: 1,04, CI: 0,26-4,2). The impact of VEGFR-2 expression on DFS failed statistical significance but a trend was observed regarding 3-year DFS in favour of patients with lower VEGFR-2 expression (3-year DFS: 81% versus 53%, n = 32, p = 0.19, HR:0,43, CI: 0,13-2,15; see Figure 2).

The high and low expressing patient groups are compared in Table 3 with respect to several clinical and molecular parameters. Median age in patients with a higher TKTL1 expression was younger compared to patients with low TKTL1 expression. No significant differences could be seen in initial tumour stage and TRG. Serum tumour markers in the patient group with higher TKTL1 expression were slightly higher, but this did not reach statistical significance (p-value for CEA p = 0.07, n = 16, p-value for CA 19.9, p = 0.18, n = 16). Patients with high TKTL1 levels eventually developed metastases or local recurrence significantly more often than patients with low TKTL1 levels (11 vs. 1 pts., p = 0.0002). TKTL1 expression was correlated to VEGFR-1/2 and to PTEN, KRAS and Survivin expression in our cohort. Expression of the latter molecular markers has previously been published by our group [28, 29]. No correlation between high TKTL1 expression and VEGFR-1 or -2 expression was demonstrated, while a tendency towards higher survivin expression in the TKTL1 overexpressing group could be detected (13 vs 17 pts., p = 0.08).