Thymidylate synthase, dihydropyrimidine dehydrogenase, ERCC1, and thymidine phosphorylase gene expression in primary and metastatic gastrointestinal adenocarcinoma tissue in patients treated on a phase I trial of oxaliplatin and capecitabine

The frozen samples that had been stored in RNALater^® were embedded in optimal cutting temperature (OCT) compound (Secure Fanatic U.S.A., Inc., Torrance, CA) and cut into serial sections with a thickness of 20 am. Sections were mounted on unbolted glass slides and stored at -80°C. For histology diagnosis, three representative sections, consisting of the beginning, middle and end of sectioning were stained with hematoxylon and eosin (H&E) by the standard method.

Before micro-dissection, sections were air-dried, fixed in 70% ethanol for 3 min and washed in water for 2 min. Afterwards, they were stained with nuclear fast red (NER, American MasterTech Scientific, Inc., Lodi, CA, USA) for 10 sec and again washed in water for 30 sec. Samples were then dehydrated in a stepwise manner with 70% ethanol, 95% ethanol and 100% ethanol for 30 sec each, followed by incubation in xylene for 5 min and complete air-drying. All H&E stained sections were evaluated by a pathologist. If the histology of the samples was homogeneous and contained more than 90% tumor tissue, samples were dissected from the slides using a scalpel. All other sections were selectively isolated by LCM (P.A.L.M. Microsystem, Leica, Wetzlar, Germany) according to the standard procedure [28]. The micro-dissected tissue was transferred to a reaction tube containing 400 μL of RNA lysis buffer.

For extraction from FFPE specimens, the tissue samples to be extracted were placed in a 0.5 mL, thin walled tube containing 400 μL of 4 M dithiothreitol (DTT)-GITC/sarc (4 M guanidinium isothiocyanate, 50 mM Tris-HCl, pH 7.5, 25 mM EDTA) (Invitrogen; #15577-018). The samples were heated at 92°C for 30 min and then transferred to a 2 mL centrifuge tube. To the tissue suspensions were added 50 μL of 2 M sodium acetate, pH 4.0, followed by 600 μL of freshly prepared phenol/chloroform/isoamyl alcohol (250:50:1). The tubes were vortex-mixed for 15 sec, placed on ice for 15 min and then centrifuged at 13,000 rpm for 8 min in a chilled (8°C) centrifuge. The upper aqueous phase (250–350 μL) was carefully removed and placed in a 1.5 mL centrifuge tube. Glycogen (10 μL) and 300–400 μL of isopropanol were added and the samples vortex-mixed for 10–15 sec. The tubes were placed at -20°C for 30–45 min to precipitate the RNA. The samples were then centrifuged at 13,000 rpm for 7 min in a chilled (8°C) centrifuge. The supernatant was poured off and 500 μL of 75% ethanol was added. The tubes were centrifuged at 13,000 rpm for 6 min in a chilled (8°C) centrifuge. The supernatant was carefully poured off so as not to disturb the RNA pellet and the samples were quick-spun for 15 sec at 13,000 rpm. The remaining ethanol was removed with a 20 μL pipette and the samples air-dried for 15 min. The pellet was re-suspended in 50 μL of 5 mM Tris. For RNA extraction from OCT embedded samples, the above procedure was followed, except that the step involving heating at 92°C for 30 min was omitted.

cDNA was prepared as previously described [29]. Quantitation of TS, TP, ERCC1 and DPD and an internal reference gene (β-actin) was done using a fluorescence based real-time detection method (ABI PRISM 7900 Sequence detection System, Perkin-Elmer (PE) Applied Biosystem, Foster City, CA, USA). The PCR reaction mixture consisted 1200 nM of each primer, 200 nM probe, 0.4 U of AmpliTaq Gold Polymerase, 200 nM each dATP, dCTP, dGTP, dTTP, 3.5 mM MgCl₂ and 1× Taqman^® Buffer A containing a reference dye, to a final volume of 20 μl (all reagents from PE Applied Biosystems, Foster City, CA, USA). Cycling conditions were 50°C for 2 min, 95°C for 10 min, followed by 46 cycles at 95°C for 15 sec and 60°C for 1 min. The primers and probes used are shown in Table 1.

The cycle after which the fluorescent signal exceeds the threshold is referred to as the Ct. The Ct is inversely proportional to the amount of cDNA in the tube, i.e., a higher Ct value means that more PCR cycles are required to reach a certain level of detection. Gene expression values (relative mRNA levels) are expressed as ratios (differences between the Ct values) between the genes of interest and an internal reference gene (β-actin) that provides a normalization factor for the amount of RNA isolated from a specimen. Each cDNA sample had a duplicate genomic sample analyzed by real-time PCR without reverse transcription. If a fluorescent signal above the threshold was seen in the no reverse transcription specimen, the results from the corresponding cDNA were not deemed reliable. Each plate contained a reference cDNA sample to allow normalization for potential plate to plate variability.

Validation of the assay was performed with regard to RNA extraction and recovery, accuracy, precision, limits of quantitation and range. These studies followed GLP guidelines and employed Stratagene reference human RNA mix, pooled tissue homogenates derived from FFPE tissue sections and multiple sections of FFPE tumor sections from a variety of tumors. Assay precision was inferred from the variation in the replicate Ct values for all genes used, and had no more than 5.3% coefficient of variation (CV). The Delta Ct values derived from these Ct values showed a maximum %CV of 9.5%. This was expected, since the Delta Ct values reflect the variation in the Ct values of two genes, the target gene and its actin reference gene.

Statistical analyses were performed using SigmaStat for Windows Version 3.11 (Systat Software Inc., Richmond, CA, USA). The median, first and third quartile values for each molecular marker were determined using all informative samples. The distribution of response vs. stable disease vs. early progressive disease as a function of molecular marker expression was analyzed by the chi-square test, whereas differences in response plus stable disease vs. early progressive disease was by Fisher's exact test. The strength of association of two variables was determined by the Pearson product moment correlation. Analysis of progression-free survival was performed by log-rank. Overall survival was analyzed by the Gehan-Breslow method. Graphs were prepared using SigmaPlot for Windows version 9.0 (Systat Software, Inc., Chicago, IL).

Ninety-one subjects participated in this clinical trial (Table 2). One patient withdrew after having a tumor biopsy performed, and did not receive any protocol therapy. Six additional patients are not assessable for response. Among 84 subjects who were considered assessable for response to protocol therapy, there were four complete and 14 partial responses (21%). Forty-five patients had stable disease (54%), while 21 (25%) experienced clinical or radiographic evidence of disease progression by the time of initial restaging after 3 cycles. For all 91 subjects, the median time to treatment failure was 5.1 months, and the median survival was 10.9 months.

Biopsies of metastatic tumor tissue were obtained in 81 subjects. Fifteen of these samples were not informative for any of the four targets of interest, either due to absence of tumor in the sample, or an insufficient quantity of tissue. The number of samples that were informative for each target was as follows: TS, 65; DPD, 63; TP, 64; ERCC1, 64. Archival paraffin-embedded formalin fixed tissue blocks from the primary tumor were available in 25 subjects. No tumor was seen in one of these. One additional subject had a biopsy of a rectal primary that was stored in RNALater^®. The median time from initial diagnosis to study entry in subjects who had both archival primary tumor tissue and metastatic tumor tissue was 671 days (25^th%, 426; 75^th%, 824).

Table 3 shows the results of the gene expression for the pre-selected molecular targets. There was much greater variability in the distribution of values for each molecular target in metastatic tissue: the ratio of the maximum to minimum values ranged from 24 to 95. In primary tumor tissue, in contrast, the ratio of maximum to minimum values for each target ranged from 9 to 14.

When mRNA expression was compared in paired samples of primary and metastatic tumor tissue, gene expression was significantly higher in metastatic tumor tissue for each molecular marker: TS, 1.9-fold; DPD, 3.8-fold; TP, 1.6-fold; ERCC1, 2.1-fold (Figure 1).

The strength of association between doublets of molecular markers was analyzed. In both primary and metastatic tumor tissue, there was a strong correlation between the expression of DPD and TP (Figure 2). No other significant correlations were observed.

Restaging computerized tomographic scans were planned after the first three cycles (nine weeks). Individual target gene expressions were stratified as above or below either the median, first or third quartile. The proportion of observations in the different categories that defined the contingency tables were analyzed by comparing complete and partial response (CR and PR) versus stable disease (SD) versus early progressive disease (PD, within the first 9 weeks), or CR plus PR plus SD versus early PD. The only association that appeared to be non-random was with TS (Table 4). Patients whose intra-tumor TS mRNA levels were above the median value had significantly greater risk of early disease progression (43% vs 17%). Subjects who had very high intra-tumor ERCC1 expression tended to have an increased risk of early PD (47% vs 22%).

Because a major endpoint of this study was tolerability of the regimen, a primary endpoint was time to treatment failure, which was defined as the interval from enrollment on study to either documented disease progression or the date a patient withdrew from the study for unacceptable toxicity. Only two patients withdrew from study due to unacceptable toxicity accompanied by a decline in performance status. Neither patient had a restaging CT scan, so the date of progression cannot be determined. When individual molecular marker expression was analyzed as a possible predictive factor for TTF, those subjects whose metastatic tumor tissue had ERCC1 gene expression above the third quartile had a significantly shorter TTF of 77 days (Figure 3). If the two patients who withdrew for toxicity are excluded, the result did not appreciably change: those with ERCC1 gene expression above the third quartile was significantly shorter: median TTF 85 days vs 166 days (p = 0.017). Various combinations of molecular markers did not discriminate any apparent differences in TTF.

Overall survival from the on-study date was analyzed according to higher versus lower mRNA expression of individual targets. The only individual molecular marker that tended to distinguish those with worse or better survival was TS (Figure 4). The median survival in subjects whose metastatic tumor tissue had TS mRNA expression below the median was 123 days longer than those with TS expression above the median. Various combinations of molecular marker expression were analyzed (e.g. TS below median and ERCC1 below third quartile vs others), but none yielded stronger prognostic information than TS mRNA expression alone.