The following 14 nucleoside standards including the internal standard 8-bromoguanosine hydrate (Br8G) were obtained from Sigma (St. Louis, MO, USA): pseudouridine (Pseu), cytidine (C), uridine (U), 1-methyladenosine (m1A), inosine (I), 5-methyluridine (m5U), guanosine (G), 1-methylinosine (m1I), 1-methylguanosine (m1G), N⁴-acetylcytidine (ac4C), N²-methylguanosine (m2G), adenosine (A), N², N²-methylguanosine (m22G), N⁶-methyladenosine (m6A). Methanol (MeOH) was HPLC-grade purchased from Tedia (Fairfield, OH, USA). Ammonium acetate (NH₄AC), ammonia (NH₃·H₂O) and potassium dihydrogenphosphate (KH₂PO₄) were all analytical reagents obtained from China. Water was deionized and purified by a Milli-Q system (Millipore, Bedford, MA, USA).

The HPLC system (Figure 1) consisted of three Shimadzu LC-10ATVP pumps (Kyoto, Japan), an autoinjector model SIL 10ADVP, an SPD-10AVP UV-Vis detector and an SCL 10AVP interface. An electric six-port valve (Rheodyne, USA) was used for the automated column switching. Valve switching and data acquisition were done on Shimadzu Class-VP version 6.10 software. The column 1 (40 mm×4.0 mm ID) was packed with a laboratory prepared boronic acid-substituted silica material. It can tolerate pH values of the buffers from 2 to 12 as well as the usual organic solvents. The column 2 (250 mm×4.6 mm ID) was packed with 5 µm Hypersil ODS₂ (Elite, Dalian, China).

Open in New Tab   Full Size Figure   Download Figure

Figure 1 Schematic diagram of the column-switching HPLC system.

Sixty healthy adults (31 males, 29 females, from 21 to 71 years, median age 52 years), who have the normal physical indices including hepatic function, renal function, chest X-ray and colonoscopy during a regular physical examination period in our institute, have been chosen as control material. Fifty-two patients with colorectal cancer and 10 patients with intestinal villous adenoma were from Ruijin Affiliated Hospital of Shanghai Second Medical University, the First and Second Affiliated Hospitals of Dalian Medical University of China. No patient had received chemotherapy or radiation therapy before surgery. Diagnoses of colorectal cancer and intestinal villous adenoma were made on the basis of usual clinical and laboratory findings and were confirmed by histopathology. Table 1 shows the clinical pathological parameters of patients with colorectal cancer.

Spontaneous urine samples were collected from healthy adults, patients with colorectal cancer and intestinal villous adenoma. In 43 patients with colorectal cancer, urine samples were also obtained 2 wk after surgery. All persons had normal renal function and were free of bacterial infection at the time when the urine was collected. After collection the samples free of preservatives were frozen immediately and stored at -20°C. Prior to analysis, the samples were thawed at room temperature and adjusted to pH 8.0 with 50 mL/L NH₃·H₂O and vortex for 3 min at 5 000 r/min. Aliquots of 1 mL centrifuged urine containing 30 µL of Br8G (0.30 mmol/L) were transferred to autosampler vials and samples of 150 µL were injected to a column-switching HPLC system.

Column 1 (Figure 1) was equilibrated for 5 min with the mobile phase delivered by pump 1. After sample injection (150 µL urine), column 1 was washed for 7 min with the same buffer. During that time, nucleosides were selectively retarded on the column 1 and the sample matrix was discharged. At the same time, column 2 was conditioned with the mobile phase delivered by pump 2 (position 1; Figure 1). After this clean-up step, column 1 was series-connected in front of the column 2. The group-specifically bound nucleosides on the column 1 were then eluted and concentrated on top of the column 2 over a period of 3 min (position 2; Figure 1). Separation of nucleosides on the column 2 was carried out with a linear gradient elution program over 40 min, while the column 1 was regenerated for a new extraction cycle (position 1; Figure 1). The nucleosides were detected at 260 nm and quantified using the internal standard method. Table 2 shows the time events used for the analytical procedure.

Peak identification was performed on the basis of retention times. Standard solutions were run daily before and after the samples to monitor reproducibility of retention times. The standard addition method was also used to confirm peak identification. The levels of the urinary nucleosides were calculated by the calibration curves, and then were transformed into nmoL/µmoL creatinine. Urinary creatinine levels were determined as described by Zheng et al[7].

The mean excreted amounts of urinary nucleosides have been calculated using MS Excel software. Differences of urinary nucleoside concentrations of healthy adults, patients with intestinal villous adenoma and colorectal cancer were compared with SPSS 10.0 software. Spearman Correlation Analysis was used to examine the relationship of urinary nucleoside concentrations from patients of colorectal cancer with the tumor size, clinical stage and differentiation. Principal component analysis (PCA) software, a home-made pattern recognition software, was used to classify healthy controls, patients with intestinal villous adenoma and colorectal cancer. It was also used to monitor changes in the urinary excretion of nucleosides before and after surgery. Principal components plots were drawn on the basis of the first principal component analysis function (PC1) against the second principal component analysis function (PC2) of the nucleosides for each urine specimen. The oblique rotational method used in the software is the Promax method[6].

Figure 2A is a standard chromatogram showing the resolution of 14 nucleosides as well as the internal standard Br8G under the condition newly developed. The calibration curves for 14 nucleosides, linear responses in peak area ratios of nucleosides to the internal standard vs the nucleoside concentrations were obtained with the correlation coefficients varying from 0.995 to 0.9995. The intra- and inter-day precisions of the method were determined by five repetitive analyses of an aqueous solution of standard nucleosides on three nonconsecutive days. The relative standard deviation (RSD) of retention times was less than 1.70% (intraday) and 3.51% (interday), and that of peak areas was less than 3.84% (intraday) and 7.85% (interday). The limits of detection ranged from 0.05 to 0.56 µmol/L, which are better than the previous reports[13,15,16], thus being suitable for the quantitative analysis.

Open in New Tab   Full Size Figure   Download Figure

Figure 2 Typical chromatograms of (A) 14 standard nucleoside mixtures (B) urinary nucleosides of a normal urine obtained under the established analysis conditions. Column-switching HPLC conditions as Table 2. Peak identification: 1 Pseu; 2 C; 3 U; 4 m1A; 5 I; 6 m5U; 7 G; 8 m1I; 9 m1G; 10 ac4C; 11 m2G; 12 A; 13 m22G; 14 Br8G; 15 m6A.

When applied to the urine specimens from 60 healthy adults, 10 patients with intestinal villous adenoma and 52 patients with colorectal cancer, a total of 14 nucleosides were positively identified. A typical chromatogram of urinary nucleosides from a normal person with peak identification is given in Figure 2B.

Fourteen nucleoside concentrations of healthy adults (group 1), patients with intestinal villous adenoma (group 2) and patients with colorectal cancer (group 3) are listed in Table 3. In the mean values of three groups, the most abundant nucleoside was Pseu, followed by m1A and m1I. Pseu level was elevated above the normal values plus two standard deviation (s) in the 58% (30/52) of patients with colorectal cancer, while 20% (2/10) of patients with intestinal villous adenoma was elevated. Clearly, the concentrations of 12 nucleosides (except U, G) were significantly elevated in patients with colorectal cancer (P < 0.05). Only four kinds of nucleoside concentrations of patients with intestinal villous adenoma were higher in comparing with those of healthy adults. Using 14 nucleoside concentrations as the data vectors for PCA technique, 71% (37/52) patients with colorectal cancer was distinguishable from healthy adults, while healthy adults are correctly classified at 96% (58/60) specificity. Serum CEA is the tumor marker being used in clinic to diagnose colorectal cancer (cut-off level: 5 mg/L). But the sensitivity of CEA method is 29% (15/52) based on the data provided by the hospitals. Based on the equation of classification of patients with colorectal cancer and healthy adults, 14 urinary nucleoside concentrations of patients with intestinal villous adenoma were calculated and the results was marked into Figure 3A, it was found that 10% (1/10) of patients with intestinal villous adenoma were in the area of patients with colorectal cancer.

Open in New Tab   Full Size Figure   Download Figure

Figure 3 PCA based on 14 nucleoside concentrations from healthy controls (+) and patients with colorectal cancer (o). A: Positions of patients with intestinal villous adenoma (*) were marked into the figure based on classification equation from healthy controls and patients with colorectal cancer; B: Positions of 32 of responsive cases after surgery (*) were marked into the figure based on classification equation from healthy controls and patients with colorectal cancer; C: Positions of 11 of progressive cases after surgery (*) were marked into the figure based on classification equation from healthy controls and patients with colorectal cancer.

We examined the relationship of urinary nucleoside concentrations from patients of colorectal cancer with the tumor size, Duke’s stages and differentiation. These results are listed in Table 4. The level of m1G, Pseu and m1A were positively correlated with the tumor size and Duke’s stages of colorectal cancer, respectively (P < 0.05). No significant correlation was noted in observed values with regard to tumor differentiation.

The changes in six urinary modified nucleoside concentrations (Pseu, m1A, m1I, m1G, ac4C, m22G) before and after surgery in 43 patients with colorectal cancer were studied. The patients were classified into two groups: response group (32 persons) and progressive disease group (11 persons). Using the paired t-test, these urinary modified nucleoside concentrations of response group before surgery were significantly higher than those of patients after surgery (P < 0.05). However, the pre- and post-surgery difference in levels of these nucleosides for progressive disease group had no significant changes.

PCA technique based on 14 nucleoside concentrations as the data vectors was used to monitor changes in the urinary excretion of nucleosides for patients with colorectal cancer before and after surgery. When urinary nucleoside concentrations of patients with colorectal cancer after surgery were fed to the principal component regression and marked to the space produced by the first two principal components (PC1, PC2) of healthy adults and patients with colorectal cancer before surgery (the process was similar to the position prediction of the patients with intestinal villous adenoma), it can be seen that points of 84% (27/32) patients with effective treatment have come back to the normal person area (Figure 3B). In the mean time, points of 91% (10/11) patients with ineffective treatment have entered into the area of colorectal cancer (Figure 3C).