应用傅里叶变换红外光谱(FTIR)联合主成分分析法(PCA), 分析直肠癌转移淋巴结的谱学特征, 并对直肠癌转移淋巴结和非转移淋巴结进行线性判别分析。 80例直肠癌转移淋巴结和80例未转移淋巴结进行FTIR光谱分析, 计算峰强并进行主成分分析, 得出在波数4 000～1 700 cm-1范围主成分1(Principal components 1, PC1)是3 260 cm-1, PC2为1 740 cm-1。 波数1 700～1 000 cm-1范围, PC1为1 640 cm-1, PC2为1 080 cm-1, 将良、 恶性淋巴结光谱3 260, 1 740, 1 640, 1 080 cm-1相对峰强比(I/I1 460)和波数1 080和1 300 cm-1进行t检验, 良、 恶性结果差异有统计学意义(p<005), 表明癌转移淋巴结中蛋白含量、 蛋白的形成、 氨基酸增多； 脂肪含量明显减少与癌组织中无氧酵解脂肪含量减少有关。 将相对峰强比(I1 080/I1 460, I1 640/I1 460, I3 260/I1 460, I1 740/I1 460, n=160)进行PCA聚类分析, 结果显示可以将良恶性淋巴结鉴别, 良性淋巴结聚类在第一和四象限, 恶性淋巴结聚类在二和三象限。 将相对峰强比、 1 080和1 300 cm-1进行线性判别分析(LDA), 将25例淋巴结作为验证集进行分析, 得出PCA/LDA模型的敏感度是875%, 特异度是885%。 结果表明傅里叶变换红外光谱分析技术可成为术中原位、 在体和快速诊断直肠癌淋巴结转移的一种简便方法。

The aim of the research is to estimate Fourier transform infrared spectroscopy (FTIR) as a novel diagnostic tool for rectal cancer metastatic lymph nodes. 160 freshly normal and malignant lymph nodes were collected from 80 rectal cancer patients for spectrum analysis. FTIR spectra were generated from tissues samples using infrared wavenumber from 4 000 to 1 000 cm-1 region. The ratios of spectral intensity and relative intensity ratios (I/1 460) were calculated. Principal component analysis (PCA) and Fisher’s Linear discriminant analysis (LDA) were applied to distinguish the malignant from normal. From 4 000 to 1 700 cm-1, principal components 1 (PC1) was 3 260 cm-1 and PC2 was 1 740 cm-1; from 1 700 to 1 000 cm-1, PC1 was 1 640 cm-1 ab PC2 was 1 080 cm-1. Four prominent significant wavenumbers at 1 080 , 1 640, 1 740 and 3 260 cm-1 separated cancer spectra from normal spectra into two distinct groups using PCA. T test associated with relative intensity ratios (I1 080/I1 460, I1 640/I1 460, I3 260/I1 460, I1 740/I1 460) revealed that these wavenumbers were also able to distinguish cancer and normal lymph nodes spectra. The bands 3 260, 1 640, 1 550, 1 080 and 1 740 cm-1, associated with Protein, nucleic acid and lipid. Compared with normal lymph nodes tissues, malignat lymph nodes significant increased (p<0.05), but 1 740 cm-1 decreased. Six parameters (wavenumbers 1 080 and P1 300 cm-1, relative intensity ratios I1 080/I1 460, I1 640/I1 460, I3 260/I1 460, I1 740/I1 460) were selected as independent factors to perform discriminant functions. The sensitivity for PCA/LDA mode in diagnosing lymph nodes was 87.5% by discriminant analysis. The results demonstrate that FTIR is as a useful technique for detection malignant and normal lymph nodes, FTIR may be applied in clinical as a noninvasive, high sensitivity and specificity method for lymph nodes diagnosis.