Integrative metabonomics as potential method for diagnosis of thyroid malignancy

Abstract

Thyroid nodules can be classified into benign and malignant tumors. However, distinguishing between these two types of tumors can be challenging in clinics. Since malignant nodules require surgical intervention whereas asymptomatic benign tumors do not, there is an urgent need for new techniques that enable accurate diagnosis of malignant thyroid nodules. Here, we used (1)H NMR spectroscopy coupled with pattern recognition techniques to analyze the metabonomes of thyroid tissues and their extracts from thyroid lesion patients (n = 53) and their adjacent healthy thyroid tissues (n = 46). We also measured fatty acid compositions using GC-FID/MS techniques as complementary information. We demonstrate that thyroid lesion tissues can be clearly distinguishable from healthy tissues, and malignant tumors can also be distinguished from the benign tumors based on the metabolic profiles, both with high sensitivity and specificity. In addition, we show that thyroid lesions are accompanied with disturbances of multiple metabolic pathways, including alterations in energy metabolism (glycolysis, lipid and TCA cycle), promotions in protein turnover, nucleotide biosynthesis as well as phosphatidylcholine biosynthesis. These findings provide essential information on the metabolic features of thyroid lesions and demonstrate that metabonomics technology can be potentially useful in the rapid and accurate preoperative diagnosis of malignant thyroid nodules.

Figure 1

The color photomicrographs of healthy adjacent thyroid tissue (A), nodular goiter (B), follicular adenoma (C) and papillary thyroid carcinoma (D).

Figure 2. Representative 600 MHz ¹H spectra of intact thyroid tissue (T) and tissue aqueous extracts (E) originating from healthy adjacent thyroid tissue (T₁, E₁), benign thyroid lesion (T₂, E₂) and malignant thyroid lesion (T₃, E₃).

The regions of δ 0.8–2.9 and δ 5.2–8.5 were vertically expanded 2 and 16 times compared with the region of δ 2.9–4.2. Keys: 1, lipid; 2, isoleucine; 3, leucine; 4, valine; 5, lactate; 6, threonine; 7, alanine; 8, lysine; 9, arginine; 10, acetate; 11, glutamate; 12, methionine; 13, glutamine; 14, aspartate; 15, glutathione (GSH); 16, choline; 17, phosphocholine (PC); 18, glycerophosphocholine (GPC); 19, taurine; 20, scyllo-inositol; 21, myo-inositol; 22, glycine; 23, phosphoethanolamine (PE); 24, inosine; 25, tyrosine; 26, phenylalaine; 27, histidine; 28, fumurate; 29, uracil; 30, guanosine; 31, hypoxanthine; 32, xanthine; 33, formate; 34, acetamide; 35, succinate; 36, citrate; 37, uridine; 38, U1.

Figure 3. 3D PCA scores plots obtained from NMR data of intact thyroid tissues (left) and tissue aqueous extracts (right).

(A) healthy adjacent thyroid tissues (), (B) benign thyroid lesions (), (C) malignant thyroid lesions (). Benign groups: NG and FA. Malignant groups: PTC.

Figure 4

ROC curve obtained from the cross-validated predicted Y-values of the ¹H NMR OPLS-DA model of intact thyroid tissue (A,C) and tissue aqueous extracts (B,D), showing the sensitivity and specificity of predictive models in discriminating between thyroid lesions and adjacent healthy thyroid tissue (A,B), and between malignant thyroid lesions and benign thyroid lesions (C,D).

Figure 5

OPLS-DA scores plots (left) and coefficient plots (right) generated from ¹H NMR spectra of intact thyroid tissue (A,C) and tissue aqueous extracts (B,D) showing the discrimination between healthy adjacent thyroid tissue (), thyroid lesions (), benign thyroid lesions () and malignant thyroid lesions (). Key to metabolite is given in Table 1.

Figure 6. GSSG and GSH contents in healthy adjacent thyroid tissue (), benign thyroid lesions () and malignant thyroid lesions ().

*p < 0.05 when compared to healthy adjacent thyroid tissues.

Figure 7. The levels of fatty acids measured by GC-FID/MS.

(A) healthy adjacent thyroid tissue (), (B) benign thyroid lesions () and (C) malignant thyroid lesions (). *p < 0.05 when compared to healthy adjacent thyroid tissue, Δp < 0.05 when compared to benign thyroid tissue.