Serum Differential Proteomic Analysis And Screening Model Establishment Using SELDI-TOF-MS In Patients With Papillary Thyroid Carcinoma

Objective To compare the differences in serum protein fingerprint between papillary thyroid carcinoma (PTC) patients with distant metastasis and those without distant metastasis, and establish a screening model that may be of help to serologic screening of PTC with distant metastasis; and to compare the differences in serum protein fingerprint between PTC patients with ¹³¹I-avid lung metastasis and those with non-¹³¹I-avid lung metastasis, and establish a screening model that may be of help to serologic screening of the ¹³¹I uptake function in lung metastasis from PTC.Methods Two pairs of serum samples were collected and tested, respectively. The first pair: Serum of PTC patients with distant metastasis and those without distant metastasis. Serum samples were collected from 36 PTC patients with distant metastasis and 29 PTC patients without distant metastasis, and randomized into a training set (26 patients with distant metastasis and 19 patients without distant metastasis) and a blinding test set (10 patients with distant metastasis and 10 patients without distant metastasis). The second pair: Serum of PTC patients with ¹³¹I-avid lung metastasis and those with non-¹³¹I-avid lung metastasis. Serum samples of 46 PTC patients with ¹³¹I-avid lung metastasis and 23 PTC patients with non-¹³¹I-avid lung metastasis were collected. Among them, 28 cases (19 patients with ¹³¹I-avid lung metastasis, and 9 patients with non-¹³¹I-avid lung metastasis) were enrolled in the training set, and another 41 cases (27 patients with ¹³¹I-avid lung metastasis, and 14 patients with non-¹³¹I-avid uptake lung metastasis) were incorporated for the blind test set. The serum protein was profiled using surface enhanced laser desorption/ionization-time of flight-mass spectrometry (SELDI-TOF-MS), and for each pair of serum samples, the difference of protein fingerprint between the two groups were compared using the Ciphergen Proteinchip 3.1 software. Bioinformatics analysis was performed to construct the decision tree model based on the data of the training set, and the blind test was also conducted for the blind test set. Results The first pair: 69 valid protein peaks were detected at the molecular range of 1500Da-20000Da, among which ten were significantly different between PTC patients with distant metastasis and those without distant metastasis (P<0.05). Among the ten peaks, one peak (M/Z=11403.4Da) displayed high expression in the PTC patients with distant metastasis, and nine peaks showed high expression in those without distant metastasis. The blind test was conducted via the decision tree model, with a sensitivity of 80% and a specificity of 80%. The second pair: 151 valid protein peaks were detected at the molecular range of 1300Da-15000Da, among which seven were significantly different between PTC patients with ¹³¹I-avid uptake lung metastasis and those with non-¹³¹I-avid lung metastasis (P<0.05). Among the seven peaks, five displayed high expression in the PTC patients with ¹³¹I-avid uptake lung metastasis, and two peaks (M/Z=3889.8Da and M/Z=4193.8Da) showed high expression in those with non-¹³¹I-avid lung metastasis. The blind test was conducted via the decision tree model, with a sensitivity of 92.6% and a specificity of 85.7 %.Conclusion The differences in the serum protein fingerprint between PTC patients with distant metastasis and those without distant metastasis and between PTC patients with ¹³¹I-avid uptake lung metastasis and those with non-¹³¹I-avid lung metastasis can be detected respectively using SELDI-TOF-MS. The decision tree models (screening models) based on the data of the training sets can make accurate judgments and predictions of PTC with or without distant metastasis and PTC with or without ¹³¹I uptake in lung metastasis. This study provides a new method for predicting clinical biological behavior of PTC and has potential applications of making clinical treatment decisions for PTC.