| Part I Early diagnosis of lung cancer based on the analysis of Multi-serum-biomarker panelsObjectives:Early diagnosis of lung cancer has always been the focus of clinical problems. Multi-biomarker panels are important to aid in the early diagnosis and ultimately battle complex diseases. Here, we aim to identify noninvasive serum-based multi-biomarkers to distinguish lung cancer from benign lung diseases.Patients and methods:(1)We developed a novel multi-biomarker detection method based on a mixed integer programming. We measured serum profiling data for405patients with pulmonary nodules and analyzed37clinical features individually and further their combinations.(2)Serum CA19-9, CA125, F-PSA,T-PSA, NSE, AFP, CEA, CA724, CYFRA21-1, PG I, PG II and pro-GRP were detected using Multiple Tumor Markers ("Male Chip"12) Quantitative Test Kit (Microarray Chemiluminescent Immunoassay) and single tumor marker test from Abbott or Roche diagnostics in1865male patients. Serum CA19-9, CA125, CA15-3, NSE, AFP, CEA, CA724and CYFRA21-1, PG Ⅰ, PG Ⅱ, beta HCG ("Female Chip"12) Quantitative Test Kit (Microarray Chemiluminescent Immunoassay) and single tumor marker test from Abbott or Roche diagnostics in1932female patients.Results:(1) Based on a mixed integer programming and leave-one-out cross-validation method, we discovered four features as the optimal small multi-biomarker panel including serum Glucose, Total Bile Acid, CEA and CYFRA21-1.The panel could detected lung cancer patients with high accuracy-98.65%.(2)The testing results of Microarray Chemiluminescent Immunoassay Chip and the traditional methods are with high consistency. Microarray Chemiluminescent Immunoassay can discriminate malignant tumor from benign diseases with high sensitivity and speciality in man and women patients.Conclusion:(1)A novel multi-biomarker panels including serum Glucose, Total Bile Acid, CEA and CYFRA21-1can be used for early diagnosis of lung cancer with high accuracy and good performance.(2) Microarray Chemiluminescent Immunoassay Chip shows potential diagnostic value of discriminating malignant tumor from benign diseases with high sensitivity and speciality. Part II Joint models based on multi-serum-parameters in evaluating the diagnosis and prognosis of lung adenocarcinoma patientsObjectives:The incidence of lung cancer increased in recent years, and the incidence of lung adenocarcinoma grows rapidly than that of lung squamous carcinoma.Lung adenocarcinoma patients has become the most common type of lung cancer, and the prognosis of patients with advanced lung adenocarcinoma is usually poor. In this section we will further investigate the diagnostic and prognostic value of serum biomarkers including conventional serological indexes, serum leptin and serum peptides in patients with lung adenocarcinoma patients. But it’s difficult to make a more accurate judgment to patients with a single marker. In this chapter we will evaluate the combination value of these serum biomarkers for diagnosis and prognosis evaluation.Methods:(1)227cases of lung adenocarcinoma samples were randomly divided into five subgroups, each selected four as the training and fit the Cox proportional hazards regression model using Akaike information criterion to gradually eliminate redundant variables. The variables with a significant Wald test (P<0.05) are selected, then remaing one test was used to fit the Cox equation. The above cross validation process were repeated1000times. Final evaluation index to measure the effectiveness of the model5000times on the test set of average p value of the log-rank test.(2) Baseline serum leptin levels were determined using immunoradioassay in27lung adenocarcinoma patients treated with pemetrexed/cisplatin chemotherapy. Serum tumor markers including CEA, AFP, CA125, CA199, CA153, CA724, CYFRA21-1,NSE and SCC were also determined.(3) Serum peptides of lung adenocarcinoma patients were screened using matrix assisted laser desorption ionization time of flight mass spectrometry(MALDI-TOF MS). Diagnostic and prognostic models for lung adenocarcinoma patients with brain metastasis were also established.Results:(1)Based on1000times by5fold cross-validation method computation, Prognostic value of43clinical and serological parameters were evaluated by5fold cross-validation computation in227lung adenocarcinoma cases. Cox models for patients are as follows:(a) Variates in the multivariate Cox PH regression models of53lung adenocarcinoma patients without metastasis: TP (HR=4.819, p=0.001), age (HR=0.554, p=0.036), TG (HR＝1.527, p=0.005), albumin (HR=0.337, p=0.012), gender (HR=0.089, p<0.001), UA (HR=0.254, p< 0.001), CYFRA21-1(HR=1.295, p=0.031), CA199(HR=0.638, p=0.014);(b) Variates in the multivariate Cox PH regression models in73lung adenocarcinoma patients with one-site metastasis:DB (HR=0.584, p=0.015), age (HR=0.722, p=0.011), NSE (HR=0.596, p=0.027), CA199(HR=0.827, p=0.035);(c) Variates in the multivariate Cox PH regression models in55lung adenocarcinoma patients with two-sites metastasis:CK (HR=0.526, p=0.011), NSE(HR=1.548, p=0.013), TG (HR=110000, p=0.010), UA (HR=0.575, p=0.006), CA153(HR=0.411, p=0.025);(d) I Variates in the multivariate Cox PH regression models in46lung adenocarcinoma patients without metastasis:TB (HR=3.193, p=0.030), DB (HR=0.085, p=0.009), CK (HR=2.624, p=0.018), smoking index (HR=1.856, p=0.051), NSE (HR=0.208, p=0.004), LDH (HR=0.324, p<0.001), CA153(HR=1.427, p=0.001), CA125(HR=25.737, p<0.001), CA199(HR=12.502, p=0.015).(2)Combination use of serum leptin, CYFRA211and SCC can be used for the diagnosis of advanced lung adenocarcinoma group, and the accuracy of our model is78.45%, the sensitivity and specificity were80.0%and76.9%respectively. Univariate and multivariate analysis showed that serum leptin is of prognostic value for advanced lung adenocarcinoma patients treated with chemotherapy with equation p=(1+e (0.002×leptin)]-1.(3)10serum peptides specific to lung adenocarcinoma were found.(a)Models including peptides m/z1969.9and2213.4Da can be used to discrminating lung adenocarcinoma patients from healthy individuals, and the Logistic regression equation is as follows:p=(1+e (0.006· mass2213.4mass1969.90.002+6.039)]-1.The accuracy of the models was85.45%and the sensitivity and specificity were79.2%and91.7%respectively.(b) Models including peptides m/z1781.8and1781.8Da is can be used to discrminating patients with brain metastasis from patients without brain metastasis, and the Logistic regression equation is as follows:p=(1+e (0.003*mass19847mass1781.8+0.008+0.008))]-1,the accuracy of the models was82.0%, and the sensitivity and specificity were84.6%and79.2%respectively.(c)Univariate and multivariate analysis showed that peptides m/z1969.9is an independent prognostic factors influencing the prognosis of lung adenocarcinoma treated with cisplatin-pemetrexed chemotherapy, and peptides m/z1969.9is is an independent prognostic factors influencing the prognosis of lung adenocarcinoma with brain metastasis.Conclusion:(1)Multi-biomarker panels is established and optimized from43serological indexes characteristics, which facilitate the The choice of chemotherapy regimens;(2) Combination of the serum leptin, CYFRA211and SCC can be used to discriminate lung adenocarcinoma patients from healthy individuals;(3)Peptides m/z1969.9and2213.4Da can be used used to discriminate lung adenocarcinoma patients from healthy individuals.Peptides m/z1781.8and1781.8Da can be used used to investigate the status of brain metastases. Univariate and multivariate analysis showed that peptides m/z1969.9is an independent factor for lung adenocarcinoma treated with cisplatin/pemetrexed chemotherapy, peptides m/z1969.9is an independent factor for lung adenocarcinoma patients with brain metastases. |
PDF Full Text Request