Limits...
Three serum metabolite signatures for diagnosing low-grade and high-grade bladder cancer

View Article: PubMed Central - PubMed

ABSTRACT

To address the shortcomings of cystoscopy and urine cytology for detecting and grading bladder cancer (BC), ultrahigh performance liquid chromatography (UHPLC) coupled with Q-TOF mass spectrometry in conjunction with univariate and multivariate statistical analyses was employed as an alternative method for the diagnosis of BC. A series of differential serum metabolites were further identified for low-grade(LG) and high-grade(HG) BC patients, suggesting metabolic dysfunction in malignant proliferation, immune escape, differentiation, apoptosis and invasion of cancer cells in BC patients. In total, three serum metabolites including inosine, acetyl-N-formyl-5-methoxykynurenamine and PS(O-18:0/0:0) were selected by binary logistic regression analysis, and receiver operating characteristic (ROC) test based on their combined use for HG BC showed that the area under the curve (AUC) was 0.961 in the discovery set and 0.950 in the validation set when compared to LG BC. Likewise, this composite biomarker panel can also differentiate LG BC from healthy controls with the AUC of 0.993 and 0.991 in the discovery and validation set, respectively. This finding suggested that this composite serum metabolite signature was a promising and less invasive classifier for probing and grading BC, which deserved to be further investigated in larger samples.

No MeSH data available.


Related in: MedlinePlus

Box plots of serum inosine, AFMK and PS(O-18:0/0:0), ROC curves based on the binary logistic regression model by the combination of three serum metabolites from the HG and LG BC dataset and the prediction plots according to the optimal cutoff value obtained from ROC curves.(A) Box plots of the relative intensities of the three metabolites in HC, LG BC and HG BC. (B) The ROC curves of the discovery set (B, left) and validation set (B, right) were obtained from the established prediction model. (C) The optimal cutoff value was obtained (0.4669) and applied to evaluate the prediction capacity (89.7% for discovery set (C, left) and 84.6% for validation set (C, right)) of the current model, where 0 and 1 on the x axis represent LG BC and HG BC patients, respectively, and blue circle represent samples.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC5382774&req=5

f3: Box plots of serum inosine, AFMK and PS(O-18:0/0:0), ROC curves based on the binary logistic regression model by the combination of three serum metabolites from the HG and LG BC dataset and the prediction plots according to the optimal cutoff value obtained from ROC curves.(A) Box plots of the relative intensities of the three metabolites in HC, LG BC and HG BC. (B) The ROC curves of the discovery set (B, left) and validation set (B, right) were obtained from the established prediction model. (C) The optimal cutoff value was obtained (0.4669) and applied to evaluate the prediction capacity (89.7% for discovery set (C, left) and 84.6% for validation set (C, right)) of the current model, where 0 and 1 on the x axis represent LG BC and HG BC patients, respectively, and blue circle represent samples.

Mentions: It was no doubt that a panel of biomarkers including these 13 serum metabolites will have more power to diagnose. However, diagnosis based on quantification of so many metabolites would not be convenient and economical in clinical practice. With bladder cancer progression (from healthy subjects to LG BC patients and then to HG BC patients), eleven metabolites demonstrated progressive increase trend (Fig. 2) and thus were further used as candidates. They could be assigned to different functions, such as proliferation, immune escape, differentiation, apoptosis and invasion, which will be discussed below. To identify a simplified serum metabolite signature that would be more practical in diagnosing HG BC, a binary logistic regression model with a stepwise optimization algorithm involving the 11 differential metabolites were subjected to variable selection based on the training set. As a result, three metabolites including inosine, AFMK and PS(O-18:0/0:0) were selected to establish a binary logistic regression model on the discovery set. The relative concentrations of these three serum metabolite biomarkers among the three groups are presented in Fig. 3A. The prediction model is as follows:


Three serum metabolite signatures for diagnosing low-grade and high-grade bladder cancer
Box plots of serum inosine, AFMK and PS(O-18:0/0:0), ROC curves based on the binary logistic regression model by the combination of three serum metabolites from the HG and LG BC dataset and the prediction plots according to the optimal cutoff value obtained from ROC curves.(A) Box plots of the relative intensities of the three metabolites in HC, LG BC and HG BC. (B) The ROC curves of the discovery set (B, left) and validation set (B, right) were obtained from the established prediction model. (C) The optimal cutoff value was obtained (0.4669) and applied to evaluate the prediction capacity (89.7% for discovery set (C, left) and 84.6% for validation set (C, right)) of the current model, where 0 and 1 on the x axis represent LG BC and HG BC patients, respectively, and blue circle represent samples.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC5382774&req=5

f3: Box plots of serum inosine, AFMK and PS(O-18:0/0:0), ROC curves based on the binary logistic regression model by the combination of three serum metabolites from the HG and LG BC dataset and the prediction plots according to the optimal cutoff value obtained from ROC curves.(A) Box plots of the relative intensities of the three metabolites in HC, LG BC and HG BC. (B) The ROC curves of the discovery set (B, left) and validation set (B, right) were obtained from the established prediction model. (C) The optimal cutoff value was obtained (0.4669) and applied to evaluate the prediction capacity (89.7% for discovery set (C, left) and 84.6% for validation set (C, right)) of the current model, where 0 and 1 on the x axis represent LG BC and HG BC patients, respectively, and blue circle represent samples.
Mentions: It was no doubt that a panel of biomarkers including these 13 serum metabolites will have more power to diagnose. However, diagnosis based on quantification of so many metabolites would not be convenient and economical in clinical practice. With bladder cancer progression (from healthy subjects to LG BC patients and then to HG BC patients), eleven metabolites demonstrated progressive increase trend (Fig. 2) and thus were further used as candidates. They could be assigned to different functions, such as proliferation, immune escape, differentiation, apoptosis and invasion, which will be discussed below. To identify a simplified serum metabolite signature that would be more practical in diagnosing HG BC, a binary logistic regression model with a stepwise optimization algorithm involving the 11 differential metabolites were subjected to variable selection based on the training set. As a result, three metabolites including inosine, AFMK and PS(O-18:0/0:0) were selected to establish a binary logistic regression model on the discovery set. The relative concentrations of these three serum metabolite biomarkers among the three groups are presented in Fig. 3A. The prediction model is as follows:

View Article: PubMed Central - PubMed

ABSTRACT

To address the shortcomings of cystoscopy and urine cytology for detecting and grading bladder cancer (BC), ultrahigh performance liquid chromatography (UHPLC) coupled with Q-TOF mass spectrometry in conjunction with univariate and multivariate statistical analyses was employed as an alternative method for the diagnosis of BC. A series of differential serum metabolites were further identified for low-grade(LG) and high-grade(HG) BC patients, suggesting metabolic dysfunction in malignant proliferation, immune escape, differentiation, apoptosis and invasion of cancer cells in BC patients. In total, three serum metabolites including inosine, acetyl-N-formyl-5-methoxykynurenamine and PS(O-18:0/0:0) were selected by binary logistic regression analysis, and receiver operating characteristic (ROC) test based on their combined use for HG BC showed that the area under the curve (AUC) was 0.961 in the discovery set and 0.950 in the validation set when compared to LG BC. Likewise, this composite biomarker panel can also differentiate LG BC from healthy controls with the AUC of 0.993 and 0.991 in the discovery and validation set, respectively. This finding suggested that this composite serum metabolite signature was a promising and less invasive classifier for probing and grading BC, which deserved to be further investigated in larger samples.

No MeSH data available.


Related in: MedlinePlus