Limits...
Novel Use of Proteomic Profiles in a Convex-Hull Ensemble Classifier to Predict Gynecological Cancer Patients' Susceptibility to Gastrointestinal Mucositis as Side Effect of Radiation Therapy.

Kodell RL, Haun RS, Siegel ER, Zhang C, Trammel AB, Hauer-Jensen M, Burnett AF - J Proteomics Bioinform (2015)

Bottom Line: We employ 20 repetitions of 10-fold cross-validation to measure classification accuracy.Pathway analysis of the 12 most prominent proteins indicated that they could be assembled into a single interaction network with direct associations.The function associated with the highest number of these 12 proteins was cell-to-cell signaling and interaction.

View Article: PubMed Central - HTML - PubMed

Affiliation: Department of Biostatistics, University of Arkansas for Medical Sciences, Little Rock, AR 72205, USA.

ABSTRACT

Background: Whole-pelvis radiation therapy is common practice in the post-surgical treatment of cervical and endometrial cancer. Gastrointestinal mucositis is an adverse side effect of radiation therapy, and is a primary concern in patient management. We investigate whether proteomic information obtained from blood samples drawn from patients scheduled to receive radiation therapy for gynecological cancers could be used to predict which patients are most susceptible to radiation-induced gastrointestinal mucositis, in order to improve the individualization of radiation therapy.

Methods: We use 132 proteins measured on 17 gynecological cancer patients in a convex-hull-based, selective-voting ensemble classifier to classify each patient into one of two classes: patients who would not (class 1) or would (class 2) develop gastrointestinal mucositis. We employ 20 repetitions of 10-fold cross-validation to measure classification accuracy.

Results: We achieved a 95% confidence interval on average prediction accuracy of (0.711, 0.771) using pre-radiation proteomic profiles to predict which patients would experience gastrointestinal mucositis. Pathway analysis of the 12 most prominent proteins indicated that they could be assembled into a single interaction network with direct associations. The function associated with the highest number of these 12 proteins was cell-to-cell signaling and interaction.

Conclusions: Pre-radiation proteomic profiles have the potential to classify cervical/endometrial cancer patients with high accuracy as to their susceptibility to gastrointestinal mucositis following radiation therapy. Further study of the network of 12 identified proteins is warranted with a larger patient sample to confirm that these proteins are predictive of gastrointestinal mucositis in this patient population.

No MeSH data available.


Related in: MedlinePlus

Frequencies (> 50) of occurrence of individual proteins in the set of 2000 best pairs from 20 runs of 10-fold cross-validation (100 per run). (Protein names in Table 3).
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4587761&req=5

Figure 1: Frequencies (> 50) of occurrence of individual proteins in the set of 2000 best pairs from 20 runs of 10-fold cross-validation (100 per run). (Protein names in Table 3).

Mentions: From the CV results for the 132 proteins as predictors along with age and BMI (summarized in Table 2), the predictors occurring most prominently across the twenty runs were examined. Table 3 and Figure 1 identify the individual proteins and Figure 2 gives the protein pairs that were most prominent in the prediction of GM. In Table 3, the corresponding UniProt accession number and gene ID are given along with each protein name, as well as the mean and standard deviation of spectral counts for each of the two classes. In the figures and in our discussion, we refer to specific proteins via the gene ID. The ten protein-pairs with the largest R2 values in the regression phase of the selective-voting algorithm [2] in each block of each run of 10-fold CV (2000 in all) were first identified. Then the frequencies of occurrence of each individual protein and each protein-pair among those 2000 pairs were tabulated; proteins and protein-pairs with frequencies exceeding 50 were retained for Table 3 and the figures. Table 3 shows that for each of the top twelve proteins, spectral counts were higher on average in class 2 than in class 1, with some proteins showing high variability among subjects. As Figures 1 and 2 show, LPA (apolipoprotein (a)) and ITIH2 (inter-alpha-trypsin inhibitor heavy chain H2) were the two most prominent proteins, both individually and in pairs. Among the other ten prominent proteins, a high proportion of their individual frequencies can be explained by their pairing with either LPA or ITIH2, except for F10 (coagulation factor X) and CLEC3B (Tetranectin). LPA is involved in proteolysis, transport, lipoprotein metabolic processes, blood circulation, apolipoprotein binding, hydrolase activity, and endopeptidase inhibitor activity. ITIH2 is involved in negative regulation of peptidase activity, hyaluronan metabolic process, and endopeptidase inhibitor activity. Neither age nor BMI appeared among the most prominent predictors, in keeping with the lack of improvement in accuracy observed when they were added to the set of proteins (compare Tables 1 and 2). Pathway analysis applied to the 12 most frequently occurring proteins in the GM prediction model (Table 3) revealed an interaction network (Figure 3) with hematological system development and function, cell-to-cell signaling and interaction, and organismal injury and abnormalities as the top associated network functions.


Novel Use of Proteomic Profiles in a Convex-Hull Ensemble Classifier to Predict Gynecological Cancer Patients' Susceptibility to Gastrointestinal Mucositis as Side Effect of Radiation Therapy.

Kodell RL, Haun RS, Siegel ER, Zhang C, Trammel AB, Hauer-Jensen M, Burnett AF - J Proteomics Bioinform (2015)

Frequencies (> 50) of occurrence of individual proteins in the set of 2000 best pairs from 20 runs of 10-fold cross-validation (100 per run). (Protein names in Table 3).
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4587761&req=5

Figure 1: Frequencies (> 50) of occurrence of individual proteins in the set of 2000 best pairs from 20 runs of 10-fold cross-validation (100 per run). (Protein names in Table 3).
Mentions: From the CV results for the 132 proteins as predictors along with age and BMI (summarized in Table 2), the predictors occurring most prominently across the twenty runs were examined. Table 3 and Figure 1 identify the individual proteins and Figure 2 gives the protein pairs that were most prominent in the prediction of GM. In Table 3, the corresponding UniProt accession number and gene ID are given along with each protein name, as well as the mean and standard deviation of spectral counts for each of the two classes. In the figures and in our discussion, we refer to specific proteins via the gene ID. The ten protein-pairs with the largest R2 values in the regression phase of the selective-voting algorithm [2] in each block of each run of 10-fold CV (2000 in all) were first identified. Then the frequencies of occurrence of each individual protein and each protein-pair among those 2000 pairs were tabulated; proteins and protein-pairs with frequencies exceeding 50 were retained for Table 3 and the figures. Table 3 shows that for each of the top twelve proteins, spectral counts were higher on average in class 2 than in class 1, with some proteins showing high variability among subjects. As Figures 1 and 2 show, LPA (apolipoprotein (a)) and ITIH2 (inter-alpha-trypsin inhibitor heavy chain H2) were the two most prominent proteins, both individually and in pairs. Among the other ten prominent proteins, a high proportion of their individual frequencies can be explained by their pairing with either LPA or ITIH2, except for F10 (coagulation factor X) and CLEC3B (Tetranectin). LPA is involved in proteolysis, transport, lipoprotein metabolic processes, blood circulation, apolipoprotein binding, hydrolase activity, and endopeptidase inhibitor activity. ITIH2 is involved in negative regulation of peptidase activity, hyaluronan metabolic process, and endopeptidase inhibitor activity. Neither age nor BMI appeared among the most prominent predictors, in keeping with the lack of improvement in accuracy observed when they were added to the set of proteins (compare Tables 1 and 2). Pathway analysis applied to the 12 most frequently occurring proteins in the GM prediction model (Table 3) revealed an interaction network (Figure 3) with hematological system development and function, cell-to-cell signaling and interaction, and organismal injury and abnormalities as the top associated network functions.

Bottom Line: We employ 20 repetitions of 10-fold cross-validation to measure classification accuracy.Pathway analysis of the 12 most prominent proteins indicated that they could be assembled into a single interaction network with direct associations.The function associated with the highest number of these 12 proteins was cell-to-cell signaling and interaction.

View Article: PubMed Central - HTML - PubMed

Affiliation: Department of Biostatistics, University of Arkansas for Medical Sciences, Little Rock, AR 72205, USA.

ABSTRACT

Background: Whole-pelvis radiation therapy is common practice in the post-surgical treatment of cervical and endometrial cancer. Gastrointestinal mucositis is an adverse side effect of radiation therapy, and is a primary concern in patient management. We investigate whether proteomic information obtained from blood samples drawn from patients scheduled to receive radiation therapy for gynecological cancers could be used to predict which patients are most susceptible to radiation-induced gastrointestinal mucositis, in order to improve the individualization of radiation therapy.

Methods: We use 132 proteins measured on 17 gynecological cancer patients in a convex-hull-based, selective-voting ensemble classifier to classify each patient into one of two classes: patients who would not (class 1) or would (class 2) develop gastrointestinal mucositis. We employ 20 repetitions of 10-fold cross-validation to measure classification accuracy.

Results: We achieved a 95% confidence interval on average prediction accuracy of (0.711, 0.771) using pre-radiation proteomic profiles to predict which patients would experience gastrointestinal mucositis. Pathway analysis of the 12 most prominent proteins indicated that they could be assembled into a single interaction network with direct associations. The function associated with the highest number of these 12 proteins was cell-to-cell signaling and interaction.

Conclusions: Pre-radiation proteomic profiles have the potential to classify cervical/endometrial cancer patients with high accuracy as to their susceptibility to gastrointestinal mucositis following radiation therapy. Further study of the network of 12 identified proteins is warranted with a larger patient sample to confirm that these proteins are predictive of gastrointestinal mucositis in this patient population.

No MeSH data available.


Related in: MedlinePlus