80306-38-3 br Table The VOCs Selected by
Table 2 The VOCs Selected by Logistic Regression Models for Prostate Cancer Diagnosis Prediction and Risk Assessment
Abbreviations: CAS ¼ Chemical Abstracts Service number; PCa ¼ prostate cancer; PCa(þ) ¼ prostate cancer-positive patients; PCa( ) ¼ prostate cancer-negative patients.
aP value ¼ the P value of selected compounds from the Wilcoxon rank-sum test.
Qin Gao et al
Figure 2 A, The ROC Curve for VOC PCa Diagnosis Logistic Model Verified in 108 Patients; B, the ROC Curve for Logistic Model With PSA Only
Abbreviations: CI ¼ confidence interval; PCa ¼ prostate cancer; PSA ¼ prostate-specific antigen; ROC ¼ receiver operator characteristic; VOCs ¼ volatile organic compounds.
Manual selection of VOCs likely introduced bias into the model as opposed to allowing the discriminatory analysis to dictate the final VOCs that would come to define each cohort. In our study, there was a blend of VOCs that were increased or reduced in either the PCa-positive or PCa-negative groups. In the Khalid study, the same cohort was used for both discovery and validation. Despite using robust statistical methods to minimize the bias associated with this approach, the sensitivity and specificity of the 4 VOCs remained
Figure 3 The ROC Curve for VOC PCa Diagnosis Logistic Model Tested in 75 Patients
Abbreviations: CI ¼ confidence interval; PCa ¼ prostate cancer; ROC ¼ receiver operator characteristic; VOCs ¼ volatile organic compounds.
poor. In this study, despite using a rural Pennsylvania patient cohort to develop the PCa detection model, the validity of the model was upheld when tested in an urban Texas 80306-38-3 with an AUC of 0.86 with sensitivity and specificity of 0.87 and 0.77, respectively. Several other biomarkers, including Iso-PSA, prostate cancer antigen 3 (PCA3), 4Kscore, Prostate Health Index (PHI), TMPRSS2:ERG, and ConfirmMDx,25-30 were developed as alter-natives to address the challenges of PSA test in PCa diagnosis.
Among those markers, IsoPSA, PHI, and 4Kscore are PSA-based assays for PCa risk assessment.25,28,30 PCA3 is reported to be a noncoding RNA that is prostate-specific and highly overexpressed in PCa31; and TMPRSS2-ERG gene fusion is the predominant molecular subtype of PCa.32 ConfimMDx is an epigenetic test for PCa diagnosis before prostate biopsy.29,33 However, these PCa diagnostic tests are still limited by either their low sensitivity and/or low specificity (Table 3). The metabolome can be considered as the amplified output of the biological system, and therefore, metab-olomics profiling is emerging as a promising strategy for PCa diagnosis.34 In this study, the VOC metabolome-based PCa diag-nosis model outperformed those aforementioned diagnostic PCa biomarkers (Table 3).
Among the 11 compounds selected in the PCa diagnosis model,
5 showed higher levels in the PCa-negative group, whereas the remaining 6 were positively related to PCa. Some of these com-pounds may be directly or indirectly involved in androgen receptor signaling. For example, androgens and androgen receptor in prostate regulate the activity/expression of the enzymes involved in lipo-genesis.35 Lipogenesis is mediated by increased activities of fatty acid biosynthetic enzymes (including ACC1, FASN, and stearoyl CoA desaturase).36-38 The urinary fatty acids and/or fatty acid metabo-lites found in this study could be a reflection of such activities, and therefore relevant in PCa.
Urinary VOCs for Prostate Cancer Diagnosis
fiComparisoninSensitivity,Specicity, and AUC From Various
Sensitivity fiSpecicity AUC Abbreviations:
Ethyl á-hydroxymyristate (Figure 4) was reduced in patients with PCa. Numerous epidemiologic studies suggest that myristic acid significantly increases the risk of prostate cancer-specific mortality among patients diagnosed with localized disease.39,40 In vivo studies in mice showed that myristoylation of Src kinase promoted Src-induced and high-fat diet-accelerated prostate tumor progres-sion.41 The low levels of ethyl á-hydroxymyristate seen in urine samples from patients with PCa could be linked to the consumption of this metabolite during the PCa tumorigenesis process. However, this would need to be validated experimentally. Moreover, ethyl á-hydroxymyristate contains the moiety of 2-hydroxymyristic acid, which was reported to inhibit the myristoylation and alters the stability of p56lckin T cells. p56lck is a protein-tyrosine kinase that is found predominantly in lymphoid cells42 and was reported to be positively expressed in PCa cell lines and tissues43 and in metastatic cancer.44