Identification of characteristic metabolic panels for different stages of prostate cancer by ¹H NMR-based metabolomics analysis

Prostate cancer (PCa) is the second most frequent cancer among men globally, which poses a major detriment to men’s health [1]. There were an estimated 1.4 million new cases of prostate cancer, and almost 375,304 cancer deaths occurred in 2020 [2]. Several significant efforts have been focused on the development of sensitive and accurate diagnostic tools for PCa since it is more likely to be cured if it is diagnosed early [3]. Currently, PCa is majorly screened by the prostate-specific antigen (PSA) blood test or a digital rectal examination (DRE) combined with a subsequent ultrasound-guided prostate biopsy (PBs) that confirms the cancerous growth presence [4]. Several limitations still exist in the traditional diagnostic methods like, higher procedural costs, longer examination time, low sensitivity and specificity leading to false positive findings, which sometimes may lead to overdiagnosis and consequent overtreatment [5, 6]. Therefore, there is an urgent need to discover novel biomarkers for improving the diagnosis, prognosis and treatment of PCa.

As dysregulated cellular metabolism is a vital hallmark of cancer [7], employing metabolomics analysis in cancer samples could provide critical insights for monitoring the cancer progression [8, 9]. Recently, multiple studies have demonstrated the potential of metabolomics analysis in PCa research. For example, Lima et al. [10] analyzed metabolomic profiling of volatile metabolites in urine samples from PCa patients and healthy controls using a metabolomics approach and found that urinary signatures of volatile organic compounds and volatile carbonyl compounds allowed for accurate discrimination between PCa and control groups. Based on the liquid chromatography-mass spectrometry (LC–MS) and chromatography-mass spectrometry (GC–MS), Huang et al. [11] suggested that serum N-oleoyl taurine and sterol metabolites levels were linked to a decreased PCa survival rate. Additionally, a serum metabolomics analysis integrated with GC–MS and magnetic resonance spectroscopy (MRS) separated benign prostatic hyperplasia (BPH) from PCa subjects, two common conditions that gave rise to increased PSA levels, and identified acylcarnitines, glycerophospholipids, and arginine that could be used as potential diagnostic markers for separating BPH from PCa [12]. Hence, this concept can be put forth that metabolomics analysis has a strong potential as a precise diagnostic tool for exploring metabolic biomarker’s expression patterns and elucidating the potential PCa metabolic mechanisms [13].

Nuclear magnetic resonance (NMR) spectroscopy is a promising noninvasive technique for metabolic profiling analysis due to innate advantages like simple sample preparation, non-destructive analysis, and high reproducibility [14]. In our previous study, ¹H-NMR based metabolomics approach was successfully applied to characterize the significant metabolic differences in tissue and biofluid samples in five different PCa stages [15] as well as prostate tissue samples in hormone-sensitive and castration-resistant PCa cases [16]. In the previous study, we identified a series of metabolic disturbances, including energy, amino acid, choline, fatty acid, and uridine metabolism in five different PCa stages [15], and amino acid, choline metabolism, and Warburg effect from hormone-sensitive to castration-resistant PCa [16]. However, little information is available about tumor progression and metastasis. The present study characterized the metabolic profiling in serum samples from PCa, PCa patients with bone metastasis (PCaB), and BPH subjects via a ¹H NMR-based metabolomics approach. The aim of this study was (1) to identify metabolic alterations among PCa, PCaB, and BPH, (2) to investigate key metabolic pathways involved in the PCa and PCaB progression, (3) to elucidate the potential biomarker panels for differentiating among BPH, PCa and PCaB subjects.

Early diagnosis plays a crucial role in the successful treatment for PCa [21]; however, the non-selective use of traditional screening tools for PCa generally leads to overdiagnosis and overtreatment that does not give expected results [22]. Several studies have reported that metabolic disturbances have been associated with increased PCa incidence [23]. Nowadays, metabolomics has emerged as an immensely powerful screening tool in PCa biomarker development due to its inherent advantages, such as noninvasive procedures, high reproducibility, and lower costs [24]. This study examined the metabolic profiles of BPH, PCa, and PCaB serum samples obtained by ¹H NMR-based metabolomics approach and revealed that serum metabolome analysis exhibited clear discriminations among the three subject groups: BPH, PCa, and PCaB. After conducting univariate statistical and VIP analyses, the most important metabolites were selected, and the diagnostic capacity of these metabolites was evaluated by ROC analysis, while the potential biomarker panels were identified in serum samples for segregating BPH, PCa, and PCaB subjects. Accordingly, the present study identified valuable metabolic panels based on serum metabolome, which exhibited superior performance in discriminating the three groups. A combination of variable metabolites from MRS and MS analysis [12] showed a good classification capacity for discriminating BPH from PCa, with 73.7% of sensitivity, 69% specificity, and 54.1% and 76.3% sensitivity and specificity, respectively. In the present study, significant metabolites in combination for discriminating BPH from PCa patients, had a 0.815 AUC value, along with 75.81% and 72% sensitivity and specificity, respectively. According to Zhang et al. [25], the prognostic value of serum sialic acid levels in predicting PCa and PCaB achieved an AUC value of 0.57, the sensitivity of 60% and the specificity of 58.6%. By combining variable metabolites found in the present study, PCa and PCaB patients were separated with a AUC value of 0.828, a sensitivity and specificity of 78.18% and 74.19%, respectively. Especially, the metabolic biomarker panels in the cluster of differential metabolites and clinical characteristics, showing the AUC values > 0.8, suggesting a relatively good clinical value for diagnosing PCa and its metastases. Based on these, our study imply that serum metabolome based on ¹H NMR-based metabolomics approach has a great potential as a diagnostic supplemental tool for PCa and its metastases. In our study, it was observed that among all identified potential metabolic biomarkers for different PCa stages, a series of metabolic pathways were mainly involved, including energy, amino acid, and ketone body metabolism. Of note, a series of characteristic metabolic changes were identified, including decreased trends of 3-HB and acetone as well as elevated trends of alanine in PCa patients compared with BPH subjects, while increased levels of 3-HB and acetone as well as decreased levels of alanine in PCaB patients compared with PCa.

Energy metabolism, which expedites the uptake and incorporation of glucose into the biomass needed to produce new cells, is critical for maintaining the abnormal growth and intensive proliferation of cancer cells [26]. Furthermore, the glycolytic breakdown of glucose, the main substrate for energy supply, results in pyruvate production that consequently gets oxidized through the tricarboxylic acid (TCA) cycle for ATP production or converted into lactate by anaerobic glycolysis [27]. According to the Warburg effect [28], most cancer cells primarily convert glucose to lactate through anaerobic glycolysis for meeting the energy requirements for cellular cell proliferation. However, our study revealed that the pyruvate concentration in serum samples was significantly enhanced in PCaB cases compared to PCa and BPH, which might further imply that PCaB may have a reverse Warburg effect in serum metabolome that still needs validation. Citrate is not only an important substrate for de novo lipid synthesis but also serves as a key TCA cycle intermediate for ATP production [29, 30]. In the present study, the citrate level of PCaB serum samples was higher than the PCa samples, which is in line with our previous study [15], that increased TCA cycle intermediates might play a pivotal role in malignant tumor cell growth and proliferation. Additionally, the concentration of creatine, an important regulator of energy metabolism, was remarkably decreased in the PCaB serum levels when compared to PCa and BPH subjects. Given high energy demand of tumor cells [31], a decrease in serum creatine may attributed to providing fuels to meet energy production. Overall, our results implies that a distinct energy metabolism occurred in tumor progression and metastasis.

Out of the many energy sources, amino acids are precisely involved in biosynthesis as well as are important reserves for supporting the survival and proliferation of cancer cells [32]. A recent study investigated the emerging roles of amino acids in epigenetic regulation and initiating immune responses that were related to tumorigenesis and metastatic pathways [33]. Our study noted that a majority of amino acids were increased in the serum samples from BPH to PCa subjects but gradually decreased when the disease progressed to PCaB. It was suggested that leucine, isoleucine, and valine are members of the branched-chain amino acids (BCAAs) that can be catabolized to TCA cycle intermediates for energy production [34], which was also supported by Teahan et al. [35] that BCAAs are potential biomarkers for PCa aggression by using NMR-based metabolomics. Hence, it was evident that increased levels of serum BCAAs are utilized as an energy source for tumor proliferation and development. In contrast, it was also observed that the serum BCAAs levels were significantly decreasing in PCaB when compared with PCa or BPH patients, thereby suggesting that downregulated BCAAs might be closely related to bone metastasis in PCa progression. Recent evidence indicated that histidine catabolism was associated with PCa progression [15] and was consistent with our findings that the histidine level was markedly decreased in the serum levels of both PCa and PCaB patients as compared to BPH patients. As proposed by Lapek et al. in 2015 [36] that histidine phosphorylation might be greatly associated with metastatic PCa. Hence, this concept could be put forth that the downregulation of PCa and PCaB histidine levels, when compared to BPH patients, might be attributed to the upregulated histidine phosphorylation in PCa and the resultant metastasis development. Several previous studies have indicated that alanine level was significantly higher in serum and biopsy tissues when progressing from BPH to PCa [37, 38]. In contrast, a decreased alanine level was associated with advanced cancer stage and poor cancer-specific survival [39]. Accordingly, our study denoted that due to alanine’s differential nature, it was observed that tumor proliferation was consistent with increased protein synthesis; as a result, the alanine serum level was distinctly elevated from BPH to PCa but decreased from PCa to PCaB subjects. Correspondingly, tyrosine and phenylalanine were two additional metabolites that exhibited increased trends in the serum of PCa and PCaB patients. Several studies have demonstrated that dysregulated tyrosine and phenylalanine metabolism is closely related to PCa progression [40, 41]. Another observation in our study was the presence of higher levels of tyrosine and phenylalanine in PCa and PCaB patients’ serum when compared with BPH patients, respectively. It was substantiated by Gomez-Cebrian et al. [42] that phenylalanine hydroxylase (PAH) is the enzyme that metabolizes excess phenylalanine into tyrosine; hence, it is also reported to have a direct association with protein acetylation and energy production [41, 43]. Moreover, our study revealed that a decreased PAH expression might be directly proportional to the enhanced levels of both phenylalanine and tyrosine in PCa and its metastatic progression.

Ketone bodies (3-HB, acetoacetate, and acetone) are alternative mitochondrial energy reserves that can be converted into acetyl-CoA and reutilized as energy substrates[44, 45]. Several studies have stated that ketone body metabolism is critical for tumor biomass expansion [46], as well as the fact that ketogenesis plays an important role in PCa progression [47]. A study by Martinez et al. [48] generated a series of cells and fibroblasts overexpressing the enzyme initiating ketone body production and suggested that the production and reutilization of ketone bodies promote tumor progression and metastasis. A prominent ketone body, 3-HB, has been proved to be specifically associated with metastatic prostate cancer [49]. The current study displayed significant reductions in 3-HB and acetone in PCa cases relative to BPH, but significant elevations were duly observed in PCaB subjects when compared to PCa cases. Collectively, the characteristic changes in 3-HB and acetone levels might be applicable for the potential detection of the PCa progressive stages. However, it was observed that the acetoacetate concentration was significantly decreased in PCaB as compared to BPH patients, which may be due to the uptake of ketone bodies from the tumor tissues in PCa proliferation and development.

Formate is a member of short-chain fatty acids and can be remodeled back for re-synthesizing serine via a one-carbon metabolism pathway [50]. Additionally, formate is an extremely vital component of one-carbon metabolism for tumor proliferation and growth [51, 52]. A study by Meiser et al. [53] revealed that mice with oxidative cancers have higher circulating formate levels than the healthy controls, thereby proposed that elevated formate overflow is a hallmark of oxidative cancers. Our study results also observed an increased formate level in the PCaB patients' serum when compared to BPH, which was also confirmed by our previous study that depicted a higher formate level in the metastatic PCa tissue when compared to other PCa stages [15]. Thus, our results might imply that upregulated formate metabolism mainly occurred in the tumor tissue and serum samples of the tumor in its metastatic stage.

LDL/VLDL is an essential member of lipoproteins, fundamental to the reverse cholesterol transport pathway and lipid homeostasis [54, 55]. Recent studies have demonstrated that lipoprotein might be considered as a risk factor for PCa [56, 57]. However, several literary insights display contradictory findings regarding the association between lipid metabolism and PCa progression and development. A study by Bull et al. [58] suggested there is a weaker association between higher lipoprotein levels and aggressive PCa risk, which was also substantiated by a meta-analysis of 14 large prospective studies that indicated an absence of association between the blood lipoprotein levels and overall PCa or high-grade PCa risk cases [59]. Our results revealed that PCaB cases had lower circulating LDL/VLDL levels than PCa, thus, suggesting that PCaB may have a disruptive lipid metabolism relative to PCa. Moreover, the lipid metabolism inconsistency along with several metabolic aberrations in PCa development could contribute to the initiation of several more studies focused on the extremely significant role of lipid metabolism in PCa progression.

In this study, BPH patients were enrolled as a control group, which is a typical clinical setting. The healthy control donors usually had lower PSA levels and may have a small chance of holding undiagnosed PCa [12]. However, BPH patients with significantly higher levels of PSA are very common among elderly and middle-aged men, which frequently leads to the lack of specificity in PSA measurements. Besides, understanding the metabolic profiles among BPH, PCa and PCaB offers vital information about metabolic biomarkers for PCa progression and its metastases since PSA measurements are not optimal. Moreover, the appropriate use of metabolic biomarkers can provide a more comprehensive evidence for the early screening and diagnosis to predict the corresponding PCa progression and make individualized treatment options. Thus, we believe this is a highly reasonable and relevant clinical setting for diagnosing and classifying of PCa. However, there are several potential limitations in this study: (1) As the majority of BPH patients did not undergo biopsy, there was no relationship analysis between differential metabolic changes and the subjects pathology in this study; but the pathologic analysis of subjects is recommended for elucidating the link of metabolic biomarkers in different PCa pathologies. (2) Although the potential metabolic biomarkers were identified in serum samples, this finding still needs validation by using other biofluids and matching tissues. (3) As our study was a prospective, single-center study with a smaller sample size, large clinical multi-center cohorts would be required for verifying the clinical potential of our findings. (4) Since ¹H NMR-based metabolomics analysis is generally limited to small metabolites at higher concentrations, multi-omics analysis integrating several useful genes, proteins, and metabolome are recommended for elucidating potential mechanisms underlying the PCa progression and its metastases.

To conclude, ¹H NMR-based metabolomics analysis of serum metabolic profiles in BPH, PCa, and PCaB patients can be successfully utilized to establish the significant metabolic signatures related to PCa progression and its metastases. The identified metabolic panels in our study might provide crucial insights into the diagnosis and classification of PCa.