Carbon-ion Evokes Metabolic Reprogramming and Individualized Response in Prostate Cancer

Background: Carbon ion radiotherapy (CIRT) is a novel and powerful tool for prostate cancer (PCa). However, the underlying mechanism for individualized treatment response after CIRT was not clear, and there was still no effective indicator to timely demonstrate the treatment response. Metabolic reprogramming is one of the main hallmarks of malignancy. Metabolic status might have a high relationship with the radiosensitivity and the individualized radiation response. The signicant changes of metabolites proles were detected after radiotherapy in the serum sample of different malignancies. But there was limited data regarding CIRT induced metabolic changes in prostate cancer. Our aim was to preliminary investigate the carbon-ion induced metabolic reprogramming in PCa patients and the individualized response of PCa patients to carbon ion. Methods: Urine samples collected from 15 pathology conrmed PCa patients before and after CIRT were enrolled into this analysis. High-throughput UPLC-MS/MS system was used for metabolites detection. XCMS online, MetDNA and MSDIAL were used for peak detection and identication of metabolites. Statistical analysis and metabolic pathway analysis were performed on Metaboanalyst. Results: A total of 1701 metabolites were monitored by high-throughput UPLC-MS/MS and 217 metabolites were identied. The PCA scores plot revealed clear discrimination of the patient’s urine metabolites proles before (pre-CIRT) and after (pre-CIRT) CIRT treatment. 35 metabolites signicantly altered after CIRT, and these metabolites mainly were amino acid. Pathway enrichment analysis further identied these metabolites could be enriched in 8 pathways (FDR<0.05, impact>2), while arginine biosynthesis and histidine metabolism pathways were the most signicant. In addition, the HCA shows that after CIRT, the patients can be clustered into two groups according to the metabolites proles. The discriminatory metabolites after CIRT in patients urine mainly enriched in the pathway of arginine biosynthesis and phenylalanine, tyrosine, and tryptophan biosynthesis. Conclusion: Metabolic reprogramming and metabolic inhibition seems of the most mass spectrometry; CTV: Clinical target volume; QC: Quality control; ESI: Electrospray ionization; FC: Fold change; FDR: False discovery rates; PCA: Principal component analysis; PLS-DA: Partial least-squares discriminant analysis; sPLS-DA: Sparse partial least-squares discriminant analysis; HCA: Hierarchical cluster analysis

relationship with the radiosensitivity and the individualized radiation response. The signi cant changes of metabolites pro les were detected after radiotherapy in the serum sample of different malignancies.
But there was limited data regarding CIRT induced metabolic changes in prostate cancer. Our aim was to preliminary investigate the carbon-ion induced metabolic reprogramming in PCa patients and the individualized response of PCa patients to carbon ion.
Methods: Urine samples collected from 15 pathology con rmed PCa patients before and after CIRT were enrolled into this analysis. High-throughput UPLC-MS/MS system was used for metabolites detection. XCMS online, MetDNA and MSDIAL were used for peak detection and identi cation of metabolites.
Statistical analysis and metabolic pathway analysis were performed on Metaboanalyst.
Results: A total of 1701 metabolites were monitored by high-throughput UPLC-MS/MS and 217 metabolites were identi ed. The PCA scores plot revealed clear discrimination of the patient's urine metabolites pro les before (pre-CIRT) and after (pre-CIRT) CIRT treatment. 35 metabolites signi cantly altered after CIRT, and these metabolites mainly were amino acid. Pathway enrichment analysis further identi ed these metabolites could be enriched in 8 pathways (FDR<0.05, impact>2), while arginine biosynthesis and histidine metabolism pathways were the most signi cant. In addition, the HCA shows that after CIRT, the patients can be clustered into two groups according to the metabolites pro les. The discriminatory metabolites after CIRT in patients urine mainly enriched in the pathway of arginine biosynthesis and phenylalanine, tyrosine, and tryptophan biosynthesis.
Conclusion: Metabolic reprogramming and metabolic inhibition seems one of the most important mechanisms of CIRT to cure PCa. Urine metabolites also showed their potentials to timely identify the individualized response of PCa patients to CIRT. Background CIRT is a novel and powerful tool for radiotherapy and has gradually been recognized as one of the best strategies for PCa with the reported excellent ve-year biochemical recurrence-free survival (RFS) and a favorable level of late gastrointestinal and genitourinary toxicities due to its inspiring physical and biological advantages [1,2]. In the past seven years, Shanghai Proton and Heavy Ion Center (SPHIC), as the rst carbon-ion treatment center in China, has treated 162 pathology con rmed PCa patients and the 3 years BFS reached 93% under CIRT. However, these PCa patients showed an individualized treatment response after carbon ion irradiation, and the underlying mechanism was not clear until now. Moreover, there was still no effective indicator to timely demonstrate the treatment response. Patients might wait several months for the serum total PSA and MRI results after the completion of irradiation, which adversely impacts decision-making. Exploring the underlying mechanism and nding timely and powerful tools would add value to early evaluate the prognosis of PCa patients treated with CIRT.
Metabolic reprogramming is one of the main hallmarks of malignancy, in which tumor cells make metabolic disorders to promote their growth and proliferation, as well as other components like microenvironment and immune cell. Metabolic reprogramming was also detected to have the potential to produce an immunosuppressive metabolic microenvironment which is conducive to tumor proliferation and escapes [3]. Data further showed that metabolic status might have a high relationship with the radiosensitivity and the individualized radiation response [4]. The signi cant changes of metabolites pro les were detected after radiotherapy in the serum samples of different malignancies such as hepatocarcinoma and breast cancer [5,6]. Moreover, carbon ion has been proved to produce different metabolic status when compared to photon [7]. But the CIRT induced metabolic changes of prostate cancer were largely unknown.
Furthermore, one pilot study from Poland evaluated the free amino acid pro les in both prostate patients' serum and urine samples and tried to nd a better biomarker other than PSA in order to improve the sensitivity and speci city for prostate cancer diagnosis [8]. The result is interesting, it was found that the metabolites might have similar or even better performance than PSA (AUC ranging from 0.53 to 0.83) [9] for PCa detection, which indicated that metabolic changes might add value to early evaluate the treatment response after radiotherapy. Recently, limited metabolomics studies evaluated the PCa treatment response to radiation and showed that the most signi cant alterations after photon irradiation were linked to metabolic pathways of nitrogen, pyrimidine, purine, porphyrin, alanine, aspartate, glutamate, and glycerophospholipid [10]. In another quality-of-life photon studies, metabolomics was alternatively used to determine individualized radiation toxicities [11]. But there were lack studies that explored the carbon-ion induced metabolic reprogramming and the individualized response in PCa patients.
Our aim was to preliminary investigate carbon-ion induced metabolic reprogramming in PCa patients and individualized response of PCa patients to carbon ion. As a kind of completely non-invasive biological uid, it contains over 2500 metabolites and allows us to observe global metabolic changes in cancer patients [12]. The discriminatory metabolites and pathways will be identi ed and analyzed. We expect this primary investigation of carbon-ion induced metabolic reprogramming and the individualized response of PCa patients will further step up the PCa CIRT treatment, and will also add value to either CIRT or photon radiotherapy in other malignancies.

Study samples and population
From July 2020 to December 2020, 15 patients with pathologically con rmed cTNM prostate adenocarcinoma were enrolled in this study. Radiotherapy was delivered with carbon ion beam by Siemens IONTRIS particle therapy device in Shanghai Proton and Heavy lon Center. The clinical target volume (CTV) included the prostate with or without proximal seminal vesicles based on different risk group types. The median CIRT doses of prostate was 60.4 GyE (range 55.2-65.6 GyE) in 12-16 fractions, and was prescribed to the 99% isodose line. Risk strati cation was based on NCCN guidelines version 2.2020. Demographic and clinical characteristics of enrolled patients are demonstrated in Table 1. ADT, androgen-deprivation therapy.

Samples collection and preparation
Urine samples were collected right before and after CIRT. All the samples were deposited under 4 ℃ immediately after collection. 0.22 µm membrane lters were used to remove contaminated bacteria from urine samples before stored at -80 ℃. All urine samples were thawed at room temperature on ice. 800 µl chilled methanol/acetonitrile (1:1, v/v) was add to 200 µl samples. The mixture was vortexed for 30 sec, sonicated for 10 min, incubated for 1 hour at − 20 ℃ and then centrifuged at 13000 g for 15 min at 4°C. Data collection and metabolites denti cation The raw data was acquired by UPLC -MS/MS and the le format was conversed using MSconvert software and Analysis Base File Converter. Conversed data les were processed by MSDIAL for identi cation. Upload MSconvert software conversed data onto XCMS online for peak detection and alignment. The parameters were set as ppm = 30, minimum peak width = 10, maximum peak width = 30, Signal/Noise threshold = 6, mzdiff = 0.01, mzwid = 0.025. Peak table were uploaded onto MetDNA for the denti cation of metabolites. Upload the acquired peak table to Metaboanalyst for statistical analysis and metabolic pathway analysis. Concentrations of metabolites were represented by peak area and normalized by data of creatinine.

Analysis
MetaboAnalyst 5.0 was used to select the metabolites statistically signi cant differences between pre-CIRT samples and post-CIRT samples, performed by volcano plot, which is combination of fold change (FC) analysis and non-parametric tests. False discovery rates (FDR) were calculated to reduce the incidence of false-positives. Unsupervised principal component analysis (PCA) was performed to detect the signi cant separation shift between compared groups. Supervised multivariate analysis, partial leastsquares discriminant analysis (PLS-DA), was performed to achieve maximum separation among the groups. Sparse PLS-DA (sPLS-DA) algorithm was used to reduce the number of variables (metabolites) to produce robust and easy-to-interpret models. Hierarchical cluster analysis (HCA) was used to separate the metabolites pro les between compared groups. Boxplots were used to show changes in urine metabolite concentrations from PCa patients showed the minimum, lower quartile, median, upper quartile, and maximum values of concentrations of metabolites.

Results
CIRT treatment signi cantly altered the urine metabolites pro les in PCa patients Moreover, HCA of metabolites can clearly discriminate the majority of pre-CIRT samples with post-CIRT samples ( Supplementary Fig. 1). Furthermore, the heat map shown in Fig. 2a reveals the concentration of metabolites in the urine sample experienced down-regulation in most patients after CIRT. Volcano plot identi es 35 signi cantly altered metabolites after CIRT, and these metabolites mainly are amino acid (Fig. 2b). 33 of the 35 urine metabolites are down-regulated after CIRT treatment, and typical metabolites include L-Glutamate, L-Glutamine, L-Cystine, glutathione, anthranilate, 5'-Methylthioadenosine. Two urine metabolites are up-regulated, including (R)-4'-Phosphopantothenoyl-L-cysteine, betaine (Fig. 2c). The above results indicate the CIRT can signi cantly alter the PCa metabolism, and especially can signi cantly decrease the amino acid metabolism.
CIRT induced metabolic changes mainly enriched in arginine biosynthesis and histidine metabolism We further performed pathway enrichment analysis of identi ed metabolites, and these metabolites could be enriched in 8 pathways (FDR < 0.05, impact > 2), including histidine metabolism, arginine biosynthesis, glutathione metabolism, cysteine and methionine metabolism, pantothenate and CoA biosynthesis, biotin metabolism, alanine, aspartate and glutamate metabolism, D-Glutamine and D-glutamate metabolism.
These metabolic pathways are part of amino acid metabolism, carbohydrate metabolism as well as cofactor and vitamin metabolism. The bubble plot shown in Fig. 3a demonstrates the signi cance and the impact of each pathway. Figure 3b demonstrates the altered pathway sorted by impact factor from top to bottom. Figure 4a-b show metabolites in arginine biosynthesis and histidine metabolism. Supplementary Figs. 2 and 3 demonstrate the details of the other 6 signi cantly altered pathways. More than 2 metabolites are identi ed in every pathway. Table 2 shows the FDR and the impact of enriched pathways of altered metabolites of the patient's urine before and after CIRT. The alteration of arginine biosynthesis (Match status = 6/14) and histidine metabolism (Match status = 7/16) pathways by CIRT are the most signi cant, with an impact factor of 0.6 (FDR = 0.0093) and 0.5 (FDR = 0.0328), respectively. L-Glutamine, L-glutamate, L-Arginine, L-Citrulline, N-(L-Arginino)succinate, L-Ornithine in arginine biosynthesis are all downregulated (Fig. 4c), and L-Histidine, L-Glutamate, urocanate, N(pi)-Methyl-Lhistidine, carnosine, imidazole-4-acetate in histidine metabolism are as well down-regulated (Fig. 4d). Metabolites pro les potentially to be a response indicator of CIRT Moreover, the relation of metabolic clustering with different risk classi cation was further explored.
According to the risk strati cation, the low-risk and medium-risk patients were considered as a relatively low-risk group, and the high-risk and very high-risk patients were considered as a relatively high-risk group (Table 3).
Firstly, the metabolites of patients' pre-CIRT urine were further analyzed by PLS-DA and be well clustered into two groups and the results matched with the risk subtype (Fig. 5a). However, the PLS-DA analysis of post-CIRT urine metabolites shows more overlap (Fig. 5b), indicating patients assessed as the same risk subtype no longer represented similar metabolites pro les. This means CIRT could signi cantly decrease the discrimination of the risk strati cation, indicating a possibility that CIRT could decrease the tumor heterogeneity.
Secondly, the HCA shows that after CIRT, the patients can be clustered into two groups according to the metabolites pro les of the patient's urine, named as PM1 and PM2 group (Fig. 5c). This clustering is different from the risk subtype. PM2 group shows a higher concentration of urine metabolites than that of PM1 group, which means the patients in two groups may have different response to CIRT. Discriminatory urine metabolites after CIRT mainly enriched in pathways of arginine biosynthesis and phenylalanine, tyrosine, and tryptophan biosynthesis Pathway enrichment was further performed to show the response diversity of PCa to CIRT. Bubble chart of discriminatory metabolites is shown in Fig. 6a. Table 4

Discussion
In this study, carbon ion showed its strong ability to inhibit the metabolism of prostate cancer. After CIRT treatment according to SOP in SPHIC, almost of the metabolites (33/35) were down-regulated. This result demonstrates the ability of CIRT treatment to inhibit the metabolism of tumors. Moreover, CIRT could generally inhibit most of the metabolism process, which will inhibit the proliferation, metastasis, and nally the progression of prostate cancer. Carbon ion is a novel tool for radiotherapy, and scarce studies revealed the in uences of carbon ion on cancer metabolism, especially the prostate tumor. Although we don't have the data from photon radiotherapy, results in this study primarily suggest that one of the reasons for the strong inhibition effects of carbon ion on prostate cancer attribute to its ability to signi cantly down-regulate metabolism.
Herein, CIRT induced pro les changes of metabolites mainly enriched in arginine biosynthesis and histidine metabolism pathways, which were signi cantly inhibited by carbon ion. Arginine biosynthesis pathway played a key role and was up-regulated in the PCa progression [13,14]. Arginine deprivation for the treatment of PCa has been investigated and proved effective [15]. Moreover, increased metabolism of L-arginine by myeloid cells can result in the impairment of lymphocyte responses to antigen during immune responses and tumor growth [16]. Thus, carbon ion induces such a high downregulation of arginine metabolism will bene t the inhibition of prostate cancer progression and have the potential to promote anti-tumor immune effects.
Interesting, we found the urine metabolites of these PCa patients have different response to CIRT treatment. And all of the patients could be clustered into two groups PM1 and PM2 after CIRT. PM2 showed relatively higher metabolites concentration. The clustered result was different from clinical risk strati cation. Because the carbon ion will decrease metabolites of urine, herein the difference of metabolites concentration between PM1 and PM2 should attribute to the tumor sensitivity to carbon ion. PCa of PM2 group patients seems not that sensitive like PM1 group patients. Although we should wait for the follow-up result, it is still necessary to mention the importance of these results, which may have the potential to give a quick judgment of the treatment endpoint right after CIRT. Until now, there are no effective and powerful clinical tools to realize it.
Metabolomics results also showed the signi cant metabolites of CIRT response and enriched pathway showed that urine of PM2 group patients presented higher metabolites related to arginine biosynthesis as well as phenylalanine, tyrosine, and tryptophan biosynthesis. These results con rmed our result above, that arginine biosynthesis is important for PCa and maybe play a central role in carbon ion response. Arginine biosynthesis may be a promising indicator for individualized response of CIRT.

Conclusion
In this study, carbon ion showed its strong ability to inhibit almost of the metabolism pathways of PCa. CIRT induced changes of metabolites pro les mainly enriched in arginine biosynthesis and histidine metabolism. Urine metabolites of PCa patients had different response to CIRT treatment. More sensitive PCa showed lower level of metabolites in urine, especially the arginine biosynthesis as well as phenylalanine, tyrosine, and tryptophan biosynthesis pathway. Carbon-ion evoked metabolic programming seems to be one of the most important underlying mechanisms of CIRT to inhibit PCa. Urine metabolites also showed their potentials to identify the individualized response of PCa patients to CIRT radiotherapy. Due to the limitation of patient number, large amount investigation is needed.      The individualized difference of metabolites pro les among patients after CIRT treatment. a PLS-DA analysis of relatively low-risk group (yellow) and the relatively high-risk group (purple) shows less overlap in pre-CIRT samples; b PLS-DA analysis of relatively low-risk group (yellow) and the relatively high-risk group (purple) shows less overlap in post-CIRT samples; c HCA of metabolites in post-treated samples.
Risk subtype is demonstrated in yellow (relatively low-risk group) and purple (relatively high-risk group). Metabolomics response subtype is demonstrated in orange (PM1) and blue (PM2). Up-regulated metabolites are shown in red and down-regulated in blue. The intensity of the color estimates the magnitude of change. d Schematic description of individualized metabolomics response difference. e Schematic description of CIRT induced disappearance of metabolomics difference between relatively high-risk and low-risk patients.