L-Arginine/NO Pathway Metabolites in Colorectal Cancer: Relevance as Disease Biomarkers and Predictors of Adverse Clinical Outcomes Following Surgery

The L-Arginine/NO pathway is involved in carcinogenesis and immunity. Its diagnostic and prognostic value in colorectal cancer (CRC) was determined using tandem mass spectrometry in 199 individuals (137 with CRC) and, during a three-day follow up, in 60 patients undergoing colorectal surgery. Citrulline was decreased and asymmetric (ADMA) and symmetric (SDMA) dimethylarginines and dimethylamine (DMA) were increased in CRC. The DMA increase corresponded with CRC advancement while arginine, ADMA, and SDMA levels were higher in left-sided cancers. Arginine, citrulline, ADMA, and DMA dropped and SDMA increased post incision. Females experienced a more substantial drop in arginine. The arginine and ADMA dynamics depended on blood loss. The initial SDMA increase was higher in patients requiring transfusions. Postoperative dynamics in arginine and dimethylarginines differed in robot-assisted and open surgery. Concomitant SDMA, citrulline, and DMA quantification displayed a 92% accuracy in detecting CRC. Monitoring changes in arginine, ADMA, and SDMA in the early postoperative period predicted postoperative ileus with 84% and surgical site infections with 90% accuracy. Changes in ADMA predicted operative morbidity with 90% and anastomotic leakage with 77% accuracy. If positively validated, L-arginine/NO pathway metabolites may facilitate CRC screening and surveillance, support differential diagnosis, and assist in clinical decision-making regarding patients recovering from colorectal surgery.

For analyzing pathway metabolites as potential differential biomarkers, data on systemic arginine, citrulline, ADMA, SDMA, and DMA in patients with inflammatory bowel disease (IBD) (n = 100; 26 with inactive and 74 with active disease) and irritable bowel syndrome (n = 18) were retrieved from our recently published paper [16].

Study Population in a Follow-Up Analysis
Among the CRC patients enrolled, 60 were followed for three days, with their blood sampled additionally after 8, 24, and 72 h post incision. Blood samples for all time points were available for only 51 patients, as nine cases of individual samples went missing. Patients from the follow-up group underwent either open colorectal surgery (OCS) or robot-assisted colorectal surgery (RACS), using the da Vinci ® Si surgical system (Intuitive Surgical, Sunnyvale CA, USA). Patient data were recorded prospectively in the departmental database. The recorded parameters included the patient's demographics, comorbidities, perioperative outcomes, and pathology report. The American Society of Anesthesiologists Physical Status Classification System (ASA grade) and the Charlson Comorbidity Index (CCI) were used for predicting complications. The standard perioperative pathway, involving mechanical bowel preparation, low molecular weight heparin, and perioperative antibiotic prophylaxis, was applied to all patients. Postoperatively, liquids were introduced orally on the first postoperative day, advancing to a solid diet on postoperative days two and three, if tolerated. Nasogastric tubes were avoided. Surgical closed drains were routinely used and removed on postoperative day one. The criteria for discharge included: A tolerance for a soft diet and no patient complaints nor reported complications. Parenteral opioids were used for pain control in the first three postoperative days; they were then gradually replaced with nonsteroidal anti-inflammatory drugs (NSAIDs). All patients were operated on under standard general anesthesia. Intravenous agents used for induction included propofol, fentanyl, and rocuronium. Anesthesia was maintained with sevoflurane. A local anesthesia or epidural was not used in the study group. All patients were given NSAIDs (metamisol) before waking up or immediately after the surgery-in the recovery room. Postoperative complications were recorded within 30 days after surgery and classified in accordance with the Clavien-Dindo Classification (CDC). Surgical site infections (SSI) were recorded prospectively for all patients and classified in accordance with the Center for Disease Control and Prevention criteria. The restoration

Ethical Approval
The study protocol was approved by the Medical Ethics Committees of Regional Specialist Hospital (#KB/nr 1/rok 2012 from 26 June, 2012).

Sample Collection
Blood was drawn by a venipuncture after overnight fasting into serum-separator tubes, clotted for 30 min, and subsequently centrifuged for 15 min at 10 • C and 720× g. The resulting serum was collected and stored frozen in aliquots at −80 • C.

LC-MS/MS
The derivatized samples were then separated on a nanoACQUITY HSS T3 column (C18-phase, internal diameter 1 mm, length 50 mm, particle size 1.75 µm) at a flow rate of 250 µL × min −1 using the Waters nanoACQUITY UPLC System (2 µL injection volume) controlled by the MassLyinx software (Waters). The LC eluents were water (A) and methanol (B), both containing 0.1% of formic acid. Elution was performed in a linear gradient as follows: 5% B for 0.5 min, from 5% to 14% B in 2.5 min, from 14% to 60% B in 1 min, from 60% to 90% B in 0.5 min, 90% B for 0.5 min, and from 90% to 5% B in 0.10 min, the re-equilibration time was 4.9 min.

Statistical Analysis
The Kolmogorov-Smirnov test was used to determine the normality of data distribution and the Levene test to assess the homogeneity of variances. One-way ANOVA with the Student-Newman-Keuls post hoc test or Kruskal-Wallis H test with the Conover post hoc test were used for multi-group comparisons. The t-test that was used for independent samples, with the Welch correction of variance was non-homogeneous, or the Mann-Whitney U test was used for two-group comparisons. If not otherwise stated, data are presented as means with a 95% confidence interval (CI) or medians with an interquartile range (IQR). The correlation analysis was conducted using Pearson correlation (r) or Spearman rank correlation (ρ) tests.
The logistic regression (backward method) was applied to select explanatory variables independently associated with a given dependent variable. Variables were entered if p < 0.05 and removed if p > 0.1. The variables retained in logistic regression models were subsequently used for constructing multi-metabolite biomarker panels.
The diagnostic power of individual metabolites and multi-metabolite panels as diagnostic or predictive biomarkers was evaluated using the receiver operating characteristic (ROC) curve analysis. Their overall accuracy was assessed in terms of the area under the ROC curve (AUC). The optimal cutoff with corresponding sensitivities and specificities as well as positive and negative likelihood ratios (LR+ and LR−, respectively) was calculated. The Youden index (J; the sum of sensitivity and specificity minus one) was estimated for comparative purposes as well. The following criteria for AUC interpretation were used: AUC from 0.90 to 1-excellent accuracy, from 0.81 to 0.90-good accuracy, from 0.71 to 0.80-fair accuracy, from 0.61 to 0.70-poor accuracy, and from 0.50 to 0.60-fail.
All tests were two-sided and p < 0.05 was considered statistically significant. The analyses were conducted using the MedCalc Statistical Software version 19.1.5 (MedCalc Software bv, Ostend, Belgium; https://www.medcalc.org; 2020). CRC patients had a significantly higher arginine, ADMA, and SDMA and lower citrulline than healthy controls. The difference in DMA did not reach statistical significance but the metabolite was significantly elevated in patients with adenomas as compared to either CRC patients or controls. SDMA, despite a low number of observations in the adenoma group, was elevated as compared to controls as well (Table 3). To determine a possible difference in the metabolite pattern between CRC and non-malignant bowel conditions, which would allow for differential diagnosis, metabolite concentrations in CRC patients were compared with pooled data from patients with adenomas, inflammatory bowel disease (IBD), and irritable bowel syndrome.

Preoperative
As compared to patients with non-malignant conditions, CRC was characterized by a significantly lower systemic citrulline and DMA but higher SDMA (Table 4).

Association between Pathway Metabolites and Clinicopathological Data
Of the evaluated metabolites, only DMA showed a significant association with the disease TNM stage, being elevated in advanced cancers ( Figure 2). Of the evaluated metabolites, only DMA showed a significant association with the disease TNM stage, being elevated in advanced cancers ( Figure 2).  In turn, arginine, ADMA, and SDMA were associated with the location of a primary tumor and were significantly more elevated in patients with left-sided cancers as compared to rectal ones or, in the case of arginine, also as compared to right-sided cancers (Table 5). Data presented as medians with an interquartile range. Arg: Arginine; Cit: Citrulline; ADMA: Asymmetric dimethylarginine; SDMA: Symmetric dimethylarginine; DMA: Dimethylamine; 1 significantly different from left-sided cancers; 2 significantly different from right-sided cancers; 3 significantly different from rectal cancers; 4 data presented as geometric means with a 95% confidence interval and analyzed using one-way ANOVA on log-transformed data; 5 data presented as arithmetic means with a 95% confidence interval and analyzed using one-way ANOVA; 6 data presented as medians with an interquartile range and analyzed using the Kruskal-Wallis H test.

Suitability of Pathway Metabolites as CRC Biomarkers
The ROC curve analysis was applied to assess the strength of association between L-arginine/NO pathway metabolites and CRC presence, as well as to evaluate the metabolites suitability as potential biomarkers for the purpose of CRC screening in the general population, CRC surveillance in patients with IBD, or for differential diagnosis.

CRC Screening in the General Population
Individually, ADMA, SDMA, and citrulline displayed a statistically significant ability to distinguish CRC patients from controls, of which SDMA had a superior diagnostic accuracy, as indicated by the highest AUC ( Figure 3) and Youden index ( Table 6). In turn, arginine, ADMA, and SDMA were associated with the location of a primary tumor and were significantly more elevated in patients with left-sided cancers as compared to rectal ones or, in the case of arginine, also as compared to right-sided cancers ( Table 5). Data presented as medians with an interquartile range. Arg: Arginine; Cit: Citrulline; ADMA: Asymmetric dimethylarginine; SDMA: Symmetric dimethylarginine; DMA: Dimethylamine; 1 significantly different from left-sided cancers; 2 significantly different from right-sided cancers; 3 significantly different from rectal cancers; 4 data presented as geometric means with a 95% confidence interval and analyzed using one-way ANOVA on log-transformed data; 5 data presented as arithmetic means with a 95% confidence interval and analyzed using one-way ANOVA; 6 data presented as medians with an interquartile range and analyzed using the Kruskal-Wallis H test.

Suitability of Pathway Metabolites as CRC Biomarkers
The ROC curve analysis was applied to assess the strength of association between L-arginine/NO pathway metabolites and CRC presence, as well as to evaluate the metabolites suitability as potential biomarkers for the purpose of CRC screening in the general population, CRC surveillance in patients with IBD, or for differential diagnosis.

CRC Screening in the General Population
Individually, ADMA, SDMA, and citrulline displayed a statistically significant ability to distinguish CRC patients from controls, of which SDMA had a superior diagnostic accuracy, as indicated by the highest AUC ( Figure 3) and Youden index ( Table 6). model allowed for a correct classification of 89% of tested subjects. The predicted probabilities calculated were subsequently used to determine the diagnostic accuracy of the model in terms of the ROC curve analysis. As depicted in Figure 3f and Table 6, the metabolic panel SDMA/citrulline/DMA displayed a superior 92% accuracy and the highest combination of sensitivity and specificity as compared to its individual components.    The logistic regression was employed to select independent predictors of CRC presence. From among the examined metabolites, SDMA, citrulline, and DMA were selected. The goodness-of-fit for the model, as indicated by the Hosmer and Lemeshow test, was χ 2 = 6.4, DF = 8, p = 0.599 and the Nagelkerke R 2 value was 0.64, indicating an overall usefulness of the model in predicting CRC. The model allowed for a correct classification of 89% of tested subjects. The predicted probabilities calculated were subsequently used to determine the diagnostic accuracy of the model in terms of the ROC curve analysis. As depicted in Figure 3f and Table 6, the metabolic panel SDMA/citrulline/DMA displayed a superior 92% accuracy and the highest combination of sensitivity and specificity as compared to its individual components.
As the concentration of SDMA, one of the components of the proposed diagnostic model, was found to depend on the location of the primary tumor (Table 5), its diagnostic performance was evaluated separately for each tumor location. As shown in Table 7, the model accuracy remained excellent for the left-sided CRC and was very good for the right-sided and rectum CRC. Differences that are more substantial were observed in model sensitivity and specificity. Concomitant quantification of SDMA, citrulline, and DMA was characterized by an excellent sensitivity and rather poor specificity for the right-sided and rectum CRC and by a very good sensitivity and a fair specificity for the left-sided CRC.

CRC Surveillance in Patients with IBD
In addition, metabolite suitability for CRC surveillance in IBD patients was determined by evaluating their power to differentiate patients with CRC and inactive IBD. Individually, citrulline and SDMA displayed a statistically significant ability to differentiate CRC from patients with inactive IBD with moderate and poor accuracy, respectively ( Figure 4). In the logistic regression, however, ADMA and SDMA were selected as independent variables. The goodness-of-fit for the model, as indicated by the Hosmer and Lemeshow test, was χ 2 = 2.8, DF = 8, p = 0.947 and the Nagelkerke R 2 value was 0.14, indicating an overall usefulness of the model in predicting CRC. The model allowed for a correct classification of 82.8% of tested subjects. The diagnostic accuracy of the model as well as the optimal combination of sensitivity and specificity were, however, inferior to individual determination of citrulline ( Figure 4, Table 8).

Differential Diagnosis
Metabolites that help differentiate between CRC and patients with conditions displaying overlapping clinical symptoms (patients with adenomas, active inflammatory bowel disease, and irritable bowel syndrome) was determined. Individually, SDMA, citrulline, and DMA displayed a

Differential Diagnosis
Metabolites that help differentiate between CRC and patients with conditions displaying overlapping clinical symptoms (patients with adenomas, active inflammatory bowel disease, and irritable bowel syndrome) was determined. Individually, SDMA, citrulline, and DMA displayed a statistically significant, although poor, ability to differentiate CRC from patients with non-malignant bowel conditions ( Figure 5 and Table 9). Table 9. Characteristics of L-arginine/NO pathway metabolites as differential CRC biomarkers. The goodness-of-fit for the model, as indicated by the Hosmer and Lemeshow test, was χ 2 = 14.6, DF = 8, p = 0.068 and the Nagelkerke R 2 value was 0.33, indicating an overall usefulness of the model in predicting CRC. The model allowed for a correct classification of 69.6% of tested subjects. As depicted in Figure 5f and Table 9, the metabolic panel SDMA/ADMA/DMA displayed a superior 78.4% accuracy and the highest combination of sensitivity and specificity as compared to its individual components.   The logistic regression identified SDMA, DMA, and ADMA as independent CRC predictors. The goodness-of-fit for the model, as indicated by the Hosmer and Lemeshow test, was χ 2 = 14.6, DF = 8, p = 0.068 and the Nagelkerke R 2 value was 0.33, indicating an overall usefulness of the model in predicting CRC. The model allowed for a correct classification of 69.6% of tested subjects. As depicted in Figure 5F and Table 9, the metabolic panel SDMA/ADMA/DMA displayed a superior 78.4% accuracy and the highest combination of sensitivity and specificity as compared to its individual components.

Time-Course of Pathway Metabolites during the Early Postoperative Period
In the early postoperative period, a serum concentration of pathway metabolites changed significantly. The changes in arginine, ADMA, and DMA concentration over time displayed a quadratic trend and a linear trend for citrulline or a cubic trend for SDMA ( Figure 6).

Time-Course of Pathway Metabolites during the Early Postoperative Period
In the early postoperative period, a serum concentration of pathway metabolites changed significantly. The changes in arginine, ADMA, and DMA concentration over time displayed a quadratic trend and a linear trend for citrulline or a cubic trend for SDMA ( Figure 6).

Impact of Surgery-and Patient-Related Factors on Metabolite Time-Course in the Early Postoperative Period
The potential impact of patient-and surgery-related factors on the early dynamics in metabolite concentrations was examined (Table 10). From among patient-related factors such as age, BMI, and an overall health condition (expressed in terms of CCI or ASA), none had a significant effect. Only sex affected arginine and Arg/ADMA, as an early drop in arginine was observed exclusively in females. From among the surgery-related factors such as length of surgery and its extent (estimated in terms of total number of resected lymph nodes) and estimated blood loss (EBL), only the latter had a significant impact on arginine and ADMA and their ratio. The higher the EBL, the more marked the initial drop in arginine and ADMA (∆ 8/0 ) and the smaller the change between 24 and 8 h post incision (∆ 24/8 ).
In addition, the change in DMA between 24 and 8 h post incision (∆ 24/8 ) tended to decrease with the increasing EBL ( Table 6). The SDMA increase at 8 h (∆ 8/0 ) tended to positively correlate with EBL (r = 0.22, p = 0.096) and was significantly higher in patients requiring transfusion (1.39 ± 1.1 vs. 0.98 ± 0.27, p = 0.024). The SDMA increase at 24 h (∆ 24/0 ) tended to be higher in males while the drop in arginine was more pronounced in females (Table 10).

Association between Changes in Metabolite Concentration and Adverse Clinical Events: Suitability of Pathway Metabolites as Predictors of Adverse Clinical Events
The postoperative complications typically appear clinically on the 5-7th day. Meanwhile, the recently introduced "fast track" ERAS perioperative care protocol allows colorectal surgeons to discharge patients home as early as on the 3rd postoperative day. Markers hinting at a possibility of postoperative complications prior to their clinical manifestation might facilitate a safe early discharge from the hospital. Therefore, a possible association between the dynamics of changes in metabolite concentration and adverse clinical outcomes such as operative morbidity (expressed in terms of the Clavien-Dindo Classification (CDC)), presence of anastomotic leakage (AL), surgical site infections (SSI), delayed restoration of bowel function (RoBF), and prolonged hospital stay (length of hospital stay (LoHS)) were evaluated. Significant differences in ratios between time-points were subsequently assessed as potential markers of adverse clinical events using the ROC curve analysis. discharge from the hospital. Therefore, a possible association between the dynamics of changes in metabolite concentration and adverse clinical outcomes such as operative morbidity (expressed in terms of the Clavien-Dindo Classification (CDC)), presence of anastomotic leakage (AL), surgical site infections (SSI), delayed restoration of bowel function (RoBF), and prolonged hospital stay (length of hospital stay (LoHS)) were evaluated. Significant differences in ratios between time-points were subsequently assessed as potential markers of adverse clinical events using the ROC curve analysis.

Operative Morbidity (Clavien-Dindo Classification) and Anastomotic Leakage
There was no change in the ADMA concentration between 8 and 24 h post incision (Δ24/8) in patients with a low level of postoperative complications expressed in terms of CDC < 3 (0.98 ± 0.2) while patients with CDC ≥ 3 experienced a drop in metabolite concentration (0.78 ± 0.2, p = 0.024). In addition, at 72 h as compared to 8 h post incision (Δ72/8), there was a slight increase in patients with CDC < 3 and a decrease in patients with CDC ≥ 3 (1.13 ± 0.3 vs. 0.88 ± 0.2, p = 0.022) (Figure 8).
There were no significant differences in the dynamics of citrulline and DMA. The Arg/ADMA ratio at 24 h post incision (∆ 24/0 ) dropped in patients with a delayed restoration of bowel function but remained at the same level in patients with RoBF < 5 days (0.81 ± 0.23 vs. 1.07 ± 0.43, p = 0.005).

Prolonged Hospitalization
The concentration of SDMA dropped at 72 h post incision, as compared to its preoperative levels in patients hospitalized < 7 days, but remained at comparable levels in those hospitalized longer (∆ 72/0 : 0.82 ± 0.25 vs. 1.02 ± 0.33, p = 0.014) (Figure 14). There were no significant differences in the time-course of other metabolites with respect to the length of hospitalization. As a LoHS ≥ 7 marker, SDMA displayed a poor overall accuracy and 58.8% sensitivity and 78.4% specificity accompanying >0.92 µM cutoff. Youden index was 0.372 and likelihood ratios LR+ and LR− were 2.7 and 0.53, respectively.   longer (Δ72/0: 0.82 ± 0.25 vs. 1.02 ± 0.33, p = 0.014) (Figure 14). There were no significant differences in the time-course of other metabolites with respect to the length of hospitalization. As a LoHS ≥ 7 marker, SDMA displayed a poor overall accuracy and 58.8% sensitivity and 78.4% specificity accompanying > 0.92 µM cutoff. Youden index was 0.372 and likelihood ratios LR+ and LR-were 2.7 and 0.53, respectively.

Discussion
Metabolic reprogramming and avoiding destruction by the immune system have been recognized as emerging cancer hallmarks. Tumor-promoting inflammation, in turn, is known as its enabling characteristic [8]. The L-arginine/NO pathway is implicated in all three. Unsurprisingly, the pathway metabolite flux is affected by the disease and reflected at the systemic level by the altered pathway metabolite concentration, which might potentially be used for cancer detection. Targeted metabolomics enables simultaneous quantification of a set of metabolites allowing for the creation of multi-metabolite biomarker panels. The diagnostic accuracy of multi-analyte panels is potentially superior to the power of their individual components. This has been demonstrated previously, e.g.,

Discussion
Metabolic reprogramming and avoiding destruction by the immune system have been recognized as emerging cancer hallmarks. Tumor-promoting inflammation, in turn, is known as its enabling characteristic [8]. The L-arginine/NO pathway is implicated in all three. Unsurprisingly, the pathway metabolite flux is affected by the disease and reflected at the systemic level by the altered pathway metabolite concentration, which might potentially be used for cancer detection. Targeted metabolomics enables simultaneous quantification of a set of metabolites allowing for the creation of multi-metabolite biomarker panels. The diagnostic accuracy of multi-analyte panels is potentially superior to the power of their individual components. This has been demonstrated previously, e.g., multi-marker diagnostic panels consisting of inflammatory mediators [21,22] or urine amino acid-related metabolites in mice CRC models [23]. Indeed, although CRC patients from our cohort differed significantly from healthy individuals in terms of citrulline (lower) and ADMA, SDMA, and DMA (higher), these metabolites individually possessed poor accuracy. The overall fair performance of SDMA was an exception but its individual impact was still outperformed by a panel consisting of SDMA, citrulline, and DMA. This panel displayed excellent accuracy as a potential CRC screening marker in the asymptomatic population. If positively validated, this finding is promising. Fecal-based assays, currently used for screening purposes, are comparable in cost and diagnostic power (characterized by lower sensitivity but higher specificity [4]) and similarly non-invasive, but are currently underused due to poor compliance of patients [7]. Blood-based metabolic panels might be better tolerated and have the advantage of being applicable in IBD patients.
In line with previous findings [12], CRC patients had a decreased systemic arginine. Serum arginine depletion has been attributed to the upregulated amino acid uptake by tumors and linked with their accelerated growth, with concomitant obstruction of an anti-tumor immune response [12]. The decrease in systemic arginine might be an early event in colonic carcinogenesis as it is also seen in active IBD [16], a condition associated with an increased risk of CRC [24]. ADMA and SDMA have been shown to accumulate in tumor tissues as well [13,23], and are elevated in the urine of CRC-bearing mice, together with arginine and citrulline [23]. In prostate cancer, however, the disease has been associated with increased arginine as compared to benign prostate hyperplasia [25]. In our study, eight patients were verified as having colorectal adenomas, and despite a low number of observations, were characterized by a statistically higher SDMA concentration than healthy controls, as well as higher DMA than controls and CRC patients. While promising, the potential of SDMA and DMA quantification for differential diagnosis requires the analysis of a larger group of adenoma patients. In addition to large adenomas, active IBD and functional bowel disorders, such as irritable bowel syndrome, are present with a set of symptoms similar to CRC. We found that CRC is distinguished, among conditions presenting overlapping symptoms, by lower systemic citrulline and DMA, and higher SDMA. The analysis of pathway metabolites' potential as differential markers and surveillance tools showed that the combination of ADMA, SDMA, and DMA allows for CRC detection in the symptomatic population with a fair accuracy, while citrulline alone is a fair surveillance marker.
SDMA and DMA are outside the mainstream research and data on their concentration in cancer are scant. A cancer-related elevation in systemic SDMA has been reported in hematological malignancies and found to be useful in predicting overall mortality [18]. In CRC, the metabolite accumulated within a tumor-adjacent tissue as compared to normal tissue, and was associated with an increased risk of metastasis [13]. In turn, metabolic profiling of CRC tissue samples has revealed that DMA accumulated in tumors with lymph node involvement, as compared to normal tissue [26]. A similar study in esophageal cancer has shown that DMA is associated with cancer progression but decreases in stage III tumors, and only its elevation in stage IV is significant as compared to normal tissue [27].
Here, systemic DMA was the only pathway metabolite reflecting CRC advancement. It displayed an association pattern analogous to the local DMA concentration in esophageal cancer. A systemic elevation of DMA is also seen in IBD [16], as demonstrated here, significantly more so than in CRC.
An interesting observation is that the baseline systemic metabolite concentration differs depending on the tumor anatomical subsite, with a higher concentration of arginine, ADMA, and SDMA in left-sided colonic tumors. There is a growing awareness of the molecular heterogeneity of cancers arising in different segments of the colorectum [28][29][30][31] that may also be reflected at the systemic level [22,32,33]. Proximal tumors are reportedly more advanced upon diagnosis, larger, and less differentiated [34], displaying poor responsiveness to chemotherapy and, consequently, associated with a worse prognosis [35]. Lower systemic arginine in right-sided cancers may indicate accelerated amino acid uptake by tumor cells and contribute to higher aggressiveness of cancers in this location. Owing to the association of arginine with immunity [36], it would also translate into the induction of immunosuppression in patients with right-sided CRC. As demonstrated here, the dependence of a potential biomarker on the primary tumor location should be taken into consideration as it might affect its diagnostic power.
The ADMA and arginine-to-ADMA ratio are considered sensitive indicators of endothelial dysfunction and therefore risk markers in cardiovascular medicine [37]. In cancer, arginine and ADMA have demonstrated potential as markers of cardiac dysfunction related to cancer therapeutics [38]. Moreover, previous studies on critically ill patients or those undergoing cardiac surgery have shown that ADMA may be a novel biomarker indicative of increased perioperative risk [39,40]. Corroborating this notion, the dynamics in ADMA concentration in our cohort differed depending on the Clavien-Dindo score, which, in turn, is inversely related to the five-year overall survival, disease-free survival, and cancer-specific survival and directly correlated with overall recurrence rates [41]. Indices based on ADMA dynamics might predict an increased risk for severe postoperative complications (CDC ≥ 3), with very good accuracy and excellent sensitivity prior to their clinical manifestation.
Likewise, biomarkers predictive of perioperative complications following colorectal resection are being extensively investigated [42]. To promote a complication-free recovery, the modern model of perioperative care, the "fast track", has been introduced and popularized worldwide [43]. This package of perioperative modifications results in fewer complications, reduction of cost, shortening of hospital stay, and quicker overall recovery. Patients undergoing colorectal surgery with such perioperative care are routinely discharged home as early as the 3rd-5th postoperative day. Unfortunately, this early discharge carries the risk of developing postoperative complications such as SSI, bowel obstruction or anastomotic leakage at home, where access to medical expertise is limited. Reliable markers of perioperative complications would help determine which patients are eligible for a safe early discharge. Therefore, we explored whether the changes in L-arginine/NO pathway metabolites in the early perioperative period may be useful as indicators of the most typical adverse clinical events such as anastomotic leakage, surgical site infections, and delayed restoration of bowel function. To the best of our knowledge, there is only one preliminary study, conducted on a mixed cohort (benign diseases included) of 16 patients undergoing laparoscopic colorectal surgery, in which arginine, ADMA, and citrulline concentration had been followed [44]. Corroborating the results of Ragina et al. [44], arginine in our study dropped directly after surgery but increased on postoperative day three. In our cohort, although ADMA displayed a similar trend to arginine, citrulline was decreasing during the whole observation period. As citrulline is considered a marker of gut functionality [45,46], its depletion in the early postoperative period, likely resulting from bowel manipulation and ileus, is to be expected. An interesting observation is that SDMA displayed an inverse pattern to ADMA. Both dimethylarginines are derived from proteins methylated by a group of protein arginine methyltransferases (PRMTs), of which type I enzymes yield ADMA and type II enzymes SDMA [47]. The pool of dimethylarginines is regulated by the rate of their synthesis as well as degradation, which includes renal excretion and enzymatic degradation to citrulline and DMA in the case of ADMA but solely renal excretion in the case of SDMA [11]. The anesthesia during surgery, blood loss, surgery-induced increase in aldosterone and antidiuretic hormone, and electrolyte imbalance associated with bowel manipulation and postoperative ileus may impair the renal function [48,49]. It is possible that such altered filtration in the early postoperative period hampers SDMA excretion, leading to metabolite accumulation in the blood. Acute kidney injury (AKI) is a relatively common complication after colorectal surgery [49], more so in males and patients with transfusions [48], with which, as shown here, the SDMA increase at 24 h post incision was positively correlated. In fact, SDMA has been proposed as a marker of renal function in children with nephrotic syndrome [50] and adults with chronic kidney disease [51]. As demonstrated in this study, quantifying perioperative SDMA concentrations might potentially be useful in monitoring patients recovering from CRC for AKI.
Colorectal surgery is followed by a temporary gut dysmotility, referred to as postoperative ileus. While some degree of postoperative ileus is an expected and physiological response to surgical trauma, its persistence is considered as one of the most common complications [52]. However, there is no consensus concerning its clinical definition and timeframe. Therefore, for the purpose of this study, a restoration of bowel function was defined as tolerance of solid diet and passage of first stool. The median value in the examined cohort was calculated and median + 1 day (thus 5th postoperative day) was established as a cutoff to define the prolonged (pathological) postoperative ileus, which is in agreement with the criteria proposed by Vather et al. [52]. The rationale behind seeking an association between perioperative alterations in the L-arginine/NO pathway and postoperative ileus is that NO synthesized by neuronal NOS is considered a main non-cholinergic and non-adrenergic inhibitory neurotransmitter in the gut, and NOS inhibitors have been shown to alleviate postoperative ileus in animal models [53]. Correspondingly, as demonstrated here, the most potent NOS inhibitor, ADMA, was more pronouncedly decreased in the early postoperative period in patients with a delayed restoration of bowel function. The NO synthesized by inducible NOS during the inflammatory response to surgical trauma and bowel manipulation further aggravates postoperative bowel dysfunction. In line with the role attributed to inflammation in hampering bowel motility, we have previously shown the ability of early postoperative concentrations of chemokines monocyte chemoattractant protein 1 (MCP-1/CCL2) and "regulated on activation, normal T cell expressed and secreted" (RANTES/CCL5) to predict postoperative ileus [54]. Here, patients with a prolonged gut dysmotility had a higher concentration of SDMA in the early postoperative period. It may be associated with a proinflammatory character of the metabolite as, in addition to being a weak inhibitor of NOS enzymes, SDMA activates NFκB and increases monocyte expression and secretion of IL-6 and TNFα [55]. Together, a panel consisting of arginine, ADMA, and SDMA displayed a superior performance in predicting persistent postoperative ileus over individual markers, with a satisfactory overall accuracy and specificity.
Independently from its role as a NO precursor, arginine is crucial for optimal functioning of the immune system. It stimulates T and NK cell proliferation, differentiation, and cytotoxicity. Accelerated arginine uptake by tumor cells yields T and NK cells deficient in arginine and abolishes their functioning. Moreover, the cancer-related increase in arginase-1 activity in macrophages further supports cancer growth by inducing immunosuppression [36]. Dysfunctional immune systems might render cancer patients more susceptible to infections. Correspondingly, our patients who developed surgical site infections were those who had earlier and more markedly decreased arginine. Additionally, NO participates in creating an immunosuppressive environment by stimulating cyclooxygenase-2 and the synthesis of proinflammatory mediators [36]. Accordingly, the NOS inhibitor, ADMA, was more pronouncedly decreased and proinflammatory SDMA was steadily increasing in patients who subsequently developed surgical site infections. A panel including all three metabolites had a very good overall accuracy and an excellent sensitivity owing to arginine and ADMA, and improved specificity as compared to individual metabolites owing to SDMA.
Anastomotic leakage is the most dreaded complication following colorectal surgery and its early indicators are sought after. Of the available biochemical markers, inflammatory indices C-reactive protein (CRP) and procalcitonin are used in the clinical practice and CRP is considered an extremely accurate biomarker of AL and thus a gold standard [56]. From among the metabolites investigated here, an index based on ADMA dynamics displayed fair accuracy and good sensitivity and specificity as an AL marker, however, it was inferior to the literature-reported performance of CRP [56,57].
As operative morbidity is dependent, among others, on the patient's age, sex, weight, and health status [58][59][60], the potential effects of age, sex, BMI, and ASA and CCI grade on metabolite dynamics in the postoperative period were examined. None were found except for arginine and the arginine-to-ADMA ratio, which decreased directly following surgery exclusively in female patients. In addition, the type of surgery affected arginine, ADMA, and SDMA. After an initial drop, their concentration increased more markedly following minimally-invasive robot-assisted surgery. Surgical resection is the mainstay of treatment for CRC and open surgery has been the gold standard of CRC resections over the past century. However, the shift to a minimally invasive surgery in this field began in the 1990s and laparoscopy has become the surgical approach of choice for most colorectal resections. It has been shown to improve short-term outcomes, without negatively affecting oncological outcomes [61]. Robotic surgery is believed to be the next step in this evolution. Robotic systems overcome certain limitations of laparoscopy and offer hand wristed instruments, three active working arms and a superb view of the operative field. All these factors make surgical robots the perfect tool for complex abdominal procedures [62,63]. Shibata et al. [64] and our group [54,65,66] have shown that robot-assisted colorectal surgery is also beneficial in terms of the inflammatory and immune response, in addition to the improvement of clinical outcomes [67]. Still, the biochemical background of the body's favorable response to robot-assisted surgery is poorly understood. To determine this advantage, we have compared changes in L-arginine/NO pathway metabolites following robot-assisted and classic open colorectal surgery. A slightly but significantly higher ADMA and SDMA concentration observed following the robot-assisted surgery in our study might contribute to reduced arginine availability for inducible NOS, and thus alleviate NO-associated inflammation and oxidative stress. Both dimethylarginines inhibit arginine uptake by the cationic amino acid transporter (CAT)-1 in addition to ADMA being a strong and SDMA a weak NOS inhibitor [11]. The more prompt restoration of arginine concentration following robotic surgery observed in this study seems to be beneficial, also in light of its role as a pharmaco-nutrient. Surgery-induced deficiency in arginine contributes to the disadvantageous immune shift towards Th2 response and postoperative immunosuppression [68]. These findings add to the existing literature, fostering a better understanding of the immune and inflammatory response that follows minimally invasive surgery.
In addition to the type of surgery, early postoperative dynamics of arginine and ADMA concentration were affected by the degree of blood loss. In particular, a metabolite drop at 24 h as compared to 8 h was more pronounced (thus yielding smaller ∆ 24/8 values) in patients with an estimated blood loss (EBL). This finding is consistent with the known causative effect of adverse circulatory conditions, including hemorrhage, on the dysfunction of endothelial cells [69]. Dysfunctional endothelium is characterized by a reduced NO synthesis by endothelial isoforms of NOS and a low-grade inflammation, conditions aggravated by decreased arginine availability. Accordingly, a restorative effect on endothelium has been attributed to L-arginine supplementation [69]. In turn, the lower the initial drop was in arginine, the higher the EBL. While counterintuitive, it may be the effect of interference by confounding factors at play during surgery. As an example, some of our patients were introduced to perioperative warming, a procedure reportedly associated with reduced intraoperative blood loss, since amino acids have been shown to induce a thermogenic response and counteract surgery-associated hypothermia [70]. Accordingly, a drop in arginine after 8 and 24 h post incision was lower, and the Arg/ADMA ratio even increased at 24 h in patients subjected to perioperative warming.

Conclusions
Concomitant determination of L-arginine/NO pathway metabolites using targeted metabolomics offers excellent accuracy in detecting CRC in a general asymptomatic population, superior over individual measurements. However, alterations in citrulline, SDMA, and DMA possess potential as markers for CRC screening and surveillance, as well as differential diagnosis with bowel diseases, presenting with a set of symptoms similar to CRC. In addition, SDMA holds promise as a marker of renal function in CRC patients recovering from colorectal surgery. Together with ADMA and arginine, SDMA may serve as a predictor of persistent postoperative ileus and surgical site infections, whereas monitoring changes in the ADMA concentration may help assess the risk of operational morbidity. Metabolite quantification in the early perioperative period, prior to clinical manifestation of common postoperative complications, may facilitate a prompt but safe discharge from the hospital. Taken together, concomitant quantification of key metabolites of the L-arginine/NO pathway in sera taken from CRC patients may assist in clinical decision making and warrants further investigation. Especially, external validation is needed to incorporate these proposed metabolite panels into the routine clinical practice. Funding: This publication was prepared under the project financed from the funds granted by the Ministry of Science and Higher education in the Regional Initiative of Excellence programme for the years 2019-2022, project number 016/RID/2019/19, the amount of funding was PLN 11998121.30. The enrollment of patients was supported by the Project, "WroVasc-Integrated Cardiovascular Centre", co-financed by the European Regional Development Fund, within the Innovative Economy Operational Program, 2007-2013 realized in the Regional Specialist Hospital, Research and Development Center in Wroclaw "European Funds-for the development of innovative economy".