D-mannose: a novel prognostic biomarker for patients with esophageal adenocarcinoma

Abstract

Metabolomic profiling is a promising approach to identify new biomarkers for cancer prognosis. However, the role of circulating metabolites as prognostic indicators in esophageal adenocarcinoma (EAC) has not been well explored. In this study, we aimed to evaluate the prognostic value of three serum metabolites, d-mannose, l-proline (LP), and 3-hydroxybutyrate (BHBA), which were significantly different between EAC patients and controls, identified through a global and targeted metabolite profiling. We measured the levels of d-mannose, LP, and BHBA in pretreatment serum from 159 EAC patients, using liquid chromatography–mass spectrometry/mass spectrometry (LC-MS/MS) methods. A multivariable Cox model was used to estimate the hazard ratios (HRs) and 95% confidence intervals (95% CIs) for the association of these metabolites with recurrence and overall survival. We found that serum levels of d-mannose were significantly associated with recurrence and overall survival in EAC patients, whereas levels of LP and BHBA were not. Compared with patients with a low (first tertile) level of d-mannose, those with a high (second plus third tertiles) level had 49% reduced risk of recurrence (HR = 0.51; 95% CI: 0.29–0.91; P = 0.02), and 56% reduced risk of death (HR = 0.44; 95% CI: 0.25–0.77, P < 0.01). The significant association of high d-mannose levels with better prognosis was consistent among patients with early-stage and advanced-stage EAC. Our results suggest that serum level of d-mannose may be used as a novel prognostic biomarker for patients with EAC. Further studies in independent populations are warranted to confirm our findings.

Introduction

Esophageal cancer (EC) accounts for approximately 455800 new cases and 400200 deaths worldwide in 2012, making it the eighth most common and sixth most deadly cancer (1). EC includes primarily two histologic subtypes, esophageal squamous cell carcinoma (ESCC) and esophageal adenocarcinoma (EAC). In recent decades, EAC is increasing rapidly in incidence in western countries, currently comprising more than 60% of new EC patients in the USA (2). The prognosis of EAC is dismal with overall 5-year survival rates of 20% for all stages combined and 40% for localized disease (3). Prediction of prognosis in patients with EAC is a clinical challenge, because patients with similar clinical characteristics vary considerably in treatment response and prognosis, demanding new novel markers for subgroup categorization, especially those based on non-invasive or minimally invasive approaches.

Metabolomics is an emerging field providing a powerful tool for discovering new biomarkers (4). Analyzing metabolites in biological fluids or tissue samples provides information on the metabolic phenotype that is related to cancer diagnosis and prognosis at the individual level and to cancer risk at the population level (5–7). In recent years, growing evidence suggests that metabolic profiles may serve as diagnostic and prognostic markers for various types of cancer (8–13). Previous studies on EC have explored metabolic differences between patients and healthy controls in serum (14), plasma (15) and urine samples (16), and they have identified potentially useful metabolite markers for screening and monitoring treatment of patients with ESCC (17) and EAC (18). A recent study found that serum metabolite profiles (valine, γ-aminobutyric acid, and pyrrole-2-carboxylic acid) were capable of predicting lymph node metastasis of ESCC (19), indicating that metabolomic signatures were potential prognostic biomarkers for EC. Nevertheless, few studies were conducted to explore potential metabolite markers for predicting the prognosis of EAC.

In our recent study (20) we performed a global metabolic profiling of serum samples from EAC patients and controls, followed by a targeted validation. The study revealed that three metabolites, carbohydrate d-mannose, amino acid l-proline (LP) and ketone body 3-hydroxybutyrate (BHBA), were consistently aberrant in patients with EAC, indicating their potential utility as risk or prognostic indicators. Therefore, we designed the current study to analyze the potential associations of these three metabolites with EAC recurrence and overall survival.

Materials and methods

Study population

To identify metabolites associated with EAC recurrence and overall survival, we conducted this prospective cohort study in Caucasian patients with newly diagnosed and histologically confirmed EAC who were recruited at The University of Texas MD Anderson Cancer Center (MDACC) between January 2004 and December 2014. Two criteria were used to select patients for this study: (1) the patient did not have distant metastasis at initial presentation, and (2) the patient was treated with curative-intent surgical resection. A total of 159 patients who fulfilled the two criteria were included in the analysis. TNM staging was determined by the sixth edition of American Joint Committee on cancer. Stage I and II were classified as early stage, and stage III and IVa were classified as advanced stage. Smoking status and body mass index (BMI) were obtained from the baseline questionnaire during initial visit to MDACC. BMI is classified as non-obese (<30 kg/m²) and obese (≥30 kg/m²) according to the standard classification of the World Health Organization. Clinical and follow-up data were abstracted by medical chart review. The median follow-up period for this cohort was 39.3 months (range: 7.6–123.3 months). This study was approved by the Institutional Review Board, and all study participants signed informed consent.

Metabolite profiling

Whole blood samples were collected before treatment and immediately separated into serum, which was stored at liquid nitrogen tanks until the sample was assayed. In our previous study (20), we performed a global metabolic profiling using Metabolon Inc. (Durham, NC) followed by a targeted metabolite analysis using validated liquid chromatography–mass spectrometry/mass spectrometry (LC-MS/MS) methods at Texas Southern University (Houston, TX). The results indicated that d-mannose, LP, and BHBA were the metabolites showing significant trend in changed levels from normal individuals to early-stage and late-stage EAC cases, suggesting their utility as early-detection markers. Sample extraction, metabolite assays, quality control, and validation of LC-MS/MS assays were performed as described previously (21). The LC-MS/MS assays met the validation requirements of Food and Drug Administration.

Statistical analysis

Serum concentrations of metabolites were divided into three groups based on the tertile cut-off points. Overall survival was defined as the time in months between date of diagnosis and date of death or date of last follow-up. Recurrence-free survival was defined as the time in months from date of diagnosis to date of to the first event of either recurrence or death. Overall survival and recurrence-free survival were compared between patients with different metabolite levels by the Kaplan–Meier method, and the significance was assessed by log-rank test. Hazard ratio (HR) and 95% confidence interval (CI) were analyzed by a multivariable Cox proportional hazards model, adjusted for age (continuous), gender (male or female), tumor grade (well, moderately, or poorly), clinical stage (I, II or III) and neoadjuvant chemoradiotherapy (yes or no). The proportional hazards assumption was checked to be valid by inspecting the log-log survival curves visually and testing the significance of the interaction term between time and the covariate. Correlations between metabolites and clinicopathologic variables were obtained by using Wilcoxon rank sum test. Interaction between BMI and serum d-mannose level on risk of death from EAC was evaluated by introducing the product of BMI and d-mannose level in the multivariable Cox regression model. Statistical analyses were performed using Stata 14.0 (Stata Corp, College Station, TX). All statistical tests were two-sided, and P < 0.05 was considered as statistically significant.

Results

Patient characteristics

The cohort consisted of 159 patients, with a median age of 61 years (range: 27–83 years). Among these cases, 57 (35.8%) patients experienced tumor recurrence and 65 (40.9%) patients died by the end of follow-up. Most of the patients (91.2%) were male, consistent with male predominance in EAC. The majority of the patients were ever smokers (74.8%) and had moderate or poorly differentiated tumors (68.6%). The distribution of patients by clinical stage was: I 24.5%, II 35.2%, III 36.5% and IVa 3.8%. All patients were treated by surgical resection, of which 112 (70.4%) patients underwent neoadjuvant chemoradiotherapy. Patients’ clinicopathologic characteristics are presented in Table 1.

Table 1.

Patient characteristics and association of d-mannose levels with selected characteristics

Characteristics	N (%)	d-mannose Mediana (range)	P valueb
Gender
Male	145 (91.2)	8.38 (1.97–20.60)
Female	14 (8.8)	8.68 (2.99–16.63)	0.92
Age
<60	78 (49.1)	8.12 (2.85–16.73)
≥60	81 (50.9)	8.88 (1.97–20.60)	0.13
Smoking status
Never	40 (25.2)	9.36 (4.00–16.63)
Ever	119 (74.8)	8.38 (1.97–20.60)	0.43
BMI (kg/m2)
<30	99 (62.3)	8.33 (1.97–19.23)
≥30	52 (32.7)	9.36 (2.85–20.60)	0.10
Depth of invasion
T1+T2	57 (35.8)	8.13 (1.97–16.73)
T3+T4	102 (64.2)	8.88 (2.85–20.60)	0.17
Lymph node involvement
No	84 (52.8)	8.38 (1.97–16.73)
Yes	75 (47.2)	8.61 (2.99–20.60)	0.43
Stage
Early-stage	95 (59.7)	8.32 (1.97–16.73)
Advanced-stage	64 (40.3)	9.34 (2.99–20.60)	0.19
Grade
Well	49 (31.4)	8.38 (4.00–20.60)
Moderately	33 (21.2)	8.87 (1.97–16.63)	0.54
Poorly	74 (47.4)	8.38 (2.85–20.07)
Recurrence
No	101 (63.5)	8.71 (1.97–20.07)
Yes	57 (35.8)	7.40 (2.85–20.60)	0.06
Vital status
Alive	93 (58.5)	9.16 (1.97–20.60)
Dead	65 (40.8)	7.40 (2.85–20.07)	0.03

Significant P values are indicated in bold font.

^aAbsolute median values in µg/ml.

^bWilcoxon rank sum test.

Metabolites as predictors of EAC survival

In Cox regression models, levels of d-mannose were significantly associated with overall survival, whereas levels of LP and BHBA were not, after adjusting for age, gender, tumor grade, clinical stage and neoadjuvant chemoradiotherapy. The levels of these three metabolites showed weak correlations (BHBA versus d-mannose: Spearman rho = 0.37, P < 0.01; LP versus d-mannose: Spearman rho = –0.25, P < 0.01; BHBA versus LP: Spearman rho = –0.43, P < 0.01). Table 2 presented the HRs for death categorized by tertiles of serum metabolites levels in the overall study population. Using patients with the first tertile (<7.35 µg/ml) of d-mannose as reference, those with the second (7.35–9.75 µg/ml) and third (>9.75 µg/ml) tertiles of d-mannose exhibited very similar risk of death, with HRs of 0.44 (95% CI: 0.23–0.84, P = 0.01) and 0.44 (95% CI: 0.23–0.86, P = 0.02), respectively. Therefore, the combination of the second and third tertiles was considered, and then d-mannose levels were divided into two groups with high (second plus third tertiles) and low (first tertile) levels. As shown in Table 2, patients with a high level of d-mannose exhibited a significantly reduced risk of death in comparison with those with a low level, with an adjusted HR of 0.44 (95% CI: 0.25–0.77, P < 0.01). The 5-year survival rates were 66.4% and 44.6% for patients with high and low levels of d-mannose, respectively. In Kaplan–Meier curve analysis, patients with a high d-mannose level had significantly longer median overall survival time than those with a low level (>123.4 versus 36.9 months, long rank P = 0.01; Figure 1A).

Table 2.

Metabolites (in tertiles) as predictors of recurrence and overall survival

Metabolites	Levels	Recurrence		HR (95% CI) a	P valuea	Survival		HR (95% CI) a	P value a
		Yes, n (%)	No, n (%)			Dead, n (%)	Alive, n (%)
d-mannose	First tertile (low)	28 (51.85)	26 (48.15)	1		31 (57.41)	23 (42.59)	1
	Second tertile	12 (23.53)	39 (76.47)	0.45 (0.22–0.93)	0.03	16 (31.37)	35 (68.63)	0.44 (0.23–0.84)	0.01
	Third tertile	17 (32.08)	36 (67.92)	0.58 (0.29–1.12)	0.11	18 (33.96)	35 (66.04)	0.44 (0.23–0.86)	0.02
	Second plus third tertiles (high)	29 (27.88)	75 (72.12)	0.51 (0.29–0.91)	0.02	34 (32.69)	70 (67.31)	0.44 (0.25–0.77)	< 0.01
LP	First tertile (low)	19 (35.19)	35 (64.81)	1		21 (38.89)	33 (61.11)	1
	Second tertile	20 (38.46)	32 (61.54)	0.95 (0.49–1.84)	0.87	25 (48.08)	27 (51.92)	1.05 (0.56–1.96)	0.88
	Third tertile	17 (32.69)	35 (67.31)	0.92 (0.46–1.82)	0.80	18 (34.62)	34 (65.38)	0.95 (0.49–1.84)	0.87
	Second plus third tertiles (high)	37 (35.58)	67 (64.42)	0.93 (0.52–1.66)	0.81	43 (41.35)	61(58.65)	1.00 (0.58–1.74)	1.00
BHBA	First tertile (low)	19 (36.54)	33 (63.46)	1		21 (40.38)	31 (59.62)	1
	Second tertile	15 (28.85)	37 (71.15)	0.61 (0.30–1.26)	0.18	18 (34.62)	34 (65.38)	0.69 (0.35–1.37)	0.29
	Third tertile	21 (40.38)	31 (59.62)	0.94 (0.49–1.78)	0.84	24 (46.15)	28 (53.85)	1.00 (0.54–1.86)	0.99
	Second plus third tertiles (high)	36 (34.62)	68 (65.38)	0.78 (0.44–1.38)	0.39	42 (40.38)	62 (59.62)	0.85 (0.49–1.48)	0.57

Abbreviations: LP, L-proline; BHBA, 3-hydroxybutyrate; HR, hazard ratio; CI, confidence interval.

Significant P values are indicated in bold font.

^aCox proportional hazards regression model, adjusted for age, gender, tumor grade, clinical stage and neoadjuvant chemoradiotherapy.

Cut off levels in the tertiles: d-mannose (first: <7.35, second: 7.35–9.75, third: >9.75 µg/ml); LP (first: <20.60, second: 20.60–25.60, third: >25.60 µg/ml); BHBA (first: <5.04, second: 5.04–17.40, third: >17.40 µg/ml).

Figure 1.

Kaplan–Meier curves for overall survival (A) and recurrence-free survival (B) grouped by low (first tertile) and high (second plus third tertiles) levels of serum d-mannose. N = number of patients with an event (death or recurrence)/total number of patients in the dataset. MOS, median overall survival time. MRFS, median recurrence-free survival time.

d-mannose as a predictor of EAC recurrence

EAC recurrence occurred in 27.9% of patients with high (second plus third tertiles) level of d-mannose and in 51.9% of those with low (first tertile) level. Patients with a high level of d-mannose had a significantly reduced risk of recurrence compared with those with a low level (HR: 0.51, 95% CI: 0.29–0.91, P = 0.02; Table 2), after adjusting for age, gender, tumor grade, clinical stage, and neoadjuvant chemoradiotherapy. The 5-year recurrence-free survival rate was 69.8% for patients with a high level of d-mannose, compared with 46.6% for patients with a low level. Kaplan–Meier curve analysis showed that patients with a high d-mannose level had significantly longer median recurrence-free survival time than those with a low level (>123.4 versus 32.9 months, long rank P < 0.01) (Figure 1B). Levels of LP and BHBA were not associated with risk of recurrence (Table 2).

d-mannose and overall survival stratified by stage and obesity

The median levels of d-mannose in patients with advanced stage were higher than those with early stage (9.34 versus 8.32 µg/ml), although the difference was not significant (P = 0.19). Thus, we analyzed the association of d-mannose levels with survival among patients with early stage and advanced stage. As shown in Table 3, compared with patients with a low (first tertile) level of d-mannose, those with a high (second plus third tertiles) level of d-mannose had HRs of 0.40 (95% CI: 0.16–0.98, P = 0.05) and 0.45 (95% CI: 0.21–0.92, P = 0.03) among early-stage and advanced-stage EAC, respectively. The results were consistent with those from the whole cohort.

Table 3.

d-mannose and overall survival stratified by stage and obesity

Characteristics	Levels of d-mannose	Dead, n (%)	Alive, n (%)	HR (95% CI)	P value	P-int
Stage						0.92
Early stage	Low (first tertile)	13 (40.63)	19 (59.38)
	High (second plus third tertiles)	15 (24.19)	47 (75.81)	0.40 (0.16–0.98) a	0.05a
Advanced stage	Low (first tertile)	18 (81.82)	4 (18.18)
	High (second plus third tertiles)	19 (45.24)	23 (54.76)	0.45 (0.21–0.92) a	0.03a
BMI						0.98
<30	Low (first tertile)	23 (58.97)	16 (41.03)
	High (second plus third tertiles)	22 (36.67)	38 (63.33)	0.46 (0.22–0.93) b	0.03b
≥30	Low (first tertile)	7 (50.00)	7 (50.00)
	High (second plus third tertiles)	8 (21.05)	30 (78.95)	0.78 (0.18–3.37) b	0.74b

Abbreviations: HR, hazard ratio; CI, confidence interval; P-int, P for interaction.

Significant P values are indicated in bold font.

^aCox proportional hazards regression model, adjusted for age, gender, tumor grade and neoadjuvant chemoradiotherapy.

^bCox proportional hazards regression model, adjusted for age, gender, tumor grade, clinical stage and neoadjuvant chemoradiotherapy.

We evaluated the association of serum metabolite levels with several selected patient characteristics. The results indicated that levels of d-mannose were not associated with patient age, gender, smoking status, depth of invasion, lymph node involvement, clinical stage, or tumor grade, as presented in Table 1, and the associations of LP and BHBA levels with these factors were presented in Supplementary Table S1, available at Carcinogenesis Online. There was some indication that levels of d-mannose were higher in obese patients (BMI ≥ 30) than in non-obese patients (BMI < 30), but the result was only borderline significant (P = 0.10). We analyzed the HRs for death stratified by obesity. As seen in Table 3, the adjusted HR for death among non-obese patients with a high level of d-mannose was 0.46 (95% CI: 0.22–0.93, P = 0.03) in comparison with those with a low level of d-mannose. The association with death risk in obese patients was not statistically significant (HR = 0.78; 95% CI: 0.18–3.37, P = 0.74). However, there was not a significant interaction between d-mannose levels and BMI on risk of death from EAC (P for interaction = 0.98).

Discussion

Metabolomic profiling has emerged as a promising approach to identify new biomarkers for cancer progression (22) and prognosis (23). Nonetheless, the prognostic value of circulating metabolites in EAC has not been well explored. Here, we quantified serum d-mannose, LP and BHBA using LC-MS/MS methods and investigated their association with recurrence and overall survival in EAC patients, which revealed that elevated levels of d-mannose might have potential to predict improved prognosis. Compared with patients with a low (first tertile) level of d-mannose, those with a high (second plus third tertiles) level had 49% reduction in risk of recurrence and 56% reduction in risk of death. To our knowledge, this is the first evidence that serum d-mannose can predict prognosis in EAC.

The relationship between d-mannose and cancer has not yet been well documented. Mannose is a monosaccharide, playing a pivotal role in human metabolism, especially in the glycosylation of proteins (24). Both exogenous mannose and glucose contribute to produce mannose-6-phosphate, which is subsequently converted to mannose-1-phosphate and Guanosine diphosphate mannose for the process of N-linked glycosylation (25). Generating mannose from glucose relies on the isomerization of fructose-6-phosphate to Man-6-P, regulated by phosphomannose isomerase in the cells (26). Previous research has mostly focused on d-mannose for the treatment of glycosylation deficiency caused by genetic defects in phosphomannose isomerase (27) and the prevention of urinary tract infections by blocking the adhesion of Escherichia coli to uroepithelial cells (28). We previously reported that levels of serum d-mannose were increased among patients with EAC compared with healthy individuals, and patients with late-stage (stages III and IV) had higher levels of d-mannose than those with early-stage (stages I and II) (20). These results were in line with the previously observed increased concentration of mannose in metastatic breast cancer individuals compared with localized early disease (29). This phenomenon could be a reflection of metabolic changes to support the aberrant anabolic requirements of cancer cells (30,31) and incompletion of the glycosylation process (32), rather than a predisposing factor.

It is paradoxical that patients with late-stage disease had higher levels of d-mannose than those with early-stage, but high levels of d-mannose were associated with better survival. In the present study, patients with stage IVb were excluded to minimize the effects of metastasis on patient outcome, and the results suggested that levels of d-mannose in patients with advanced stage (stages III and IVa) were still higher than those with early stage, although the difference was not significant (P = 0.19). Since clinical stage is an established prognostic indicator for EAC, we evaluated the association of d-mannose levels with overall survival stratified by stage. The results showed that patients with high level of d-mannose were associated with better survival in both early and advanced stages compared with those with low level of d-mannose. Hence, the prognostic value of d-mannose was independent of clinical stage.

The mechanisms underlying the association between d-mannose and EAC prognosis remains unclear. It is worth noting that we collected the blood samples prior to chemotherapy or radiotherapy, limiting the metabolic changes induced by these treatments as well as treatment-related complications. Moreover, all enrolled patients were treated with surgery, and we adjusted for neoadjuvant chemoradiotherapy in Cox regression analysis, so it is unlikely that the therapy strategies affected the results. Our data highlighted that EAC patients with the lowest level (first tertile) of d-mannose were associated with poor overall survival. One potential explanation for this association was the concentration of mannose that we measured in serum by LC-MS/MS method included both free mannose and that binding to mannose-binding lectin (MBL), which is an important molecule of the innate immune system. Low preoperative MBL levels in serum of cancer patients were predictive of postoperative infection (33,34), which was significantly associated with poor survival (35). Furthermore, low MBL concentrations were significantly associated with higher risks of severe infection in patients with cancer who were undergoing chemotherapy (36,37). Additional studies are necessary to separately evaluate the associations of free mannose and MBL with the prognosis of EAC, as well as whether d-mannose has prognostic significance in other types of cancer.

Our study did have some limitations. First, only three serum metabolites were explored in our study, and other prognostic metabolic markers may be missed. Future metabolomic profiles generated from a comprehensive study will provide a more robust way to identify clinically significant prognostic markers in EAC. Additionally, the serum samples were collected after diagnosis prior to treatment, and the duration of underlying EAC before diagnosis may affect the metabolite levels. However, it remains a challenge to estimate exactly the variations in duration of underlying, undiagnosed EAC in the clinical setting. Clinical stage might be a surrogate for the duration and was adjusted in Cox regression analysis in our study.

In summary, we demonstrated that assessing the serum level of d-mannose may be potentially used as to identify EAC patients at high risk of recurrence and mortality, which in turn, can help to optimize treatment strategies and clinical decision-making. Prospective validation in independent cohorts is warranted to confirm these findings, and further biological investigation is needed to clarify the underlying mechanisms.

Supplementary materials

Supplementary data are available at Carcinogenesis online.

Funding

Supported by Center for Translational and Public Health Genomics, Duncan Family Institute, MD Anderson Cancer Center.

Conflict of Interest Statement: None declared.

Author notes

*To whom correspondence should be addressed. Tel: [phone]; Fax: [phone]; Email: .