Differences in Metabolomic Profiles by Birthplace in Mexican-Origin Hispanic Men Who Participated in a Weight Loss Lifestyle Intervention

Abstract

Birthplace, as a proxy for environmental exposures (e.g., diet), may influence metabolomic profiles and influence risk of cancer. This secondary analysis investigated metabolomic profile differences between foreign and U.S.-born Mexican-origin (MO) Hispanic men to shed light on potential mechanisms through which foreign- and U.S.-born individuals experience differences in cancer risk and risk factors. Plasma samples from MO Hispanic men (N = 42) who participated in a previous lifestyle intervention were collected pre-and post-intervention. Metabolomic profiles were characterized from samples using ultra performance liquid chromatography–quadrupole time of flight mass spectrometry (UPLC-QTOF). Models were visualized using supervised orthogonal projections to latent structures–discriminant analysis (OPLS-DA). Progenesis QI was used for peak integration and metabolite identification. Plasma metabolomic profiles differed between foreign- and U.S.-born pre-intervention (R2 = .65) and post-intervention (R2 = .62). Metabolomic profiles differed pre- versus post-intervention (R2 = .35 and R2 = .65) for the foreign- and U.S.-born group, respectively. Both endogenous metabolites and dietary components characterized differences between foreign- and U.S.-born participants pre- and post-intervention. Plasma metabolomic profiles from MO Hispanic men differed by birthplace. These results advance our understanding of relevant exposures that may affect cancer risk among MO Hispanic men born abroad or in the United States.

Introduction

Obesity rates in the United States are among the highest for Hispanic men, with 45.7% considered obese (Hales et al., 2020). This places Hispanic men at elevated risk of obesity-related cancers (“Centers for Disease Control and Prevention. Cancers Associated with Overweight and Obesity Make up 40 Percent of Cancers Diagnosed in the United States,” 2017). Differences in cancer-related risk factors (e.g., obesity, type 2 diabetes) are higher for U.S.-born compared with foreign-born Hispanic men, and this risk increases with greater residency in the United States (Afable-Munsuz et al., 2013; Anderson et al., 2016; Lee et al., 2013; Lopez et al., 2014). In addition, foreign-born Hispanic men have substantially lower odds of being overweight and obese compared with those born in the United States (Akresh, 2008). In an aging Mexican-origin population, Afable-Munsuz et al. reported that relative to first-generation adults (foreign-born adults), second- and third-generation adults (U.S.-born) had 1.8 higher odds of having type 2 diabetes, respectively (Afable-Munsuz et al., 2013). Anderson et al. reported that type 2 diabetes risk among individuals born in Mexico increased with longer residency in the United States (Afable-Munsuz et al., 2013; Anderson et al., 2016). A study published by Flores et al. examined the odds of several liver cancer risk factors in a sample of foreign and U.S.-born Mexican Americans (Flores et al., 2018). Compared with Mexican men living in Mexico, foreign-born and U.S.-born Mexican American men residing in the United States had 2.1 and 3.9 greater odds for obesity, 3.3 and 5.4 greater odds for abdominal obesity, and 1.7 and 3.0 greater odds for diabetes, respectively. Importantly, type 2 diabetes is an established risk factor for cancer (Scully et al., 2020), suggesting that birthplace or time in U.S. residence may drive cancer risk in Hispanic males (Haile et al., 2012). These data highlight the need to investigate the effects of birthplace and associated exposures on biological systems that may drive these relationships among Hispanic men.

One such mechanism driving obesity and obesity-related cancer risk may be metabolism and its characteristic metabolome. As a rapidly advancing area of “omics” science, metabolomics is the systematic study of small-molecule metabolites (with an atomic mass less than 1.5 kDA) in living systems, defined as the “metabolome” (Wishart et al., 2007). Previous research studies have utilized metabolomic analysis as a tool to identify signatures of overweight and obesity. Findings from such studies have reported changes in fatty acid and amino acid concentrations among overweight and obese adults (Rangel-Huerta et al., 2019). Higher levels of stearic acid, lower levels of oleic acid, and changes in critical compounds of fatty acid (FA) synthesis and oxidation (carnitine, propionyl-, butyryl-, and hexanoyl-carnitine) have been observed in the metabolomic profiles of overweight and obese men (Kim et al., 2010). Studies have reported an association between higher concentrations of branched-chain amino acids (BCAA) and essential amino acids (AA) among overweight and obese men compared with lean men (Kim et al., 2010; Xie et al., 2014). Signatures in individuals with obesity are consistently marked by dramatically higher concentrations of phenylalanine, alanine, valine, leucine/isoleucine, tyrosine, glutamate/glutamine, aspartate/asparagine, and arginine (Bagheri et al., 2019; Kim et al., 2010; Newgard et al., 2009; Xie et al., 2014).

To our knowledge, no studies have evaluated metabolomic profiles of Hispanic men with overweight and obesity born in the United States or abroad. The primary purpose of this study was to investigate whether plasma metabolomic profiles differ between foreign versus U.S.-born Hispanic males. Given that participants underwent a lifestyle intervention, metabolomic profiles were also compared after receiving the intervention for foreign and U.S. born. Identifying metabolomic signatures in Hispanic men by birthplace and in response to a lifestyle intervention holds promise to tailor future cancer prevention interventions and address the cancer disparities observed by nativity status among this specific population.

Materials and Method

Study Sample

The protocol, research design, and results of the ANIMO (Spanish term for motivation or encouragement) pilot study have been published elsewhere (Garcia et al., 2018, 2019). Briefly, the 24-week randomized controlled trial investigated the feasibility and preliminary efficacy of gender- and culturally relevant weight loss intervention in Hispanic men compared with a wait-list control (WLC) group. The ANIMO pilot study was a 12-week randomized controlled trial with a 12-week follow-up period conducted July 2016 to February 2017. The intervention protocol included 12 weeks of individual counseling for lifestyle modification (e.g., diet and physical activity) and biweekly phone calls across a 12-week follow-up period. Eligibility criteria included (1) self-identifying as a Hispanic male; (2) being 18 to 64 years of age; (3) having a body mass index (BMI) of 25 to 50 kg/m²; (4) providing informed consent and completing a health risk assessment before participation; and (5) speaking, reading, and writing English or Spanish. Given the original design of the study includes a waitlist control group who received the intervention after the intervention group had undergone the study protocol, in the current analysis, we refrain from labeling periods by time points but instead, we defined them according to intervention status. In other words, the preintervention period is the point in time before participants underwent the intervention protocol, while the post-intervention period was after participants had received the intervention. Participants were then grouped into two different categories based on their birthplace. Participants who self-reported as first-generation immigrants were categorized as foreign-born participants (n = 19), with those who reported being second-generation or beyond categorized as U.S.-born (n = 23). Foreign-born participants were mostly of Mexican origin, with the only exception being a participant who was of Puerto Rican descent. Exploratory analyses were conducted based on acculturation status as measured by the Acculturation Rating Scale for Mexican Americans-II (ARSMA-II), a measure of acculturation based on language, ethnic identity, and ethnic interaction that has been validated among this population (Cuellar et al., 1995).

Metabolomic Analysis

Plasma Processing

Stored plasma samples that were available from the ANIMO weight loss pilot study pre- and post-intervention were prepared for metabolomic profiling by methanol precipitation as previously described with minor modifications (Want et al., 2010). First, 1 mL ice-cold methanol was added to 200 µL of the sample. Plasma was then vortexed at 4°C and centrifuged at 17,600 relative centrifugal force (RCF) for 10 min at 4^oC, and 100 μL of supernatant was transferred to a clean tube. The supernatant was dried in a SpeedVac Vacuum Concentrator, and sample residues were reconstituted in 100 µL deionized water for ultra performance liquid chromatography-quadrupole time of flight mass spectrometry (UPLC-QTOF) analysis.

UPLC-QTOF of Plasma

Plasma metabolite extracts (5 μL) were injected onto a 2.1 × 100 mm (1.7 μm) Acquity UPLC HSS T3 column (Waters Corporation, Milford, MA) and eluted using optimized gradients (A = water, 0.1% formic acid; B = acetonitrile, 0.1% formic acid). Samples were analyzed using a UPLC system (UPLC Acquity, Waters MS Technologies Corporation, Manchester, United Kingdom) coupled online to a Xevo G2S Q-TOF (Waters MS Technologies Corporation, Manchester, United Kingdom). Samples were analyzed for metabolomics in positive and negative electrospray modes, with a scan range of 50 to 1,000 m/z. Fragmentation information was generated using MSE experiments collected at both low (5V) and high (50V) collision energies. Quality control (QC) samples (pooled plasma) were injected at the start of the analyses for column conditioning and at multiple points during a single analytical run. Metabolomic profile alignment, peak detection, and integration were achieved using Progenesis QI software (Nonlinear Dynamics, Durham, North Carolina).

Statistical Analysis

Given the small sample size, nonparametric tests were used to test baseline differences in demographic and clinical characteristics between the groups. Wilcoxon rank-sum test was used to calculate p values for differences in median values of continuous variables including age, BMI, physical activity, and caloric intake. Fisher’s exact tests were used for categorical variables such as education, language preference, BMI category, smoking status, alcohol intake, cancer history, cardiovascular disease history, and type 2 diabetes status.

Multivariate statistical analyses were carried out by SIMCA 16.0 (Umetrics, Sweden) for metabolomics data analysis. All variables were imported into SIMCA. First, unsupervised principal component analysis was utilized to visualize the intrinsic variance of metabolites and to confirm the reliability of the QC samples. Then, supervised orthogonal projections to latent structures discriminant analysis (OPLS-DA) were then used to extract maximum information from the dataset and to isolate the metabolites responsible for differences among foreign-born and U.S.-born participants. Models were visualized using supervised OPLS-DA; model quality was described using R² cut points of 0.75, 0.50, and 0.25 for substantial, moderate, or weak levels of predictive accuracy, respectively (Hair et al., 2011). Score plots (S-plots), which provided the visualization of metabolites influence in the OPLS-DA model, were created and served to select the top putative significant metabolites. Accurate m/z values, retention times, isotopic patterns, full MS, and MS/MS fragmentation spectra were acquired for each metabolite to facilitate identification. The Progenesis QI software’s in-house library was used for peak integration and metabolite identification in addition to the Human Metabolite Data Base (HMDB).

Results

Participant’s Characteristics

Significant differences were observed between the foreign and U.S.-born groups for several demographic characteristics at baseline. Foreign-born participants were 12 years older (Z = 2.68, p = .007), reported less formal education (p = .002), were mostly Spanish-preferred language speakers (p < .001), and reported lower acculturation (p < .001) compared with U.S.-born participants (Table 1). There were no significant differences in several cancer risk factors including BMI, smoking, alcohol intake, or history of cancer. There was no difference in type 2 diabetes status or history of cardiovascular disease.

Table 1.

Baseline Characteristics of Participants From the ANIMO Pilot Study by Birthplace (N = 42).

Characteristic	Foreign-born n= 19 (45%)	U.S.-born n = 23 (55%)	P values
Demographics
Age, years (median, IQR)	47 (23, 61)	35 (22, 63)	.007
Education, n (%)
High school/GED or less	15 (79.0%)	6 (26.1%)	.002
Some college/associate degree	2 (15.8%)	12 (52.2%)
College graduate and beyond	2 (5.2%)	5 (21.7%)
Preferred language, n (%)			<.001
English	0 (0.0%)	17 (73.9%)
Spanish	18 (94.7%)	5 (21.7%)
Other	1 (5.3%)	1 (4.4%)
Acculturation			<.001
No acculturation	14 (73.7%)	1 (4.4%)
Some level of acculturation	5 (26.3%)	22 (95.6%)
Clinical characteristics
BMI at baseline(kg/m2; median, IQR)	32.6 (25.0, 42.2)	34.8 (26.2, 40.6)	.44
BMI status, n (%)
Overweight (25–29.9 kg/m2)	3 (15.8%)	2 (8.7%)	.644
Obese (≥30 kg/m2)	16 (84.2%)	21 (91.3%)
Smoking status, n (%)			1.000
Nonsmoker in the past year	16 (84.2%)	19 (82.6%)
Current smoker	3 (15.8%)	4 (17.4%)
Alcohol intake, n (%)			.489
Never/past	5 (26.3%)	7 (30.4%)
<1 drink per week	12 (63.2%)	11 (47.8%)
>1 drink per week	2 (10.5%)	5 (21.8%)
Physical activity
Leisure (min/week, median, IQR)	0 (0, 360)	0 (0,600)	.19
Cancer history, n (%)
Yes	0 (0.0%)	2 (8.7%)	.493
No	19 (100.0%)	21 (91.3%)
CVD history, n (%)
Yes	0 (0.0%)	0 (0.0%)	1.000
No	19 (100.0%)	23 (100.0%)
Diabetes, n (%)
Yes	1 (5.3%)	2 (8.7%)	1.000
No	18 (94.7%)	21 (91.3%)
Diet (Median, IQR)
Energy intake at baseline (kcals/day)	2,785.2 (510.6, 6211.7)	2,424.5 (466.9, 5574.7)	.62

Note. IQR = interquartile range; BMI = body mass index; CVD = cardiovascular disease; GED = general educational development.

^aExcluded for implausible energy intake; foreign born (n = 2), US born (n = 2).

Analytical Performance of Plasma Metabolites

There were 8,504 metabolite features in positive ion mode and 5,217 in negative ion mode. The coefficient of variation (CV) for each metabolite feature was calculated using pooled quality control. Metabolite features with a CV > 30% were excluded leaving 1,327 metabolites for the analysis in positive mode and 2,376 in negative ion mode. The analysis was conducted with the metabolite features in positive ion mode only given that all models describing differences in metabolomic profiles were very weak in negative ion mode (R² < .25 for all models).

Nontargeted Positive Ion Mode Metabolic Pattern Analysis

Because participants underwent a weight loss intervention, OPLS-DAs were conducted to compare metabolite clustering for individuals who achieve clinically meaningful weight loss (5% of body weight; Williamson et al., 2015) to those who did not. Results showed models had weak levels of predictive accuracy (R² = .34) suggesting no significant clustering between the groups (data not shown). Next, metabolic profiles were compared across groups by intervention status and birthplace. Stratification of study sample is summarized in Figure 1.

Figure 1.

Stratification of Sample for Metabolomics Profile Comparison Among Participants From the ANIMO Pilot Study (N = 42)

OPLS-DAs were conducted to compare metabolite clustering for foreign-born versus U.S.-born participants pre-intervention (n = 40) and post-intervention (n = 38; Figure 2). Metabolomic profiles between foreign- and U.S.-born groups were found to be moderately different pre-intervention (R² = .65; Figure 2A) and post-intervention (R² = .62; Figure 2C).

Figure 2.

OPLS-DA and S-Plots for Foreign-Born versus U.S.-Born Pre-Intervention and Post-Intervention. Multivariate Statistical Analysis From Ultra Performance Liquid Chromatography-Quadrupole Time of Flight Mass Spectrometry Metabolic Profiling. (A) OPLS-DA Score of Foreign-Born (Green) vs. U.S.-Born (Blue) at Pre-Intervention; (B) S-Plot With Metabolites That Were Examined (Circled in Red); (C) OPLS-DA Score of Foreign-Born (Green) vs. U.S.-Born (Blue) at Post-Intervention; (D) S-Plot With Metabolites That Were Examined (Circled in Red) in C

There were 33 metabolite features that were selected from the S-plots and that were associated with birthplace prior to the intervention, of which eight had quality peak shape. Table 2 lists the putative IDs that were obtained for four metabolite features that were exclusively associated with the foreign-born group and four metabolite features exclusively associated with the U.S.-born group pre-intervention. Endogenous metabolites, such as alpha-keto and amino acids, phospholipids, fatty acids, as well as some dietary compounds, were putative IDs for metabolites characteristic of the foreign-born group during the pre-intervention period, independent of the study arm.

Table 2.

Plasma Metabolites Associated With Birthplace in the ANIMO Study Pre-Intervention (N = 42).

Retention time	M/Z	Putative IDs or Class	Source
Foreign-born associated metabolites
3.96	102.9840	3-Mercaptopyruvic acid	Alpha keto acid
3.96	117.9877	(2Z)-2-[(sulfooxy)methyl]but-2-enoic acid 3-methyl-4-(sulfooxy)but-2-enoic acid	Sulfated FA phenylsulfate Sulfated FA lipid
3.96a,b	120.0011	Maleylacetoacetic acid 2,4,5-trihydroxy-3-methoxybenzoic acid (-)-1-(Methylthio)propyl 1-propenyl disulfide / (-)-1-(Methylthio)propyl 1-propenyl disulfide 2-(2H-1,3-benzodioxol-5-yl)-2-oxoacetic acid	Tyrosine metabolism intermediate Lipid metabolism and transport/ Diet Diet (green vegetables) Organoheterocyclic compounds/ Human gut microbiota
3.96	165.9827	Urolithin b sulfate 1,2,4-Trithiolane Methylarsonite	Organic compound from amino acid phenylalanine/ Diet (red wine, berries fruits and nuts) Diet (Green vegetables and mushrooms) Organic compound (arsenic detox pathway)
U.S.-born associated metabolites
0.64c	169.9827	Taurine	Diet (animal and fish protein)
0.79	170.9705	Oxalacetic acid Ethyl hydrogen sulfate	Diet (soy products) Diet (alcoholic beverages, wine)
1.47	726.6804	Pipercitine DG(i-22:0/i-20:0/0:0)	Diet (herbs and spices) Diacylglycerols
2.25	230.9484	Gentisic acid 2,4-Dihydroxybenzoic acid	Diet (thyme, basil, peas and carrots) Diet (avocado, beer, wine and coffee)

^aAssociated with foreign-born group pre-intervention (Table 4). ^bAssociated with U.S.-born group pre-intervention (Table 5). ^cAssociated with foreign-born group post-intervention (Table 4).

Given that these participants had participated in a lifestyle intervention, we sought to understand whether differences in the metabolome by birthplace remained at the post-intervention time point. There were 17 metabolite features that were selected from the S-plots and that were associated with birthplace post-intervention of which seven had quality peak shape. Table 3 lists the putative IDs that were obtained for four metabolite features that were exclusively associated with the foreign-born group and three exclusively associated with the U.S.-born group post-intervention. Diet-related metabolites, such as tea, fruits, poultry, pork, cereal products, and soy products, were some of the putative IDs for metabolites characteristic of the foreign-born group during the post-intervention period. Most of the metabolites identified for the U.S.-born group, pre- and post-intervention, had putative IDs related to dietary sources, including animal and fish protein, soy products, alcoholic beverages, some herbs such as thyme and basil, peas, carrots, and coffee characteristic of the pre-intervention period. Potential IDs for metabolites characteristic of the U.S.-born group at the post-intervention time point included dietary components from onions, fruits, herbs and spices, and some potential lipid-related metabolites.

Table 3.

Plasma Metabolites Associated With Birthplace in the ANIMO Study Post-Intervention (N = 42).

Retention time	M/Z	Putative IDs or class	Source
Foreign-born associated metabolites
0.60a	233.9103	Ginsenoside M7cd Notoginsenoside R9	Diet (tea) Diet (tea)
0.78	102.0381	4,5,-Dihydro-2-mehtylthiazole Coniferyl alcohol 3-Methoxybenzenepropanoic acid	Thiazolines Diet (sages, chestnuts, cereals and cereal products, gingers, and cashew nuts) Diet (duck poultry, and pork)
0.78a	292.0679	3-Hydroxy-9-(4-hydroxyphenyl)-1H,3H-naphtho[1,8-cd]pyran-1-one	Diet (fruits)
6.75	333.8549	Trichloroepoxyethane Selenate Ficin	Epoxide Diet (chives, beans, soy products) Diet (beer, meat tenderizer, precooked cereals)
U.S.-born associated metabolites
1.46b	167.9232n	Methyl 1-(methylthio)propyl disulfide Ethyl 1-(methylthio)ethyl disulfide	Dietary (onion family) Dietary (fruits)
1.47b	267.7996	Trigraecum	Diet (herbs and spices)
1.47b	390.7740	PC(16:1(9Z)/P-18:1(9Z)) PENMe(20:2(11Z,14Z)/16:0)	Lipid metabolism Phosphatidylcholine biosynthesis

^aAssociated with U.S.-born group pre-intervention (Table 5). ^bAssociated with foreign-born group pre-intervention (Table 4).

To further understand the effect of the intervention on the metabolome by birthplace, OPLS-DA was conducted to determine whether there were any differences between pre- and post-intervention among the foreign and U.S.-born groups separately. The pre versus post-intervention model for the foreign-born group had weak predictability (R² = .35; metabolites shown in Table 4), while the model for the U.S.-born group had moderate predictability (R² = .64; metabolites shown in Table 5). For the foreign-born group analysis, there were 35 metabolites selected from the S-plots. Of these, 14 had quality peaks with four metabolites also associated with birthplace (shown in Tables 2 and 3). Of the 14 identified metabolites, 10 metabolites were exclusively associated with the pre-intervention time point and four were exclusively associated with the post-intervention time point. Potential IDs for the metabolites characteristic of the pre-intervention time point for the foreign-born group included lipid-related metabolites such as those involved in lipid metabolism, phosphatidylcholine biosynthesis, and glycerophospholipids. Some putative IDs for the rest of the pre-intervention metabolites included onions, fruits, coffee, beef, pork, shrimp, legumes, vegetables, and cruciferous vegetables, and there was a human gut microbiome-related metabolite. Putative IDs for metabolites characteristic of the post-intervention time point included diet-related metabolites such as animal protein from fish, duck, poultry, and pork, some food additive, herbs and spices, and onions and garlic. Importantly, metabolites identified post-intervention were not associated with birthplace in the primary analysis.

Table 4.

Metabolites Pre- Versus Post-Intervention Among the Foreign-Born Group of the ANIMO Study (n = 19).

Retention time	M/Z	Putative IDs or class	Source
Pre-intervention-associated metabolites
1.46a	167.9232n	Methyl 1-(methylthio)propyl disulfide Ethyl 1-(methylthio)ethyl disulfide	Dietary (onion family) Dietary (fruits)
1.47	189.8945	LysoPA (24:0/0:0)	Lipid
1.47a	267.7996	Trigraecum	Diet (herbs and spices)
1.47a	390.7740	PC(16:1(9Z)/P-18:1(9Z)) PENMe(20:2(11Z,14Z)/16:0)	Lipid metabolism Phosphatidylcholine biosynthesis
1.49	299.7907	PGP(5-iso PGF2VI/i-13:0)	Glycerophospholipids
2.01	122.9630	Thiophene	Diet (coffee, beef, pork, shrimp, papaya)
2.01	135.9876	2,4-Dihydroxyacetophenone 5-sulfate Vanillin 4-sulfate	Polyphenol/ diet (fruits) Polyphenol/ diet (fruits)
2.01	163.9536	3-Isoxazolidinone 2-Aminoacrylic acid	Diet (legumes) Diet (fruits and vegetables)
3.79	194.0003	Formononetin 7-sulfate Benzyl isothiocyanate	Diet (isoflavone) Diet (cruciferous vegetables)
3.96b	120.0011	Maleylacetoacetic acid 2,4,5-trihydroxy-3-methoxybenzoic acid (-)-1-(Methylthio)propyl 1-propenyl disulfide / (-)-1-(Methylthio)propyl 1-propenyl disulfide 2-(2H-1,3-benzodioxol-5-yl)-2-oxoacetic acid	Tyrosine metabolism intermediate Lipid metabolism and transport/ diet Organosulfur compound/ diet (green vegetables) Organoheterocyclic compounds/ human gut microbiota
Post-intervention-associated metabolites
0.64c	169.9827	Taurine	Diet (animal and fish protein)
0.67	688.7475	Quillaic acid	Diet (food additive)
0.81	494.9361	p-Lacto-N-octaose TG(20:2n6/O-18:0/22:5(7Z,10Z,13Z,16Z,19Z))	Diet (duck, poultry, and pork) Triglyceride
0.83	306.9422	5-(4-Chloro-3-hydroxy-1-butynyl)-2,2’-bithiophene Di-2-propenyl hexasulfide	Diet (herbs and spices) Diet (onions and garlic)

^aAssociated with US-born group post-intervention (Table 3). ^bPost-intervention is the 12-week time point for the intervention group and the 24-week time point for the waitlist control. cAssociated with US-born group pre-intervention (Table 2).

Table 5.

Metabolites Pre- Versus Post-Intervention Among the U.S.-Born Group Only From the ANIMO Study (n = 23).

Retention time	M/Z	Putative IDs or Class	Source
Pre-intervention-associated metabolites
0.60a	233.9103	Ginsenoside M7cd Notoginsenoside R9	Diet (tea) Diet (tea)
0.78a	292.0679	3-Hydroxy-9-(4-hydroxyphenyl)-1H,3H-naphtho[1,8-cd]pyran-1-one	Diet (fruits)
2.01b	135.9876	2,4-Dihydroxyacetophenone 5-sulfate Vanillin 4-sulfate	Polyphenol/ diet (fruits) Polyphenol/ diet (fruits)
2.01b	163.9536	3-Isoxazolidinone 2-Aminoacrylic acid	Diet (legumes) Diet (fruits and vegetables)
3.79b	194.0003	Formononetin 7-sulfate Benzyl isothiocyanate	Diet (isoflavone) Diet (cruciferous vegetables)
3.96bc	120.0011	Maleylacetoacetic acid 2,4,5-trihydroxy-3-methoxybenzoic acid (-)-1-(Methylthio)propyl 1-propenyl disulfide / (-)-1-(Methylthio)propyl 1-propenyl disulfide 2-(2H-1,3-benzodioxol-5-yl)-2-oxoacetic acid	Tyrosine metabolism intermediate Lipid metabolism and transport/ diet Organosulfur compound/ diet (green vegetables) Organoheterocyclic compounds/ human gut microbiota
Post-intervention-associated metabolites
0.67	202.9306	Ethylphosphate 2-Acetylthiophene	Diet (duck, poultry, pork) Diet (asparagus, kohlrabi)
0.69	313.8973	PA(i-24:0/i-24:0) TG(14:1(9Z)/22:6(4Z,7Z,10Z,13Z,16Z,19Z)/O-18:0)	Phosphatidic acid-biosynthesis of triacylglycerols and phospholipids Dietary fat transfer
0.69	330.8712	PC(20:4(8Z,11Z,14Z,17Z)-2OH(5S,6R)/24:1(15Z)) PG(LTE4/i-22:0)	Glycerophospholipids Phosphatidylglycerols
0.78	159.9598	Methaneselenol	Diet (onions and garlic)
0.78	291.9254	Orthoperiodic acid PGP(18:1(9Z)/16:0)	Diet (milk and milk products) Glycerophospholipid/diet (fish oils, milk fats, vegetable oils and animal fats)
0.84	143.0024	L-Cysteine 4-[(4’-O-Acetyl-alpha-L-rhamnosyloxy)benzyl] isothiocyanate	Amino acid Diet (supplements)

^aAssociated with foreign-born group post-intervention (Table 3). ^bAssociated with foreign-born group pre-intervention (Table 4). ^cAssociated with foreign-born group pre-intervention (Table 2).

For the U.S.-born group, there were 20 metabolites selected from the S-plots. Of these, 12 had quality peaks with two metabolites having overlapping associations with birthplace (Table 3) and four with overlapping associations with foreign-group pre-intervention (Table 4). Of the 12 metabolites identified among the U.S.-born group (Table 5), six of these were identified pre-intervention and six post-intervention. Similarly, the majority of the metabolites characteristic of the pre-intervention time point had putative IDs from dietary sources. These included tea, fruits, legumes, vegetables, cruciferous vegetables, and green vegetables, with only one putative ID related to tyrosine metabolism and another to human gut microbiota. Post-intervention diet-related putative IDs included animal protein (duck, poultry, pork), vegetables such as asparagus, onions, and garlic, milk and milk products, and dietary fats such as fish, milk, vegetable and animal fats, and oils. Post-intervention, there were two lipid-related metabolites—one potentially involved in the dietary fat transfer or triacylglycerol synthesis and another potentially glycerophospholipids or phosphatidylglycerols. None of the post-intervention metabolites among U.S.-born participants were associated with birthplace in the primary analysis.

Additional exploratory analyses were conducted based on the participant’s acculturation status. Metabolomic profiles between participants who had at least some level of acculturation, (defined as a score of > 2 in the ARSMA-II), compared with those who were “very Mexican oriented” (score = 1 in the ARSMA-II), were found to be moderately (R² = .55) different at baseline. There were 19 metabolite features selected from the S-plots, and of these, two had quality peaks: 163.9536 m/z and 135.9876 m/z. Putative IDs for 163.9536 m/z included compounds from legumes, fruits, or vegetables and for 135.9876 m/z possible IDs included polyphenol compounds or fruits.

Discussion

This secondary analysis of a lifestyle intervention study suggests that there are differences in plasma metabolomic profiles of foreign-born compared with U.S.-born Hispanic men predominantly from Mexico. Metabolomic differences based on birthplace, independent of the study arm, were attributed to metabolites of which the majority were diet related. Pre-intervention, the foreign-born group demonstrated characteristic metabolites such as α-keto and amino acids, phospholipids, fatty acids, as well as green vegetables, onions, fruits, herbs. and spices. For the U.S.-born group, pre-intervention metabolites included animal and fish protein, soy products, alcoholic beverages, herbs such as thyme and basil, peas, carrots, and coffee. Post-intervention, foreign-born participants demonstrated metabolites of tea, fruits, poultry, pork, cereal products, and soy products, while U.S.-born demonstrated metabolites of onions, garlic, fruits, herbs and spices, and some lipid-related metabolites. Overall, our study findings highlight that environmental exposures associated with birthplace, primarily diet, are evident in the metabolome of predominately Mexican-origin Hispanic men. In the future, these results could inform strategies to target specific factors associated with birthplace to help identify risk factors for disease among this group.

Our finding that metabolomic differences exist among individuals from different countries of origin has been demonstrated in other studies. Dumas et al. reported polyphenol metabolites were positively associated with a U.S. sample population but not those in Japan or China (Dumas et al., 2006). In our study, not surprisingly, polyphenols were associated with both the highly acculturated group (135.9876 m/z; data not shown) and the U.S.-born group pre-intervention. Surprisingly, although reports suggest that consumption of polyphenols in Mexico is much lower than in the United States (Zamora-Ros et al., 2018), in this analysis, polyphenols were also characteristic of the foreign-born group pre-intervention. Furthermore, 2,4-dihydroxybenzoic acid, a potential ID for a metabolite associated with the U.S.-born group pre-intervention (230.9484 m/z), has been associated with alcoholic beverages (Humia et al., 2019). Given that acculturation status among Hispanics has been associated with alcohol intake (r = .10, p < .001), these results are not surprising.

The majority of the metabolic differences between foreign and U.S.-born participants are metabolites derived from dietary compounds. The foreign-born group was characterized by metabolites from dietary sources such as onions (167.9232 n), herbs and spices (267.7996 m/z), and fruits (167.9232 n, 135.9876 m/z, 163.9636 m/z), which are foods characteristic of a traditional Mexican dietary pattern (Santiago-Torres et al., 2015; Valerino-Perea et al., 2019). Metabolites associated with the U.S.-born groups included metabolites associated with flavonoid-rich foods such as wine, tea, and coffee (165.9827 m/z, 233.9103 m/z). Both the foreign- and the U.S.-born group had overlapping associations pre-intervention with metabolites putatively from cruciferous and green vegetables (194.0003 m/z, 120.0011 m/z). Metabolites associated with onions and garlic (167.9232 n, 159.9598 m/z), which were characteristic of the foreign-born group in the primary analysis, were associated with the U.S.-born post-intervention. These findings may indicate that the intervention promoted the adoption of dietary behaviors that are more commonly associated with the Mexican culture (Santiago-Torres et al., 2015; Valerino-Perea et al., 2019). The fact that there was a greater change in metabolomic profiles pre- versus post-intervention among the U.S.-born group (R² = .64) compared with the foreign-born group (R² = .35), could be explained by an unintentional dietary shift to a more traditional Mexican diet observed in the U.S.-born group, whereas the foreign-born group navigated the intervention program while maintaining some of their healthy Mexican dietary practices. Given the lifestyle intervention was meant for weight loss and despite no metabolomic differences were observed with weight loss in this study (data not shown), the U.S.-born participants did experience 2.94% (3.47 kg) greater weight loss compared with foreign-born participants, independent of the study arm. Furthermore, this may suggest that dietary behaviors can be re-introduced through dietary interventions among those who are more acculturated such as U.S.-born Hispanics, and potentially induce weight loss. Importantly, this finding supports the testing of a more traditional Mexican dietary pattern to reduce the higher disease risk observed among U.S.-born Hispanics. This dietary pattern has been previously reported to reduce chronic diseases including lower cancer risk and mortality (Lopez-Pentecost et al., 2020) as well as reduce obesity-related cancer risk factors such as inflammation and insulin insensitivity (Santiago-Torres et al., 2016, 2017). Thus, an intervention that promotes the adoption of this dietary pattern and explores its effects on weight loss could serve as a promising strategy for disease risk reduction among this population.

While there were distinct metabolites associated with foreign versus U.S.-born individuals, we identified some similarities in metabolites among the groups. Overall, both groups had lipid metabolism and phospholipid-related metabolites. For example, a metabolite potentially related to lipid metabolism (120.0011 m/z) was observed for both the foreign and the U.S.-born pre-intervention, and another one (390.7740 m/z) was observed for the foreign-born group pre-intervention as well as the U.S.-born post-intervention. Interestingly, lipid metabolism metabolites have previously been reported to contribute to the underlying inflammatory pathway through which diet potentially influences disease risk (Tabung et al., 2020). Similarly, we identified metabolites that have previously been associated with hyperinsulinemic dietary patterns such as phospholipid-related metabolites. One metabolite potentially identified as glycerophospholipid (299.7907 m/z) was associated with the foreign-born group, while in the U.S.-born group, one metabolite was potentially identified as a diacylglycerol pre-intervention (726.6804 m/z), and two were potentially identified as glycerophospholipids (291.9254 m/z, 330.8712 m/z), one as triacylglycerols (313.8973 m/z). and all three were observed post-intervention. Given the analyses of the current study are not designed to quantify the concentration differences for the metabolites observed, it is important to highlight that it is unknown whether the groups had higher or lower concentrations of these metabolites but we know these were characteristic of the groups at the aforementioned time points.

The study sample of the current study was composed of individuals with overweight and obesity. Therefore identifying tyrosine and phenylalanine as potential IDs for several metabolites (120.0011 m/z and 117,9877 m/z, respectively) pre-intervention for both groups was not surprising as these are congruent with the literature. Particularly, these amino acid metabolites have been observed in metabolomic profiles of overweight and obese men in previous studies (Bagheri et al., 2019; Kim et al., 2010; Newgard et al., 2009; Xie et al., 2014). However, given age has been reported to influence the metabolome (Darst et al., 2019), an important consideration for the current study is the statistically significant difference in age between the groups, with foreign-born participants being approximately a decade older and thus may have driven some of these differences.

Taken together, the results presented highlight the need to identify whole systemic profile-based analysis that may more closely serve as biomarkers of disease risk. Similarly, in current nutrition research, studying dietary patterns, which have proven to be more indicative of disease risk and overall health, is being recommended over investigating individual nutrients (U.S. Department of Health and Human Services and U.S. Department of Agriculture, 2015). As such, investigating metabolite patterns as a whole, their association with disease risk and as identifiers of intervention responsiveness, instead of single metabolites, may hold greater promise for future metabolomic studies. Particularly, future interventions should focus on metabolomic patterns and changes thereof linked to specific cancer-related endpoints to identify targets aimed at reducing cancer disparities among the Hispanic population.

This study has several limitations, first, for each metabolite, several potential metabolite IDs were identified, and it was not possible to confirm identification utilizing pure standards to compare fragmentations, given our limited sample availability. Second, this was a pilot study with a relatively small sample size; thus, the results are largely exploratory and hypothesis-generating, and a much larger sample is needed to confirm these findings. The current study was not inherently designed to detect differences between birthplace, and therefore findings may be due to confounding factors such as age and intervention adherence. Moreover, our results are sex-dependent, given the fact that only Hispanic men were included in the study and therefore we cannot extrapolate our findings to the general population. However, this limitation rendered one of the strengths of the study given that sex can be a confounding factor in metabolomic profiles. Additional strengths include that the untargeted analytical approach yielded metabolites not previously associated with birthplace or other exposures such as diet.

In conclusion, this study identified different metabolites that were related to birthplace among a sample of 42 Hispanic men with obesity that participated in a previous lifestyle intervention study. Overall, our results highlight the feasibility of metabolomics to identify metabolic signatures associated with birthplace among Hispanic men residing in the Southwestern United States. These results advance our understanding of potential underlying differences through which individuals from the same racial background may possibly experience differences in cancer and cancer-related risk factors based on their birthplace.