Urinary Metals Concentrations and Biomarkers of Autoimmunity among Navajo and Nicaraguan Men

Metals are suspected contributors of autoimmune disease among indigenous Americans. However, the association between metals exposure and biomarkers of autoimmunity is under-studied. In Nicaragua, environmental exposure to metals is also largely unexamined with regard to autoimmunity. We analyzed pooled and stratified exposure and outcome data from Navajo (n = 68) and Nicaraguan (n = 47) men of similar age and health status in order to characterize urinary concentrations of metals, compare concentrations with the US National Health and Nutrition Examination Survey (NHANES) male population, and examine the associations with biomarkers of autoimmunity. Urine samples were analyzed for metals via inductively coupled plasma mass spectrometry (ICP-MS) at the US Centers for Disease Control and Prevention. Serum samples were examined for antinuclear antibodies (ANA) at 1:160 and 1:40 dilutions, using an indirect immunofluorescence assay and for specific autoantibodies using enzyme-linked immunosorbent assay (ELISA). Logistic regression analyses evaluated associations of urinary metals with autoimmune biomarkers, adjusted for group (Navajo or Nicaraguan), age, and seafood consumption. The Nicaraguan men had higher urinary metal concentrations compared with both NHANES and the Navajo for most metals; however, tin was highest among the Navajo, and uranium was much higher in both populations compared with NHANES. Upper tertile associations with ANA positivity at the 1:160 dilution were observed for barium, cesium, lead, strontium and tungsten.


Introduction
In epidemiology, there is a long-standing interest in determining whether various diseases are concentrated in populations defined by racial, ethnic, demographic, and/or geographic characteristics. extractable nuclear antigens [41]. The same study reported detectable urinary concentrations of lead, cadmium, uranium and mercury, but did not assess the association between metals exposure and autoimmune biomarkers [41].
In addition to shared concerns regarding metals exposures among an indigenous North American and a Central American community, increasing prevalence of chronic diseases relative to US white populations have been reported among Navajo Nation and in Mesoamerica (including Nicaragua) [43][44][45][46]. These epidemiological similarities, as well as the availability of both urinary metals and ANA data in both populations, prompted the current collaborative, hypothesis-generating study to investigate the relationships between environmental metal contaminants and serological biomarkers of autoimmune disease in groups from both regions.
Using data from two geographically distinct groups, men from the Navajo Birth Cohort Study (NBCS) and men from an occupational study of sugarcane workers in Nicaragua, the present study had the following objectives: (1) Characterize exposure to metals among healthy Nicaraguan and Navajo men of working age; (2) evaluate biomarkers of autoimmunity (ANA and specific autoantibodies) in each group; and (3) examine the relationship between metals exposure and biomarkers of autoimmunity in the pooled population.

Study Population
The study population included 47 male sugarcane workers from northwestern Nicaragua and 68 fathers from the Navajo Nation. All study participants were at least 18 years of age. Previously, we described the enrollment of 284 sugarcane workers from one company in the department of Chinandega, Nicaragua, in 2010 [47]. In March 2015, we re-sampled 50 of these same workers randomly selected from among those who were still employed by the company in 2015 and worked in one of four job tasks: (1) Cane cutting, (2) irrigation, (3) seeding/seed-cutting or (4) agrichemical application. We collected blood and urine samples post-shift. Three workers refused the blood draw, resulting in complete data for 47 participants. All workers completed questionnaires reporting no disease diagnoses, chronic or acute health conditions. The NBCS is a congressionally mandated collaborative research study originally supported by the Agency for Toxic Substances and Disease Registry (CDC/ATSDR) with research led by the University of New Mexico Community Environmental Health Program in partnership with the Southwest Research and Information Center, the CDC Division of Laboratory Sciences, the Navajo Nation Department of Health and other Navajo agencies, the Navajo Area Indian Health Service (IHS), and the previously acknowledged six IHS and PL-638 healthcare facilities on Navajo Nation. The purpose of the study is to investigate birth outcomes and child development through age one year in relation to non-occupational exposures to uranium wastes from past mining and milling operations on the Navajo Nation. This study encourages father participation. Nearly 200 NBCS fathers were recruited between February 2013 and December 2015 in six geographic areas of Navajo Nation representing mined and unmined regions. At enrollment, fathers provided blood and urine samples for clinical and research analyses, including investigation of the association between metals exposure, immune function, and autoimmunity. A subset of sixty-eight of these fathers were selected for the current analysis by matching age and health status to the previously recruited Nicaraguan workers, all of whom were men. Due to these selection criteria, the Navajo fathers in this study are not representative of the NBCS fathers.
The Institutional Review Boards at the Boston University Medical Campus and the Nicaraguan Ministry of Health approved the study protocols for the studies in Nicaragua. NBCS received approvals from the University of New Mexico Health Sciences Center Human Research Protections Office  and the Navajo Nation Human Research Review Board (NNR 11.323), which continues oversight of the project. All participants provided informed consent prior to participation in research activities.

Laboratory Analysis of Urinary Metals
All urine samples from both studies were analyzed for metals at the US Centers for Disease Control and Prevention Division of Laboratory Science in Atlanta, Georgia, USA. Metals (antimony, total arsenic, barium, cadmium, cesium, cobalt, lead, manganese, molybdenum, strontium, thallium, tin, tungsten, and uranium) were analyzed using an inductively coupled plasma dynamic reaction cell mass spectrometer (ICP-DRC-MS) (PerkinElmer NexION 300D, Waltham, USA) [48]. Speciated arsenic (arsenobetaine, arsenocholine, arsenite (As III), arsenate (As V), monomethylarsonic acid (MMA), and dimethylarsinic acid (DMA)) was also determined using inductively coupled mass spectrometry (ICP-MS) coupled with high performance liquid chromatography [49]. Limits of detection are published in Supplementary Table S1. Urine creatinine was analyzed via enzymatic method using the Roche/Hitachi Modular P Chemistry Analyzer.

Laboratory Analysis of Biomarkers of Autoimmunity
The presence of antinuclear antibody (ANA) was determined by indirect immunofluorescence (IIF) at the University of New Mexico Health Sciences Center using Hep2 cells as substrate and fluorescein-labeled anti-IgG (H + L) as the detecting reagent (INOVA Diagnostics, San Diego, CA, USA). Sera were diluted 1:40 and 1:160 in phosphate-buffered saline. A dilution of 1:80-1:160 is standard for clinical studies [50][51][52][53], while the 1:40 dilution is intended for exploratory, hypothesis generating [54] studies. Slides were viewed independently by two observers (E.E. & R.L.R.) using an Olympus fluorescence microscope at 600-fold magnification (Olympus, Lake Success, NY, USA). Staining patterns (nuclear, nucleolar, cytoplasmic) were graded on a scale of 0-4+ intensity, based on positive and negative control standard sera that were run with each assay.
Specific autoantibodies were also detected by enzyme-linked immunosorbent assay (ELISA) using an in-house procedure, as previously described [55,56], and used by members of our team in Navajo Nation studies [34]. Antibody binding to chromatin, histones, denatured DNA (dDNA or single-stranded DNA), and native DNA (nDNA or double-stranded DNA) was quantified in duplicate serum samples in a 1:200 dilution after incubation in antigen-coated wells for 2 h at room temperature [55,56]. Bound antibodies were detected with peroxidase-conjugated anti-human IgG (Southern Biotech Associates, Birmingham, AL, USA) followed by color development using a peroxidase secondary substrate and reported as optical density (O.D.) units. Positive and negative controls were included in each assay. Two standard deviations above the mean of the negative controls was considered elevated reactivity.

Statistical Analysis
We created a dichotomous outcome variable for each serum dilution (1:160 and 1:40) to categorize ANA positivity. Positivity was defined as a ≥2+ intensity fluorescence score, which is an appropriate cut-off for a healthy population without known autoimmune disease [57]. To account for differences in urine concentration among individuals, we ran two distinct models. In the first, we normalized metal concentrations to urinary creatinine concentration (g/L), as is conventional practice. In the second model, we statistically adjusted for urinary creatinine as a variable in our regression models, as suggested by Barr et al. 2005 [58], to account for population differences in urine concentration and time of day of urine collection. Metal values below the limit of detection (LOD) were substituted with LOD/ √ 2. Median concentrations of each metal were compared across groups (Navajo and Nicaraguan men) and to the United States adult male population (20 years and older) using data from the 2011-2012 National Health and Nutrition Examination Survey (NHANES), conducted by the US Centers for Disease Control and Prevention [59].
ANA and ELISA autoantibody positivity were first characterized for each group. We then assessed the association between each metal and ANA positivity using multivariable logistic regression. We pooled data on the 47 Nicaraguan and 68 Navajo men, creating more power in our sample (N = 115), and examined the spread of metals concentrations in the pooled data, calculating exposure tertiles for each metal (normalized to urine creatinine). Metals tertile, the categorical exposure variable, was used to predict an aggregate outcome variable indicative of any ANA response at one or more of the three cellular sites (positivity at the cytoplasmic, nuclear or nucleolar site) at the 1:40, and 1:160 dilutions, respectively.
Additional predictors included in all models were age, group (Navajo or Nicaragua) and seafood consumption within the past three days (yes/no). Recent seafood consumption data were lacking for the Navajo participants, however, over 60% of the NBCS mothers reported never having consumed fish, ever. Given cultural beliefs and geographic location in the high desert of Southwestern US, which both preclude consumption of seafood, Navajo participants were coded as unexposed. Data on current cigarette smoking were also lacking among Navajo participants. However, among the Nicaraguan participants current smoking behavior was predictive of increased metals exposure, but was not associated with ANA positivity. Consequently, we ruled out smoking behavior as a potential confounder in our pooled analyses. Given the hypothesis-generating nature of this study, the results were evaluated, based on the strength of association, rather than solely on statistical significance. As such, multiple comparison correction was overly conservative and the data were incompatible with the assumptions of penalized regression models. Data were analyzed using Statistical Analysis Software version 9.4 (SAS Institute Inc. Cary, NC, USA).

Population Characteristics
All participants were male with an average age of 32 years (range: 23-51 years) ( Table 1). The Nicaraguan group was composed of workers employed in active jobs that required them to pass a health screening. Questionnaires also confirmed health status. The selected Navajo group reported having no diagnosed diseases including cardiovascular disease, high blood pressure, diabetes or kidney disease.

Urinary Metals
All urine samples had detectable levels of barium, cesium, cobalt, lead, molybdenum, and strontium (Table 1). We calculated the arithmetic median concentration and range of each metal in both groups and the pooled cohort (Table 1). We also calculated the geometric means and 95% confidence limits for each metal in in both groups, comparing these results with US males in NHANES (Figures 1 and 2). Concentrations of most metals were higher among the Nicaraguan men compared to concentrations among Navajo men (Table 1), and compared to the US adult males (Figures 1 and 2). Median concentrations of arsenic, barium and strontium were approximately twice as high among the Nicaraguan men as compared to US adult male geometric mean concentrations in NHANES and the Navajo men (Figures 1 and 2 and Table 1). The concentrations of all arsenic species, cesium, cobalt, lead, manganese, molybdenum and uranium were also higher among the Nicaraguans, as compared to NHANES and the Navajo. Metals' concentrations that were lower among Nicaraguan men, compared with NHANES and Navajo men were antimony, cadmium, thallium, tin and tungsten. Antimony, cadmium, tin and tungsten were all highest among the Navajo compared with both NHANES and the Nicaraguans, with tin notably higher and detected in 99% of samples among the Navajo as compared to 48% among the Nicaraguans. Uranium concentrations were somewhat similar between the Nicaraguan (0.02 µg/g creatinine) and Navajo (0.01 µg/g creatinine) groups, although 2-3-fold higher compared to NHANES (0.004 µg/g creatinine) (Figures 1 and 2).

Autoimmune Biomarkers
Any ANA positivity based on a staining intensity of ≥2+ at the 1:160 serum dilution was similar between the two groups (7.4% among the Navajo; 8.5% among the Nicaraguans) ( Table 2). ANA positivity at the nuclear site was also similar between groups at both dilutions. However, cellular site specific staining patterns differed between the two groups; this was most apparent at the 1:40 dilution in which the prevalence of ANA positivity was naturally higher than at the 1:160 dilution. Any ANA positivity was higher among the Nicaraguans (80.9%), compared to the Navajo (55.9% positivity), although nucleolar staining was detected in 12% of the Navajo samples at 1:40 dilution, but in none of the Nicaraguan samples at either serum dilution. In contrast, the Nicaraguans had more anticytoplasmic (74.5%) antibody reactivity, as compared to the Navajo (47.1%) at the 1:40 dilution. Overall, in the pooled cohort (n = 115), 66.1% were ANA positive at any cellular site at the 1:40 serum dilution compared to 7.8% positive at any cellular site at the 1:160 serum dilution (Table 2).

Autoimmune Biomarkers
Any ANA positivity based on a staining intensity of ≥2+ at the 1:160 serum dilution was similar between the two groups (7.4% among the Navajo; 8.5% among the Nicaraguans) ( Table 2). ANA positivity at the nuclear site was also similar between groups at both dilutions. However, cellular site specific staining patterns differed between the two groups; this was most apparent at the 1:40 dilution in which the prevalence of ANA positivity was naturally higher than at the 1:160 dilution. Any ANA positivity was higher among the Nicaraguans (80.9%), compared to the Navajo (55.9% positivity), although nucleolar staining was detected in 12% of the Navajo samples at 1:40 dilution, but in none of the Nicaraguan samples at either serum dilution. In contrast, the Nicaraguans had more anti-cytoplasmic (74.5%) antibody reactivity, as compared to the Navajo (47.1%) at the 1:40 dilution. Overall, in the pooled cohort (n = 115), 66.1% were ANA positive at any cellular site at the 1:40 serum dilution compared to 7.8% positive at any cellular site at the 1:160 serum dilution (Table 2).  Specific autoantibodies measured by ELISA are shown in Table 2 and Figure 3. Prevalence of reactivity to the tested antigens was low. In the Nicaraguan group, three participants had slightly elevated reactivity to dDNA, one to histones and one to chromatin; in the Navajo group, low-level reactivity to dDNA was detectable in seven participant samples and to the other antigens (chromatin, histones and nDNA) in two participants. Specific autoantibodies measured by ELISA are shown in Table 2 and Figure 3. Prevalence of reactivity to the tested antigens was low. In the Nicaraguan group, three participants had slightly elevated reactivity to dDNA, one to histones and one to chromatin; in the Navajo group, low-level reactivity to dDNA was detectable in seven participant samples and to the other antigens (chromatin, histones and nDNA) in two participants.    (Table 3) It was not possible to run the models, including urine creatinine as a co-variate for several of the arsenic metabolites, due to low percent detection. Instead we focus discussion on the results of the model evaluating creatinine-normalized concentrations, but provide the results of both normalizing creatinine, and including it as a variable in Supplemental Table S1, Associations between urinary metal concentrations and ANA positivity differed at the two dilutions (1:160 and 1:40). At the 1:160 dilution, odds of ANA positivity at any cellular site were highest and significant among participants with third tertile exposure as compared to participants with first tertile exposure for cesium ( (Table 3). Third tertile barium exposure also had a high association, although not significant (OR = 2.33, 95% CI: 0.83, 6.55). Lower (OR's ranging from 1.01 to 1.78), albeit positive associations between any ANA positivity and third tertile concentrations were also observed for total arsenic, arsenobetaine, arsenocholine, As (III), As (V), antimony, cadmium, cobalt, thallium, tin and uranium. Although the odds of any ANA positivity were higher among participants with second tertile urinary concentrations of DMA (OR = 1.46, 95% CI: 0.56, 3.82), there was no association between the third tertile exposure to DMA and ANA positivity (OR = 0.45, 95% CI: 0.10, 1.97). No association was seen between urinary concentrations of MMA and ANA positivity at the 1:160 dilution.

Associations between Metals and ANA Positivity in Pooled Analyses
The results at the 1:40 dilution were somewhat disparate in that the only comparably high odds ratios for ANA associated with metals in third tertile of exposure were for urinary DMA (OR = 2.17, 95% CI: 0.85, 5.47), which was an opposite trend as at the 1:160 dilution. Although, neither was significant. Associations at 1:40 dilution between any ANA positivity and third tertile urinary concentrations of total arsenic, arsenobetaine, cobalt, strontium and thallium were also positive, but lower-similar to the results seen at the 1:160 dilution (OR's ranged from 1.08 to 1.72). However, antimony in the second tertile exposure was associated with any ANA positivity at the 1:40 dilution (OR = 2.25, 95% CI: 1. 16, 4.35), third tertile exposure was not (Table 3). This was also true at the 1:160 dilution. At the 1:40 dilution, there was no association between urinary tin or uranium and ANA positivity. Finally, the odds ratios were undefined given low counts in each tertile for urinary molybdenum and manganese at the 1:160 dilution and are not included in Table 3.  Note: All models were adjusted for age, seafood consumption within the past three days and cohort. (Nicaragua or NBCS). * Statistically significant at an alpha level of 0.05.

Discussion
This is the first biomonitoring study to examine the prevalence and associations between urinary concentrations of metals and autoimmune biomarkers in Nicaraguan or Navajo men. With respect to metals, urinary concentrations of most metals assessed in this study were highest among the Nicaraguan group. As Nicaragua is a volcanic region that is high in strontium [38], lava and bedrock are probable environmental sources and may explain the higher urinary concentrations of strontium in the Nicaraguan group (median 222.28 µg/g creatinine), as compared to the Navajo group (median 130.16 µg/g creatinine) and NHANES (median 91.8 µg/g creatinine). Volcanic emissions are also known to contain barium, manganese [57], and arsenic [35]. The high urinary concentrations of total arsenic among the Nicaraguans (median 13.5 µg/g creatinine, as compared to NHANES median of 6.1 µg/g creatinine) may also reflect the presence of arsenic in drinking water sources in the region [35]. Dietary habits, such as seafood consumption (as evidenced by the high concentrations of the organic arsenic species arsenobetaine and DMA relative to inorganic arsenic species) [59] or exposure to arsenical pesticides historically used in the region. The low arsenobetaine concentrations among the Navajo are consistent with a population that does not consume seafood [60,61]. It is notable, however, that the inorganic arsenic species are high among the Navajo compared with NHANES.
Navajo participants were expected and found to have relatively high urine uranium, due to proximity to abandoned uranium mine and milling sites. Although, they were not selected with any consideration of exposure history, but rather with a focus on health status. It is possible that the even higher uranium concentrations in Nicaraguan participants may be due to the volcanic activity in the area, but uncertain at best.
While antimony, cadmium, and tungsten concentrations were slightly higher among the Navajo as compared with NHANES and the Nicaraguan men, urinary concentrations of most metals in the Navajo group were lower or similar to NHANES, with the noted exceptions of tin and uranium. A subsequent analysis of urine metals compared Navajo men selected for this study to Navajo men in the larger Navajo Birth Cohort Study (NBCS). The results indicated that there was no difference in urinary concentrations between the subgroups with the exception of tin. Urine tin concentrations from NBCS men selected for analysis were marginally greater (Mann-Whitney U test, p-value 0.045) than concentrations among NBCS men who were not selected for this study. Compared to the NHANES 2011-2012 report of median tin levels among adult men, urinary tin concentrations were high in the Navajo group and 15-fold higher than in the Nicaraguan group. While, tin exposure can occur via contact with dust and soil containing tin [62], the primary exposure route is thought to be via the consumption of food and beverages from tin-containing cans and food packaging [63].
The prevalence of any ANA positivity in the current sample is similar to that observed in other studies of healthy individuals suggesting the findings in this study are generalizable beyond our somewhat unique exposure scenarios. The prevalence of ANA positivity (8%) in the present male population with an average age of 32 years is also similar to the~8% prevalence of ANA positivity among the male US NHANES population age 30-39 years (1999-2004) [64]. Satoh et al. [64] used a cut point of a 3+ intensity score and a dilution of 1:80, while we used a lower intensity ANA score (2+) but higher serum dilution (1:160) that may be more clinically relevant. Dinse et al. [16] estimated 11% prevalence among adult males during the same time period as Satoh et al. [64] at 1:80 dilution and 3+ ANA intensity.
The dilution (1:160 v. 1:40) had a substantial effect on our results. In the present study, two-thirds more of the pooled population had positive antibody binding to any cellular site at the 1:40 dilution compared with the 1:160 dilution. Qualitatively similar observations were made by Mariz et al. [65] in healthy Brazilians where the percentage of ANA positivity increased from 7.6% at 1:160 dilution to 46% at a lower (1:80) dilution. Similarly, Marin et al. [66] found 35.4% of healthy individuals had nuclear staining at 1:40 dilution, and at the 1:160 dilution the same population had only 3.2% staining.
The 1:40 dilution is more sensitive and useful for research examining possible associations with environmental exposures, but has a higher possibility of false positives. The basis for serum reactivity at lower dilution is due, in part, to autoantibodies of the lens epithelium-derived growth factor (LEDGF) nuclear antigen, which has no clinical significance [65,67], and to non-specific binding at high concentration of polyclonal immunoglobulin in this assay format. Nevertheless, the finding that third tertile exposures to DMA, arsenic, arsenobetaine, cobalt, strontium, and thallium were associated, albeit weakly, with ANA at the 1:40 dilution suggests that, long-term exposure to relatively low levels of these metals, may promote production of autoimmune biomarkers.
Specific autoantibodies tested by ELISA in this study (histones, dDNA, chromatin and nDNA) were observed infrequently and only at low levels among Nicaraguan and Navajo men. While, antibodies to histones and dDNA are sensitive indicators of xenobiotic-induced antibodies, especially by ingested medications [68]. Similar immune perturbations did not occur to environmental metals at the exposure duration and levels encountered in the current study. Antibodies to chromatin and nDNA, considered indicators of idiopathic autoimmune disease, were detected at low levels only in two Navajo men. The lack of a strong response to these antigens is consistent with the health status of the groups studied.
Based on the strength of association, the metals that were seemingly associated with ANA positivity at the 1:160 dilution in the present study are barium, cesium, lead, strontium and tungsten. Although, other metals were also positively associated with ANA, the associations were weak or close to null. Lead was the only metal examined in our study with toxicological literature supporting the association [19,69]. However, a recent epidemiological analysis by Dinse et al. [16] found no association between blood lead in NHANES and ANA among men and women, a result which is inconsistent with those of the present study. No literature was found examining the associations of barium, cesium, strontium or tungsten with ANA or any autoimmune outcomes in humans or in animals. Although, there is literature suggesting an association between exposure to arsenic [23,70] and cadmium [18,71], and biomarkers of autoimmunity, the associations observed in the present study were positive but weak. Future investigations are needed to ascertain the relationship between arsenic, cadmium and the autoimmune biomarker production.
Overall, the results of our analyses suggest a possible association between exposures to certain metals (barium, lead, strontium, cesium and tungsten) and the production of antinuclear antibodies, when assessed for positivity at the 1:160 dilution. Given the possibility of interactions between metals, additional research assessing the impact of metal mixtures on autoimmune outcomes is warranted.
This study has several limitations. Although, the prevalence of ANA positivity at the 1:160 dilution is consistent with previously published population-based, healthy or undiagnosed patients' studies [57,64], the small sample size, and relatively low prevalence of positive outcomes (ANA and specific autoantibody reactivity), requires cautionary interpretation of statistical significance. We did not include body weight as a variable in our models, which could influence urinary metals concentrations. The small sample size also limited our ability to assess interactive effects between metals or the association between metals and specific autoantibodies. The present study was also limited to quantifying a small selection of autoantibodies; future studies should consider inclusion of a broader range of autoantibodies specific to different target tissues such as thyroid or pancreatic antigens. The participants in this study were exclusively male, and neither the healthy Nicaraguan workers, nor the Navajo men selected for health status reflect their populations as a whole. Given that women have higher autoimmunity biomarker production, and are most affected by autoimmune disease [64], future studies examining environmental exposures and autoimmune biomarkers should include female participants [1]. Finally, only single urine measures at one point in time were collected for all participants. No diurnal or seasonal variation in urine metal concentrations were examined, and we were not able to examine mercury exposure among the metals.

Conclusions
The low prevalence of ANA positivity in this pooled cohort of relatively young, healthy men using a conservative assay cut-off criterion is consistent with rates published in the literature. Similarly, reactivity to specific autoantibodies (histone, dDNA, nDNA and chromatin) was low. We found urinary concentrations of arsenic, barium and strontium were twice as high among Nicaraguan men, compared with the US population as measured by NHANES, and manganese 50% higher in the Nicaraguan group as compared to the Navajo group. Other adverse outcomes that we did not study may be associated with these high urine concentrations, and follow-up on exposure sources may be warranted. Notable associations between ANA positivity at the 1:160 dilution and upper tertile concentrations of several metals including barium, cesium, lead, strontium, and tungsten suggest that more research on metals and autoimmune biomarkers may also be warranted.

Conflicts of Interest:
Some authors declare the potential for perceived conflict of interest by having received funding from the Association of Sugar Producers in Central America. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. The authors declare no conflict of interest.