Maternal pre-pregnant obesity is associated with cord blood metabolomics in a multi-ethnic cohort

Background Maternal obesity has become a growing global health concern that impacts fetal health and subsequently predisposes the offspring to medical conditions later in life. However, the metabolic link between maternal pre-pregnant obesity and offspring has not yet been fully elucidated. Objective This study aims to investigate metabolomics changes in fetal cord blood associated with obese (BMI>30) and normal pre-pregnant weight (18.5<BMI<25) mothers. Design In this study, we conducted a case-control study using coupled untargeted and targeted metabolomics approach, from the newborn cord blood metabolomes associated with a matched discovery cohort of 28 cases and 29 controls for maternal pre-pregnant obesity. The subjects are recruited from multi-ethnic populations in Hawaii, including rarely reported Native Hawaiian and other Pacific Islanders (NHPI). The results are subsequently validated in by an indepdent cohort of 12 cases and 18 controls. Results Maternal obesity is the most important factor contributing to differences in cord blood metabolomics. Using elastic net regularization based logistic regression model, we identify 29 metabolites as early-life biomarkers manifesting intrauterine effect of maternal obesity, with accuracy as high as 0.947 even after adjusting for clinical confounding (maternal ethnicity etc). This obese model is validated in a separate cohort (N=30) with accuracy of 0.822. Among the metabolites, six metabolites of them (galactonic acid, butenylcarnitine, 2-hydroxy-3-methylbutyric acid, phosphatidylcholine diacyl C40:3, 1,5-anhydrosorbitol, and phosphatidylcholine acyl-alkyl 40:3) are individually significantly different between the maternal obese vs. norm-weight groups. Interstingly, hydroxy-3-methylbutyric acid shows significnatly higher levels in cord blood from the NHPI group in the dicovery cohort, compared to asian and caucasian groups. This trend is also observed in the validation cohort. Conclusions The work here demonstrates the significant associations between maternal pre-pregnant obesity and offspring metabolomics alternation at birth, revealing the inter-generational impact of maternal obesity.


70
Obesity is a global health concern. While some countries have a relative paucity of obesity, in the United 71 States, obesity affects 38% of adults (1, 2). As such, maternal obesity has risen to epidemic proportions in 72 recent years and can impose significant risk to both the mother and unborn fetus. Recently, research has 73 7 The plasma samples were thawed and extracted with 3-vol cold organic mixture of ethanol: chloroform and 150 centrifuged at 4 °C at 13500 rpm for 20 min. The supernatant was split for lipid and amino acid profiling 151 with an Acquity ultra performance liquid chromatography coupled to a Xevo TQ-S mass spectrometry 152 (UPLC-MS/MS, Waters Corp., Milford, MA). Metabolic profiling of other metabolites including organic 153 acids, carbohydrates, amino acids, and nucleotides were measured with an Agilent 7890A gas 154 chromatography coupled to a Leco Pegasus time of flight mass spectrometry (Leco Corp., St Joseph, MI). 155 The raw data files generated from LC-MS (targeted) and GC-MS (untargeted) were processed with 156 TargetLynx Application Manager (Waters Corp., Milford, MA) and ChromaTOF software (Leco Corp.,St 157 Joseph, MI) respectively. Peak signal, mass spectral data, and retention times were obtained for each 158 metabolite. The detected metabolites from GC-MS were annotated and combined using an automated mass 159 spectral data processing (AMSDP) software package (26). The levels of lipids and amino acids detected 160 from LC-MS were calculated with calibration curves established with reference standards. 161

Metabolomics data processing 162
We conducted data pre-processing similar to the previous report (27). Briefly, we used K-Nearest 163 Neighbors (KNN) method to impute missing metabolomics data (28). To adjust for the offset between high 164 and low-intensity features, and to reduce the heteroscedasticity, the logged value of each metabolite was 165 centred by its mean and autoscaled by its standard deviation (29). We used quantile normalization to reduce 166 sample-to-sample variation (30). We applied partial least squares discriminant analysis (PLS-DA) to 167 visualize how well metabolites could differentiate the obese from normal samples. To explore the 168 contribution of different clinical/physiological factors to metabolomics data, we conducted source of 169 variation analysis. We used comBat Bioconductor R package (31) to adjust for the batch effects in the 170 metabolomics data. 171

Classification modeling and evaluation 172
To reduce the dimensionality of our data (230 metabolites vs 57 samples), we selected the unique 173 metabolites associated with separating obese and normal status. To achieve this, we used a penalized 8 logistic regression method called elastic net that was implemented in the glment R package (32). Elastic 175 net method selects metabolites that have non-zero coefficients as features, guided by two penalty parameters 176 alpha and lambda (32). Alpha sets the degree of mixing between lasso (when alpha=1) and the ridge 177 regression (when alpha=0). Lambda controls the shrunk rate of cofficients regardless of the value of alpha. 178 When lambda equals zero, no shrinkage is performed and the algorithm selects all the features. As lambda 179 increases, the coefficients are shrunk more strongly and the algorithm retrives all features with non-zero 180 coefficients. To find optimal parameters, we performed 10-fold cross-validation that yield the smallest 181 prediction minimum square error (MSE). We then used the metabolites selected by the elastic net to fit the 182 regularized logistic regression model. Three parameters were tuned: cost, which controls the trade-off 183 between regularization and correct classification, logistic loss and epsilon, which sets the tolerance of 184 termination criterion for optimization. 185 To construct and evaluate the model, we divided samples into 5 folds. We trained the model on four folds 186 (80% of data) using leave one out cross validation (LOOCV) and measured model performance on the 187 remaining fold (20% of data). We carried out the above training and testing five times on all folds 188 combination. We plotted the receiver-operating characteristic (ROC) curve for all folds prediction using 189 pROC R package. To adjust confounding other clinical covariants such as ethnicity, gravidity and parity, 190 we reconstructed the metabolomics model above by including these factors. 191

Analysis on metabolite features 192
We used Classification And REgression Training (CARET) R package to rank metabolites based on the 193 model-based approach (33). In this approach, each metabolite was assigned a score that estimates its 194 contribution to the model performance (34). These scores were scaled to have a maximum of 100. We 195 performed metabolomic pathway analysis on metabolites chosen by the elastic net method using 196 Consensus Pathway DataBase (CPDB). We used rcorr function implemented in Hmisc R package to 197 compute the correlations among clinical and metabolomics data. in the metabolomics, we conducted source of variation analysis. Indeed, maternal obesity is the 223 predominatly most important factor contributing to metabolomic difference, rather than other factors 224 ( Figure 1A). To test if these metabolites allow clear separation between the obese and normal-weight 225 subjects, we used elastic net regularization based logistic regression, rather than the partial least squares 226 (PLS) model, a routine supervised multivariate method which only yielded modest accuracy AUC=0.62 227 ( Figure 1S). Elastic net regularization overcomes the limitation of either ridge and lasso regularization 228 alone, and combines their strengths to identify an optimized set metabolites [25]. Using the optimized 229 regularization parameters ( Figure.  Isovalerylcarnitine (C5), PC ae C40:2, L-arabitol, Octadecenoylcarnitine (C18:1) ( Figure 3A, Table 2) . 260 The remaining 14 metabolites are lower in obese associated cord blood samples: malic acid, L-aspartic acid, 261 citric acid, PC ae C34:0, isoleucine, PC ae C36:2, oleic acid, PC aa C36:5, PC ae C34:3, PC ae C40:6, 262 C5:1-DC, 2-hydroxybutyric acid, myoinositol, and C16:1 -OH ( Figure 3A, Table 2). The individual 263 metabolite levels of Hexanoylcarnitine (C6(C4:1-DC)), O-butanoyl-carnitine (C4:1), PC aa C40:3, 264 Propionylcarnitine (C3), PC ae C40:3, galactonic acid, and 2-hydroxy-3-methylbutyric acid increased 265 significantly in obese cases (p<0.05, t-test). 266 To elucidate the biological processes in newborns that may be effected by maternal obesity, we performed 267 pathway enrichment analysis on the 29 metabolite features, using Consensus pathway database (CPDB) 268 tool (37). We combined multiple pathway databases including KEGG, Wikipathways, Reactome, EHNM 269 and SMPDB. A list of 10 pathways are enriched with adjusted p-value q<0.05 ( Figure 3B). Among them, 270 alanine and aspartate metabolism is the most significantly enriched pathway (q=0.004). Transmembrane 271 transport of small molecules and SLC-mediated transmembrane transport are also significantly enriched 272 (q=0.004 and q=0.01 respectively). 273

The influence of ethinicity on metabolite levels 275
Our earlier correlational analysis suggested that maternal ethnicity may be correlated with 2-hydroxy-3-276 methylbutyric acid level (Figure 2A). To confirm this, we conducted 2-way ANOVA statistical tests and 277 indeed obtained significant p-value (P=0.023, chi-square test). We thus stratified the levels of 2-hydroxy-278 3-methylbutyric acid by ethnicity (Figure 4). There is no significant difference in normal pre pregnant-279 weight subjects across the three ethnic groups ( Figure 4A). However, in cord blood samples associated with 280 obese mothers, the concentration of 2-hydroxy-3-methylbutyric acid is much higher in NHPI, as compared 281 to Caucasians (p=0.05) or Asians (p=0.04) ( Figure 4B). 2-hydroxy-3-methylbutyric acid originates mainly 282 from ketogenesis through the metabolism of valine, leucine and isoleucine (38). Since all subjects have 283 fasted 8 hours before the C-section, we expect the confounding from diets is minimized among the three 284 ethnical groups. Thus the higher 2-hydroxy-3-methylbutyric acid level may indicate the higher efficiency 285 of ketogenesis in babies born from obese NHPI mothers. 286 287

Validation on an independent cohort 288
To test the robustness of our results, we applied our model on a new cohort of 30 patients (18 normal-289 weight and 12 obese). We then performed new metabolomics measurements and processed the data as This study aims to distinguish key cord blood metabolites associated with maternal pre-pregnancy obesity. 301 The novelty of the study is manifested in several folds. First, we have collected a unique multi-ethnic 302 population in Hawaii over a period of 3 years (2015-2018), which includes Asian, NHPI and Caucasians, 303 following very strict inclusion/exclusion criteria (esp. on matching gestational weight gain). Secondly, we 304 utilize state of the art metabolomics technology platform coupling GC-MS and LC-MS platforms, which 305 allows us to detect hundreds of metabolites simultaneously. Lastly, we use the state of art method called 306 elastic net based logistic regression that drastically improves the classification accuracy on cord blood 307 metabolomics data. 308 To ensure the quality of metabolomics data, our study set most stringent inclusion and exclusion crtieria to 309 exclude as many confouding factors as possible. To avoid the confounding from labor and vaginal delivery, 310 we only targeted mothers having elective C-sections. We also excluded obese mothers who had known 311 complications during pregnancy, such as pre-gestational diabetes, smoking, and hypertension. These 312 criteria helped to improve the quality of the metabolomics data. To minimize confounding due to maternal 313 diet, all subjects fasted 8 hours before the Cesarean section. 314 Such careful experimental design did yield good data quality, as the source of variation analysis did show 315 that maternal obesity is the only dominate factor contributing to metabolomics diffeence in the cord blood. 316 Additionally, we conducted rigorous statistical modeling and found that metabolites can distinguish the two 317 maternal groups with accuracy as high as AUC=0.97 under cross-validation (or 0.947 after adjusting for 318 confounding effects). Metabolomics pathway analysis on the metabolite features in the model identified 10 319 significant pathways. Among them, alanine and aspartate metabolism was previously reported to be 320 associated with obesity (39). Transmembrane transport was identified as another significant pathway. The 321 transmembrane transport pathway corresponds to the acylcarnitine metabolites in the features. 322 Acylcarnitines are known transmembrane transporters of fatty acids across the mitochondrial membrane 323 (40). Among all metabolites and physiological/demographic features selected by the combined model, 324 galactonic acid has the largest impact on the model performance (importance score =86%). Galactonic acid, 325 was previously shown to be associated with diabetes in a mouse model, due to a proposed mechanism of 326 oxidative stress (41). On the other hand, maternal ethnicity has the largest impact among physiological 327 factors (importance score =84%). 328 A very few cord blood metabolomics studies have been carried out to associate with maternal obesity 329 directly, or birth weight (22,42,43). In a recent Hyperglycemia and Adverse Pregnancy Outcome (HAPO) 330 Study, Lowe et al. reported that branched-chain amino acids such as valine, phenylalanie, leucine/isoleucine 331 and AC C4, AC C3, AC C5 are associated with maternal BMI in a meta-analysis over 4 large cohorts (400 332 subjects in each) (43). In another study to associate cord blood metabolomics with low birth weight (LBW), 333 Ivorra et al. found that newborns of LBW (birth weight < 10th percentile, n = 20) had higher levels of 334 phenylalanine and citrulline, compared to the control newborns (birth weight between the 75th-90th 335 percentiles, n = 30) (22). They also found lower levels of choline, proline, glutamine, alanine and glucose 336 in new borns of LBW, however, there was no significant differences between the mothers of the two groups. 337 In our study, isoleucine is also identified as one of the 29 metablite features related to maternal obesity; 338 although alanine iteself is not selected by the model to be a maternal obesity biomarker in cord blood, we 339 did find that alanine and aspartate metabolism are enriched in the cord blood samples associated with 340 maternal obesity group. 341 Notably, our study has identified 5 metabolites which are previously not reported in the literature with 342 association to obesity or maternal obesity: galactonic acid, L-arabitol, indoxyl sulfate, 2-hydroxy-3-343 methylbutyric acid and citric acid. Except citric acid, all the other four metabolites are increased in obese 344 associated cord blood samples. 2-hydroxy-3-methylbutyric acid concentrations varied by ethnicity, but only 345 in babies born from obese pre-pregnant mothers. 2-hydroxy-3-methylbutyric acid is known to accumulate 346 in high levels during ketoacidosis and fatty acid breakdown. Therefore, the higher elevation of 2-hydroxy-347 3-methylbutyric acid is likely due to increased cellular ketoacidosis and fatty acid breakdown in new borns 348 from obese pre-pregnant mothers. To the best of our knowledge, this is the first study that shows differences 349 in the 2-hydroxy-3-methylbutyric acid concentration levels among different ethnicities. Additionally, 350 Indoxyl sulfate is a metabolite of the amino acid tryptophan. As tryptophan is commonly found in fatty 351 food, red meat and cheese, it is possible that high levels of indoxyl sulfate detected in the cord blood 352 associated with obese pre-pregnant mothers could be due to the maternal high fat diet. Oppositely, citric 353 acid, a compound associated with the citric acid cycle (44), is decreased in the cord blood associated with 354 obese pre-pregnant mothers. This could be related to the lower vegitable and fruit consumptions among 355 obese pre-pregnant mothers. In all, the data suggest that maternal obesity may impact offspring cord blood 356 metabolites. Further research into the specific mode of action of these metabolites would be beneficial in 357 understanding its association with maternal obesity. 358

359
This study may benefit from some improvmenet in the future follow-up s. We determined the subjects' 360 ethnicity by self-reporting rather than genotyping, due to the restriction of the currently approved IRB 361 protocol. Additionally, there has been debates on the use of BMI as an indicator of obesity (45), more direct 362 measures of body fat could be considered such as skin-fold thickness measurements, bioelectrical 363 impedance and energy x-ray absorptiometry (46,47). Nevertheless, this study has established relationships 364 between cord blood metabolomics with maternal pre-pregnant obesity, which in turn is associated with 365 social economical disparities.