Blood biomarkers representing maternal-fetal interface tissues used to predict early-and late-onset preeclampsia but not COVID-19 infection

Abstract

Background: A well-known blood biomarker (soluble fms-like tyrosinase-1 [sFLT-1]) for preeclampsia, i.e., a pregnancy disorder, was found to predict severe COVID-19, including in males. True biomarker may be masked by more-abrupt changes related to endothelial instead of placental dysfunction. This study aimed to identify blood biomarkers that represent maternal-fetal interface tissues for predicting preeclampsia but not COVID-19 infection.

Methods: The surrogate transcriptome of tissues was determined by that in maternal blood, utilizing four datasets (n = 1354) which were collected before the COVID-19 pandemic. Applying machine learning, a preeclampsia prediction model was chosen between those using blood transcriptome (differentially expressed genes [DEGs]) and the blood-derived surrogate for tissues. We selected the best predictive model by the area under the receiver operating characteristic (AUROC) using a dataset for developing the model, and well-replicated in datasets both with and without an intervention. To identify eligible blood biomarkers that predicted any-onset preeclampsia from the datasets but that were not positive in the COVID-19 dataset (n = 47), we compared several methods of predictor discovery: (1) the best prediction model; (2) gene sets of standard pipelines; and (3) a validated gene set for predicting any-onset preeclampsia during the pandemic (n = 404). We chose the most predictive biomarkers from the best method with the significantly largest number of discoveries by a permutation test. The biological relevance was justified by exploring and reanalyzing low- and high-level, multiomics information.

Results: A prediction model using the surrogates developed for predicting any-onset preeclampsia (AUROC of 0.85, 95 % confidence interval [CI] 0.77 to 0.93) was the only that was well-replicated in an independent dataset with no intervention. No model was well-replicated in datasets with a vitamin D intervention. None of the blood biomarkers with high weights in the best model overlapped with blood DEGs. Blood biomarkers were transcripts of integrin-α5 (ITGA5), interferon regulatory factor-6 (IRF6), and P2X purinoreceptor-7 (P2RX7) from the prediction model, which was the only method that significantly discovered eligible blood biomarkers (n = 3/100 combinations, 3.0 %; P =.036). Most of the predicted events (73.70 %) among any-onset preeclampsia were cluster A as defined by ITGA5 (Z-score ≥ 1.1), but were only a minority (6.34 %) among positives in the COVID-19 dataset. The remaining were predicted events (26.30 %) among any-onset preeclampsia or those among COVID-19 infection (93.66 %) if IRF6 Z-score was ≥-0.73 (clusters B and C), in which none was the predicted events among either late-onset preeclampsia (LOPE) or COVID-19 infection if P2RX7 Z-score was <0.13 (cluster C). Greater proportions of predicted events among LOPE were cluster A (82.85 % vs 70.53 %) compared to early-onset preeclampsia (EOPE). The biological relevance by multiomics information explained the biomarker mechanism, polymicrobial infection in any-onset preeclampsia by ITGA5, viral co-infection in EOPE by ITGA5-IRF6, a shared prediction with COVID-19 infection by ITGA5-IRF6-P2RX7, and non-replicability in datasets with a vitamin D intervention by ITGA5.

Conclusions: In a model that predicts preeclampsia but not COVID-19 infection, the important predictors were genes in maternal blood that were not extremely expressed, including the proposed blood biomarkers. The predictive performance and biological relevance should be validated in future experiments.

Keywords: Biomarker; COVID-19; Machine learning; Preeclampsia; Transcriptome.

Graphical abstract

Fig. 1

Predictive modeling pipeline. *, developed model; †, applied model; ‡, two models were developed using either the maternal-blood transcriptome or blood-derived surrogate; DEG, differentially expressed gene.

Fig. 2

Distribution of weights used to adjust the gene expression probability. The weight was determined by Matthew’s correlation coefficient (MCC) and rounded to two decimal places for binning MCCs. *, ratio of the number of genes per MCC bin and the average number per tissue; †, probability of distribution.

Fig. 3

Predictive performance between models using the maternal-blood transcriptome and blood-derived surrogate in all datasets. Dashed lines show the area under receiver operating characteristics curve (AUROC) of 0.5 and the average per dataset among models using the same set of candidate predictors. The best model was evaluated in each set of candidate predictors by the AUROC. If the AUROC interval was ≥0.5 and more than the average in the development and replication datasets, particularly those without an intervention (i.e., vitamin D supplementation), the model was well-replicated. CI, confidence interval; DI-VNN, deep-insight visible neural network; ENR, elastic net regression; GBM, gradient boosting machine; PC, principal component; RF, random forest.

Fig. 4

Emulation of the most predictive biomarkers from the principal component-gradient boosting machine (PC-GBM). The number is the standardized value of the splitting biomarker. A dashed-line arrow from node D to the IRF6 mRNA node is applied only if P2RX7 is not measured. *, not fulfilling the criteria (i.e., top one to 20 of surrogate genes and top one to 5 of blood genes); a, acetylation; EOPE, early-onset preeclampsia (PE); FGR, fetal growth restriction; g, glycosylation; LOPE, late-onset PE, pa, palmitoylation; PE, preeclampsia; ph, phosphorylation; r, ribosylation; u, ubiquitination.

Fig. 5

Networks and pathways in the context of the maternal-fetal interface. We used proteins in the shortest paths connecting all of the input pairs (biomarkers and the surrogate transcriptome as indicated by colored-highlighted names). Nodes represent proteins, for which the same colors of the nearest nodes indicate the same overrepresented pathway. The pathway descriptors are adjacent to the nodes in the same colors. The edges indicate both functional and physical protein associations with the directed paths . The edge color indicates the type of interaction evidence. Proteins that overrepresented vitamin D-related pathways are surrounded by gray-colored highlights, with pointers to the descriptors. The colors of the areas indicate the tissue context. *, edge information instead of the pathway in the STRING database.