Spatiotemporal association of low birth weight with Cs-137 deposition at the prefecture level in Japan after the Fukushima nuclear power plant accidents: an analytical-ecologic epidemiological study

Development and evaluation of machine learning models for predicting large-for-gestational-age newborns in women exposed to radiation prior to pregnancy

INTRODUCTION

The correlation between radiation exposure before pregnancy and abnormal birth weight has been previously proven. However, for large-for-gestational-age (LGA) babies in women exposed to radiation before becoming pregnant, there is no prediction model yet.

MATERIAL AND METHODS

The data were collected from the National Free Preconception Health Examination Project in China. A sum of 455 neonates (42 SGA births and 423 non-LGA births) were included. A training set (n = 319) and a test set (n = 136) were created from the dataset at random. To develop prediction models for LGA neonates, conventional logistic regression (LR) method and six machine learning methods were used in this study. Recursive feature elimination approach was performed by choosing 10 features which made a big contribution to the prediction models. And the Shapley Additive Explanation model was applied to interpret the most important characteristics that affected forecast outputs.

RESULTS

The random forest (RF) model had the highest average area under the receiver-operating-characteristic curve (AUC) for predicting LGA in the test set (0.843, 95% confidence interval [CI]: 0.714-0.974). Except for the logistic regression model (AUC: 0.603, 95%CI: 0.440-0.767), other models' AUCs displayed well. Thereinto, the RF algorithm's final prediction model using 10 characteristics achieved an average AUC of 0.821 (95% CI: 0.693-0.949).

CONCLUSION

The prediction model based on machine learning might be a promising tool for the prenatal prediction of LGA births in women with radiation exposure before pregnancy.

Maternal urinary metal(loid)s and risk of preterm birth: A cohort study in the Tibetan Plateau

Maternal metal(loid)s exposure has been related to preterm birth (PTB), but the results are still inconclusive. Previous studies have mainly discussed the harmful metal(loid)s, neglecting beneficial ones. We examined the association of maternal metal(loid)s with PTB and gestational age (GA) in a birth cohort from the Tibetan Plateau. We measured 29 metal(loid)s in urine samples from 1081 pregnant women in the third trimester. Information regarding demographics, socioeconomic status, diet, medication, and lifestyle was collected through standardized interviews. The associations of single metal(loid)s with PTB or GA were evaluated using a generalized linear mixed-effects model or linear mixed-effects model. Elastic net and Bayesian kernel machine regressions were used to explore the joint associations. Magnesium (Mg), Copper (Cu), and Tin (Sn) were the main "harmful" metal(loid)s positively and negatively associated with PTB or GA, respectively. Mg was the dominant "harmful" metal(loid)s associated with PTB in a J-shape. A one-fold increase in Mg was associated with a 38% increased risk of PTB [OR (95% CI) = 1.38 (1.15, 1.65), PFDR＜0.05] and 0.17 weeks shortening of GA [β (95% CI) = -0.25 (-0.35, -0.14), PFDR＜0.05]. Cesium (Cs), rubidium (Rb), and Molybdenum (Mo) were the main "beneficial" metals. Cs dominated the "beneficial" associations and was negatively associated with PTB in a linear manner. A one-fold increase in Cs was associated with a 67% decreased risk of PTB [OR (95% CI) = 0.43 (0.27, 0.67), PFDR＜0.05] and 0.24 weeks of prolonged GA [β (95% CI) = 0.35 (0.13, 0.56), PFDR＜0.05]. Ethnicity and living altitude modified the association of Mg and Cu with PTB or GA. In conclusion, Maternal urinary metal(loid)s were bi-directionally associated with PTB in a population in the Tibetan Plateau. Mg and Cs were the dominant "harmful" and "beneficial" metal(loid)s, respectively.

Influence of radiation exposure to delayed fetal growth in wild Japanese monkeys after the Fukushima accident

Wild Japanese monkeys (Macaca fuscata) were exposed to radiation after the Fukushima Daiichi nuclear accident in 2011. To clarify the biological effects of radiation exposure on their fetal growth, pregnant monkeys and their fetuses were analyzed. These animals were collected between 2008 and 2020 (before and after the accident in 2011) living in Fukushima City, approximately 70 km from the nuclear power plant. Multiple regression analyses were conducted with fetal body weight (FBW) and fetal head circumference (FHS) as objective variables, and maternal and fetal factors as explanatory variables. The maternal factors were relative exposure dose rate (REDR), age, body weight, body length, fat index, and parity. The fetal factors were crown ramp length (CRL) and sex. Multiple regression analyses showed that FBR and FHS growth were positively associated with CRL, maternal body length, and negatively associated with REDR. Since the relative growth of FBR and FHS to CRL decreased with increasing REDR, radiation exposure due to the nuclear accident may have contributed to the delayed fetal growth observed in Japanese monkeys.

Low-dose ionizing radiation and adverse birth outcomes: a systematic review and meta-analysis

OBJECTIVE

Ionizing radiation is a human carcinogen, and there is evidence that exposure to low-dose ionizing radiation increases the risk of adverse birth outcomes.

METHODS

We undertook a systematic review and meta-analysis to synthesize the research of maternal and paternal exposure to low-dose radiation on low birth weight, miscarriage, pre-term delivery, and stillbirth. Our literature search used four databases (PubMed, Environmental Index, GeoBASE, and the Cumulative Index to Nursing and Allied Health Literature). We included study populations exposed to occupational and medical sources of radiation, nuclear disasters, and those living near nuclear power plants. We considered papers published between January 1st, 1990, and June 30th, 2021. The quality of the studies was assessed, and we performed meta-analysis using random effects models to generate summary measures of association. Forest plots were created to assess the heterogeneity in these measures, and funnel plots were used to assess publication bias.

RESULTS

Overall, 26 studies were identified, and these yielded measures of association from 10, 11, and 8 studies for low birth weight, miscarriage, and stillbirth outcomes, respectively. It was not possible to perform meta-analyses for pre-term delivery due to a small number of studies. The meta-analysis summary relative risk (RR) of having a low-birth-weight infant among those ever exposed to radiation relative to those unexposed, after adjusting for publication bias, was 1.29 (95% CI 0.97-1.73). The corresponding risk estimates for miscarriage and stillbirth were 1.15 (95% CI 1.02-1.30), and 1.19 (95% CI 0.98-1.45), respectively.

CONCLUSIONS

Our findings suggest that ionizing radiation increases the risk of adverse birth outcomes. Future work should strive to provide data needed to better understand the shape of the exposure-response curve.

Effects of External Radiation Exposure on Perinatal Outcomes in Pregnant Women After the Fukushima Daiichi Nuclear Power Plant Accident: the Fukushima Health Management Survey

BACKGROUND

This study aimed to investigate the effects of maternal exposure to external radiation on perinatal outcomes among women who experienced the Fukushima Daiichi Nuclear Disaster (FDND) using the Fukushima Health Management Survey (FHMS).

METHODS

Data from the Pregnancy and Birth Survey and Basic Survey in the FHMS were combined to analyze external maternal radiation exposure following the FDND, and the relationship between radiation dose and perinatal outcomes was analyzed using binomial logistic regression analysis. Missing dose data were supplemented using multiple imputation.

RESULTS

A total of 6,875 individuals responded to the survey. Congenital anomalies occurred in 2.9% of patients, low birth weight (LBW) in 7.6%, small for gestation age (SGA; <10th percentile) in 8.9%, and preterm birth in 4.1%. The median maternal external radiation dose was 0.5 mSv (maximum, 5.2 mSv). Doses were classified as follows: <1 mSv (reference), 1 to <2 mSv, and ≥2 mSv. For congenital anomalies, the crude odds ratio for 1 to <2 mSv was 0.81 (95% confidence interval [CI], 0.56-1.17) (no participants with congenital anomaly were exposed to ≥2 mSv). At 1 to <2 mSv and ≥2 mSv, the respective adjusted odds ratios were 0.91 (95% CI, 0.71-1.18) and 1.21 (95% CI, 0.53-2.79) for LBW, 1.14 (95% CI, 0.92-1.42) and 0.84 (95% CI, 0.30-2.37) for SGA, and 0.91 (95% CI, 0.65-1.29) and 1.05 (95% CI, 0.22-4.87) for preterm birth.

CONCLUSION

External radiation dose due to the FDND was not associated with congenital anomalies, LBW, SGA, or preterm birth.

Network exploration of gene signatures underlying low birth weight induced metabolic alterations

BACKGROUND

This study explored underlying gene signatures of low birth weight (LBW) by analyzing differentially expressed genes (DEGs) between LBW and normal birth weight (NBW) subjects.

METHODS

Subjects with different birth weight was collected from GEO database. P < .05 and | logFC | ≥ 1.0 were used for screening DEGs. David (2021 Update) was used to perform GO annotation and KEGG signaling pathway enrichment analysis. The protein-protein interaction network of DEGs was constructed using the STRING database, in which hub genes were mined through Cytoscape software.

RESULTS

A total of 326 DEGs were identified, including 287 up-regulated genes and 39 down-regulated genes. The GO biological processes enriched by DEGs mainly involved epidermal growth, keratinization and intermediate fibrous tissue. The DEGs were significantly enriched in intracellular insoluble membranes, desmosomes and extracellular space. Their molecular functions mainly focused on structural molecular activity, structural components of epidermis and structural components of cytoskeleton. PI3K/AKT signaling pathway and tight junction were highlighted as critical pathways enriched by DEGs. Ten hub genes which included KRT14, EGF, DSP, DSG1, KRT16, KRT6A, EPCAM, SPRR1B, PKP1, and PPL were identified from the constructed protein-protein interaction network.

CONCLUSION

A total of 326 DEGs and 10 hub genes were identified as candidates for metabolic disorders in LBW individuals. Our results indicated PI3K/AKT signaling pathway as an intrauterine adaptive mechanism for LBW individuals. We observed activated PI3K/AKT pathway in LBW individuals, which would promote growth and development at the early stage of life, but adversely introduce extra metabolic stress and thereby potentially induce metabolic disorders in adulthood.

Development and Evaluation of a Machine Learning Prediction Model for Small-for-Gestational-Age Births in Women Exposed to Radiation before Pregnancy

Exposure to radiation has been associated with increased risk of delivering small-for-gestational-age (SGA) newborns. There are no tools to predict SGA newborns in pregnant women exposed to radiation before pregnancy. Here, we aimed to develop an array of machine learning (ML) models to predict SGA newborns in women exposed to radiation before pregnancy. Patients’ data was obtained from the National Free Preconception Health Examination Project from 2010 to 2012. The data were randomly divided into a training dataset (n = 364) and a testing dataset (n = 91). Eight various ML models were compared for solving the binary classification of SGA prediction, followed by a post hoc explainability based on the SHAP model to identify and interpret the most important features that contribute to the prediction outcome. A total of 455 newborns were included, with the occurrence of 60 SGA births (13.2%). Overall, the model obtained by extreme gradient boosting (XGBoost) achieved the highest area under the receiver-operating-characteristic curve (AUC) in the testing set (0.844, 95% confidence interval (CI): 0.713–0.974). All models showed satisfied AUCs, except for the logistic regression model (AUC: 0.561, 95% CI: 0.355–0.768). After feature selection by recursive feature elimination (RFE), 15 features were included in the final prediction model using the XGBoost algorithm, with an AUC of 0.821 (95% CI: 0.650–0.993). ML algorithms can generate robust models to predict SGA newborns in pregnant women exposed to radiation before pregnancy, which may thus be used as a prediction tool for SGA newborns in high-risk pregnant women.

Methodological considerations for interrupted time series analysis in radiation epidemiology: an overview

Interrupted time series analysis (ITSA) is a method that can be applied to evaluate health outcomes in populations exposed to ionizing radiation following major radiological events. Using aggregated time series data, ITSA evaluates whether the time trend of a health indicator shows a change associated with the radiological event. That is, ITSA checks whether there is a statistically significant discrepancy between the projection of a pre-event trend and the data empirically observed after the event. Conducting ITSA requires one to consider specific methodological issues due to unique threats to internal validity that make ITSA prone to bias. We here discuss the strengths and limitations of ITSA with respect to bias and confounding, data quality, and statistical aspects. We provide recommendations to strengthen the robustness of ITSA studies and reduce their susceptibility to producing spurious results as a consequence of arbitrary modelling decisions.

Letter to the editor regarding "spatiotemporal association of low birth weight with Cs-137 deposition at the prefecture level in Japan after the Fukushima nuclear power plant accidents"

In the previous report, association between increased low birth weight prevalence and radiocesium deposition after 2011 Fukushima nuclear disaster was found. However, the statistical analyses therein raised several questions. First, ecological variables are not justified enough to adjust potential confounding. Second, the spatiotemporal regression model does not consider temporal reduction in radiation dose rate. Third, dose-response plot between dose rates and odds ratios overestimates R² and underestimates p value.

Response to the "letter to the editor" by Sani Rachman Soleman et al., "spatiotemporal association of low birth weight with Cs-137 deposition at the prefecture level in Japan after the Fukushima nuclear power plant accidents"

We thank Sani Rachman Soleman et al. for three specific points of criticism concerning our investigation of the ecological association between low birth weight (LBW) and radioactive contamination in Japan after the Fukushima Daiichi Nuclear Power Plant (FDNPP) accidents: 1. Ecological variables are not justified enough to adjust potential confounding. 2. The spatiotemporal regression model does not consider temporal reduction in radiation dose rate. 3. Dose-response plot between dose rates and odds ratios overestimates R2 and underestimates p-value. This criticism is a good starting point to explain some of the technical backgrounds of our approach in more detail.

Short term increase in low birth weight babies after Fukushima

An analysis of low birth weight (LBW) births in ten contaminated prefectures of Japan, 1995-2018, finds a statistically significant increase in the LBW proportion in 2012-2013, but no increase after 2013. In the rest of Japan (37 prefectures), no increase in LBW births was found after the Fukushima accident.