Current applications of artificial intelligence in assisted reproductive technologies through the perspective of a patient's journey

PURPOSE OF REVIEW

This review highlights the timely relevance of artificial intelligence in enhancing assisted reproductive technologies (ARTs), particularly in-vitro fertilization (IVF). It underscores artificial intelligence's potential in revolutionizing patient outcomes and operational efficiency by addressing challenges in fertility diagnoses and procedures.

RECENT FINDINGS

Recent advancements in artificial intelligence, including machine learning and predictive modeling, are making significant strides in optimizing IVF processes such as medication dosing, scheduling, and embryological assessments. Innovations include artificial intelligence augmented diagnostic testing, predictive modeling for treatment outcomes, scheduling optimization, dosing and protocol selection, follicular and hormone monitoring, trigger timing, and improved embryo selection. These developments promise to refine treatment approaches, enhance patient engagement, and increase the accuracy and scalability of fertility treatments.

SUMMARY

The integration of artificial intelligence into reproductive medicine offers profound implications for clinical practice and research. By facilitating personalized treatment plans, standardizing procedures, and improving the efficiency of fertility clinics, artificial intelligence technologies pave the way for value-based, accessible, and efficient fertility services. Despite the promise, the full potential of artificial intelligence in ART will require ongoing validation and ethical considerations to ensure equitable and effective implementation.

Live birth outcomes following single-step blastocyst warming technique - optimizing efficiency without impacting live birth rates

PURPOSE

To evaluate the impact of a single-step (SS) warming versus standard warming (SW) protocol on the survival/expansion of vitrified blastocysts and their clinical outcomes post-frozen embryo transfer (FET).

METHODS

Retrospective analysis was performed on 200 vitrified/warmed research blastocysts equally divided amongst two thawing protocols utilizing the Fujifilm Warming NX kits (Fujifilm, CA). SW utilized the standard 14-minute manufacturer's guidelines. SS protocol required only a one-minute immersion in thaw solution (TS) before the embryos were transferred to culture media. A time-interrupted study was performed evaluating 752 FETs (SW: 376 FETs, SS 376 FETs) between April 2021-December 2022 at a single academic fertility clinic in Boston, Massachusetts. Embryologic, clinical pregnancy, and live birth outcomes were assessed using generalized estimated equation (GEE) models, which accounted for potential confounders.

RESULTS

There was 100% survival for all blastocysts (n = 952 embryos) with no differences in blastocyst re-expansion regardless of PGT status. Adjusted analysis showed no differences in implantation, clinical pregnancy, spontaneous abortion, or biochemical pregnancy rate. A higher odds of multiple gestation [AdjOR(95%CI) 1.06 (1.01, 1.11), p = 0.019] were noted, even when adjusting for number of embryos transferred [AdjOR(95%CI) 1.05 (1.01, 1.10)]. Live birth outcomes showed no differences in live birth rates or birthweight at delivery.

CONCLUSIONS

The study found equivalent outcomes for SS and SW in all parameters except for a slight rise in the rate of multiple gestations. The results suggest that SS warming is an efficient, viable alternative to SW, reducing thaw times without adverse effects on live birth rates or neonatal birth weights.

Predicting polycystic ovary syndrome with machine learning algorithms from electronic health records

Introduction

Predictive models have been used to aid early diagnosis of PCOS, though existing models are based on small sample sizes and limited to fertility clinic populations. We built a predictive model using machine learning algorithms based on an outpatient population at risk for PCOS to predict risk and facilitate earlier diagnosis, particularly among those who meet diagnostic criteria but have not received a diagnosis.

Methods

This is a retrospective cohort study from a SafetyNet hospital's electronic health records (EHR) from 2003-2016. The study population included 30,601 women aged 18-45 years without concurrent endocrinopathy who had any visit to Boston Medical Center for primary care, obstetrics and gynecology, endocrinology, family medicine, or general internal medicine. Four prediction outcomes were assessed for PCOS. The first outcome was PCOS ICD-9 diagnosis with additional model outcomes of algorithm-defined PCOS. The latter was based on Rotterdam criteria and merging laboratory values, radiographic imaging, and ICD data from the EHR to define irregular menstruation, hyperandrogenism, and polycystic ovarian morphology on ultrasound.

Results

We developed predictive models using four machine learning methods: logistic regression, supported vector machine, gradient boosted trees, and random forests. Hormone values (follicle-stimulating hormone, luteinizing hormone, estradiol, and sex hormone binding globulin) were combined to create a multilayer perceptron score using a neural network classifier. Prediction of PCOS prior to clinical diagnosis in an out-of-sample test set of patients achieved an average AUC of 85%, 81%, 80%, and 82%, respectively in Models I, II, III and IV. Significant positive predictors of PCOS diagnosis across models included hormone levels and obesity; negative predictors included gravidity and positive bHCG.

Conclusion

Machine learning algorithms were used to predict PCOS based on a large at-risk population. This approach may guide early detection of PCOS within EHR-interfaced populations to facilitate counseling and interventions that may reduce long-term health consequences. Our model illustrates the potential benefits of an artificial intelligence-enabled provider assistance tool that can be integrated into the EHR to reduce delays in diagnosis. However, model validation in other hospital-based populations is necessary.

The Role of Artificial Intelligence and Machine Learning in Assisted Reproductive Technologies

Artificial intelligence (AI) and machine learning, the form most commonly used in medicine, offer powerful tools utilizing the strengths of large data sets and intelligent algorithms. These systems can help to revolutionize delivery of treatments, access to medical care, and improvement of outcomes, particularly in the realm of reproductive medicine. Whether that is more robust oocyte and embryo grading or more accurate follicular measurement, AI will be able to aid clinicians, and most importantly patients, in providing the best possible and individualized care. However, despite all of the potential strengths of AI, algorithms are not immune to bias and are vulnerable to the many socioeconomic and demographic biases that current healthcare systems suffer from. Wrong diagnoses as well is furthering of healthcare discrimination are real possibilities if both the capabilities and limitations of AI are not well understood. Armed with appropriate knowledge of how AI can most appropriately operate within medicine, and specifically reproductive medicine, will enable clinicians to both create and utilize machine learning-based innovations for the furthering of reproductive medicine and ultimately achieving the goal of building of healthy families.

Temporal trends in urinary concentrations of phenols, phthalate metabolites and phthalate replacements between 2000 and 2017 in Boston, MA

Endocrine disrupting chemicals (EDCs) can adversely affect human health and are ubiquitously found in everyday products. We examined temporal trends in urinary concentrations of EDCs and their replacements. Urinary concentrations of 11 environmental phenols, 15 phthalate metabolites, phthalate replacements such as two di(isononyl)cyclohexane-1,2-dicarboxylate (DINCH) metabolites, and triclocarban were quantified using isotope-dilution tandem mass spectrometry. This ecological study included 996 male and 819 female patients who were predominantly White/Caucasian (83 %) with an average age of 35 years and a BMI of 25.5 kg/m2 seeking fertility treatment in Boston, MA, USA. Patients provided a total of 6483 urine samples (median = 2, range = 1-30 samples per patient) between 2000 and 2017. Over the study period, we observed significant decreases (% per year) in urinary concentrations of traditional phenols, parabens, and phthalates such as bisphenol A (β: -6.3, 95 % CI: -7.2, -5.4), benzophenone-3 (β: -6.5, 95 % CI: -1.1, -18.9), parabens ((β range:-5.4 to -14.2), triclosan (β: -18.8, 95 % CI: -24, -13.6), dichlorophenols (2.4-dichlorophenol β: -6.6, 95 % CI: -8.8, -4.3); 2,5-dichlorophenol β: -13.6, 95 % CI: -17, -10.3), di(2-ethylhexyl) phthalate metabolites (β range: -11.9 to -22.0), and other phthalate metabolites including mono-ethyl, mono-n-butyl, and mono-methyl phthalate (β range: -0.3 to -11.5). In contrast, we found significant increases in urinary concentrations of environmental phenol replacements including bisphenol S (β: 3.9, 95 % CI: 2.7, 7.6) and bisphenol F (β: 6, 95 % CI: 1.8, 10.3), DINCH metabolites (cyclohexane-1,2-dicarboxylic acid monohydroxy isononyl ester [MHiNCH] β: 20, 95 % CI: 17.8, 22.2; monocarboxyisooctyl phthalate [MCOCH] β: 16.2, 95 % CI: 14, 18.4), and newer phthalate replacements such as mono-3-carboxypropyl phthalate, monobenzyl phthalate, mono-2-ethyl-5-carboxypentyl phthalate and di-isobutyl phthalate metabolites (β range = 5.3 to 45.1), over time. Urinary MHBP concentrations remained stable over the study period. While the majority of biomarkers measured declined over time, concentrations of several increased, particularly replacement chemicals that are studied.

Predicting polycystic ovary syndrome (PCOS) with machine learning algorithms from electronic health records

Introduction

Predictive models have been used to aid early diagnosis of PCOS, though existing models are based on small sample sizes and limited to fertility clinic populations. We built a predictive model using machine learning algorithms based on an outpatient population at risk for PCOS to predict risk and facilitate earlier diagnosis, particularly among those who meet diagnostic criteria but have not received a diagnosis.

Methods

This is a retrospective cohort study from a SafetyNet hospital's electronic health records (EHR) from 2003-2016. The study population included 30,601 women aged 18-45 years without concurrent endocrinopathy who had any visit to Boston Medical Center for primary care, obstetrics and gynecology, endocrinology, family medicine, or general internal medicine. Four prediction outcomes were assessed for PCOS. The first outcome was PCOS ICD-9 diagnosis with additional model outcomes of algorithm-defined PCOS. The latter was based on Rotterdam criteria and merging laboratory values, radiographic imaging, and ICD data from the EHR to define irregular menstruation, hyperandrogenism, and polycystic ovarian morphology on ultrasound.

Results

We developed predictive models using four machine learning methods: logistic regression, supported vector machine, gradient boosted trees, and random forests. Hormone values (follicle-stimulating hormone, luteinizing hormone, estradiol, and sex hormone binding globulin) were combined to create a multilayer perceptron score using a neural network classifier. Prediction of PCOS prior to clinical diagnosis in an out-of-sample test set of patients achieved AUC of 85%, 81%, 80%, and 82%, respectively in Models I, II, III and IV. Significant positive predictors of PCOS diagnosis across models included hormone levels and obesity; negative predictors included gravidity and positive bHCG.

Conclusions

Machine learning algorithms were used to predict PCOS based on a large at-risk population. This approach may guide early detection of PCOS within EHR-interfaced populations to facilitate counseling and interventions that may reduce long-term health consequences. Our model illustrates the potential benefits of an artificial intelligence-enabled provider assistance tool that can be integrated into the EHR to reduce delays in diagnosis. However, model validation in other hospital-based populations is necessary.

Non-invasive Genetic Screening: Current Advances in Artificial Intelligence for Embryo Ploidy Prediction

This review discusses the use of artificial intelligence (AI) algorithms in non-invasive prediction of embryo ploidy status for preimplantation genetic testing (PGT) in in vitro fertilization (IVF) procedures. The current gold standard, preimplantation genetic testing for aneuploidy (PGT-A), has limitations such as invasive biopsy, financial burden, delays in reporting results, and difficulty in interpreting results. Non-invasive ploidy screening methods, including blastocoel fluid sampling, spent media testing, and AI algorithms using embryonic images and clinical parameters, are explored. Various AI models have been developed using different machine learning (ML) algorithms, such as random forest classifier (RFC) and logistic regression (MLR) and have shown variable performance in predicting euploidy. Static embryo imaging combined with AI algorithms has demonstrated good accuracy in ploidy prediction, with models like ERICA and STORK-A outperforming human grading. Time-lapse embryo imaging analyzed by AI algorithms has also shown promise in predicting ploidy status. However, the inclusion of clinical parameters is crucial to improve the predictive value of these models. Mosaicism, an important aspect of embryo classification, is often overlooked in AI algorithms and should be considered in future studies. The integration of AI algorithms into microscopy equipment and Embryoscope platforms will facilitate non-invasive genetic testing. Further development of algorithms that optimize clinical considerations and incorporate minimal necessary covariates will enhance the predictive value of AI in embryo selection. AI-based ploidy prediction has the potential to improve pregnancy rates and reduce costs in IVF cycles.

The impact of cryopreserved sperm on intrauterine insemination outcomes: is frozen as good as fresh?

Introduction

Frozen sperm utilization might negatively impact cycle outcomes in animals, implicating cryopreservation-induced sperm damage. However, in vitro fertilization and intrauterine insemination (IUI) in human studies are inconclusive.

Methods

This study is a retrospective review of 5,335 IUI [± ovarian stimulation (OS)] cycles from a large academic fertility center. Cycles were stratified based on the utilization of frozen (FROZEN, n = 1,871) instead of fresh ejaculated sperm (FRESH, n = 3,464). Main outcomes included human chorionic gonadotropin (HCG) positivity, clinical pregnancy (CP), and spontaneous abortion (SAB) rates. Secondary outcome was live birth (LB) rate. Odds ratios (OR) for all outcomes were calculated utilizing logistic regression and adjusted (adjOR) for maternal age, day-3 FSH, and OS regimen. Stratified analysis was performed based on OS subtype [gonadotropins; oral medications (OM): clomiphene citrate and letrozole; and unstimulated/natural]. Time to pregnancy and cumulative pregnancy rates were also calculated. Further subanalyses were performed limited to either the first cycle only or to the partner's sperm only, after excluding female factor infertility, and after stratification by female age (<30, 30-35, and >35 years old).

Results

Overall, HCG positivity and CP were lower in the FROZEN compared to the FRESH group (12.2% vs. 15.6%, p < 0.001; 9.4% vs. 13.0%, p < 0.001, respectively), which persisted only among OM cycles after stratification (9.9% vs. 14.2% HCG positivity, p = 0.030; 8.1% vs. 11.8% CP, p = 0.041). Among all cycles, adjOR (95% CI) for HCG positivity and CP were 0.75 (0.56-1.02) and 0.77 (0.57-1.03), respectively, ref: FRESH. In OM cycles, adjOR (95% CI) for HCG positivity [0.55 (0.30-0.99)] and CP [0.49 (0.25-0.95), ref.: FRESH] favored the FRESH group but showed no differences among gonadotropin and natural cycles. SAB odds did not differ between groups among OM and natural cycles but were lower in the FROZEN group among gonadotropin cycles [adjOR (95% CI): 0.13 (0.02-0.98), ref.: FRESH]. There were no differences in CP and SAB in the performed subanalyses (limited to first cycles or partner's sperm only, after excluding female factors, or after stratification according to female age). Nevertheless, time to conception was slightly longer in the FROZEN compared to the FRESH group (3.84 vs. 2.58 cycles, p < 0.001). No significant differences were present in LB and cumulative pregnancy results, other than in the subgroup of natural cycles, where higher LB odds [adjOR (95% CI): 1.08 (1.05-1.12)] and higher cumulative pregnancy rate (34% vs. 15%, p = 0.002) were noted in the FROZEN compared to the FRESH group.

Conclusion

Overall, clinical outcomes did not differ significantly between frozen and fresh sperm IUI cycles, although specific subgroups might benefit from fresh sperm utilization.

Applications of Artificial Intelligence in Ovarian Stimulation: A Tool for Improving Efficiency and Outcomes

Since the birth of the first baby following in vitro fertility (IVF), the field of assisted reproductive technologies (ART) has seen significant advancements in the past 40 years. Over the last decade, the healthcare industry has increasingly adopted machine learning (ML) algorithms to improve patient care and operational efficiency. Artificial intelligence (AI) in ovarian stimulation is a burgeoning niche which is currently benefiting from increased research and investment from both the scientific and technology communities, leading to cutting-edge advancements with promise for rapid clinical integration. AI-assisted IVF is a rapidly growing area of research that can improve ovarian stimulation outcomes and efficiency by optimizing the dosage and timing of medications, streamlining the IVF process, and ultimately leading to increased standardization and better clinical outcomes. This review article aims to shed light on the latest breakthroughs in this area, discuss the role of validation and potential limitations of the technology, and examine the potential of these technologies to transform the field of ART. Integrating AI responsibly into IVF stimulation will result in providing higher value clinical care with the goal of having a meaningful impact on enhancing access to more successful and efficient fertility treatments.

Quality assurance (QA) for monitoring the performance of assisted reproductive technology (ART) staff using artificial intelligence (AI)

PURPOSE

Deep learning neural networks have been used to predict the developmental fate and implantation potential of embryos with high accuracy. Such networks have been used as an assistive quality assurance (QA) tool to identify perturbations in the embryo culture environment which may impact clinical outcomes. The present study aimed to evaluate the utility of an AI-QA tool to consistently monitor ART staff performance (MD and embryologist) in embryo transfer (ET), embryo vitrification (EV), embryo warming (EW), and trophectoderm biopsy (TBx).

METHODS

Pregnancy outcomes from groups of 20 consecutive elective single day 5 blastocyst transfers were evaluated for the following procedures: MD performed ET (N = 160 transfers), embryologist performed ET (N = 160 transfers), embryologist performed EV (N = 160 vitrification procedures), embryologist performed EW (N = 160 warming procedures), and embryologist performed TBx (N = 120 biopsies). AI-generated implantation probabilities for the same embryo cohorts were estimated, as were mean AI-predicted and actual implantation rates for each provider and compared using Wilcoxon singed-rank test.

RESULTS

Actual implantation rates following ET performed by one MD provider: "H" was significantly lower than AI-predicted (20% vs. 61%, p = 0.001). Similar results were observed for one embryologist, "H" (30% vs. 60%, p = 0.011). Embryos thawed by embryologist "H" had lower implantation rates compared to AI prediction (25% vs. 60%, p = 0.004). There were no significant differences between actual and AI-predicted implantation rates for EV, TBx, or for the rest of the clinical staff performing ET or EW.

CONCLUSIONS

AI-based QA tools could provide accurate, reproducible, and efficient staff performance monitoring in an ART practice.

Advancements in the future of automating micromanipulation techniques in the IVF laboratory using deep convolutional neural networks

PURPOSE

To determine if deep learning artificial intelligence algorithms can be used to accurately identify key morphologic landmarks on oocytes and cleavage stage embryo images for micromanipulation procedures such as intracytoplasmic sperm injection (ICSI) or assisted hatching (AH).

METHODS

Two convolutional neural network (CNN) models were trained, validated, and tested over three replicates to identify key morphologic landmarks used to guide embryologists when performing micromanipulation procedures. The first model (CNN-ICSI) was trained (n = 13,992), validated (n = 1920), and tested (n = 3900) to identify the optimal location for ICSI through polar body identification. The second model (CNN-AH) was trained (n = 13,908), validated (n = 1908), and tested (n = 3888) to identify the optimal location for AH on the zona pellucida that maximizes distance from healthy blastomeres.

RESULTS

The CNN-ICSI model accurately identified the polar body and corresponding optimal ICSI location with 98.9% accuracy (95% CI 98.5-99.2%) with a receiver operator characteristic (ROC) with micro and macro area under the curves (AUC) of 1. The CNN-AH model accurately identified the optimal AH location with 99.41% accuracy (95% CI 99.11-99.62%) with a ROC with micro and macro AUCs of 1.

CONCLUSION

Deep CNN models demonstrate powerful potential in accurately identifying key landmarks on oocytes and cleavage stage embryos for micromanipulation. These findings are novel, essential stepping stones in the automation of micromanipulation procedures.

The use of voting ensembles to improve the accuracy of deep neural networks as a non-invasive method to predict embryo ploidy status

PURPOSE

To determine if creating voting ensembles combining convolutional neural networks (CNN), support vector machine (SVM), and multi-layer neural networks (NN) alongside clinical parameters improves the accuracy of artificial intelligence (AI) as a non-invasive method for predicting aneuploidy.

METHODS

A cohort of 699 day 5 PGT-A tested blastocysts was used to train, validate, and test a CNN to classify embryos as euploid/aneuploid. All embryos were analyzed using a modified FAST-SeqS next-generation sequencing method. Patient characteristics such as maternal age, AMH level, paternal sperm quality, and total number of normally fertilized (2PN) embryos were processed using SVM and NN. To improve model performance, we created voting ensembles using CNN, SVM, and NN to combine our imaging data with clinical parameter variations. Statistical significance was evaluated with a one-sample t-test with 2 degrees of freedom.

RESULTS

When assessing blastocyst images alone, the CNN test accuracy was 61.2% (± 1.32% SEM, n = 3 models) in correctly classifying euploid/aneuploid embryos (n = 140 embryos). When the best CNN model was assessed as a voting ensemble, the test accuracy improved to 65.0% (AMH; p = 0.1), 66.4% (maternal age; p = 0.06), 65.7% (maternal age, AMH; p = 0.08), 66.4% (maternal age, AMH, number of 2PNs; p = 0.06), and 71.4% (maternal age, AMH, number of 2PNs, sperm quality; p = 0.02) (n = 140 embryos).

CONCLUSIONS

By combining CNNs with patient characteristics, voting ensembles can be created to improve the accuracy of classifying embryos as euploid/aneuploid from CNN alone, allowing for AI to serve as a potential non-invasive method to aid in karyotype screening and selection of embryos.

Female pattern hair loss and polycystic ovarian syndrome: more than just hirsutism

PURPOSE OF REVIEW

To explore the recent updates in the diagnosis, management, and clinical implications of androgenic alopecia among patients diagnosed with polycystic ovarian syndrome (PCOS).

RECENT FINDINGS

PCOS diagnosis continues to be the most common cause of infertility among reproductively aged women, serving as the most common endocrinopathy among this population. Female pattern hair loss (FPHL) has been seen to be associated and more common among patients with PCOS, however, there are limited studies examining the impact of FPHL among PCOS patients. Although hyperandrogenism is associated with FPHL, the pathophysiology continues to be unclear as FPHL can be present with normal biochemical androgen markers. Treatment can be complex, as common treatments to promote hair growth can exacerbate undesired hirsutism, which can be overcome by cosmetic treatments. New second-line treatment options such as low level laser therapy and platelet rich plasma have been emerging, with limited data supporting efficacy.

SUMMARY

PCOS is a complex endocrinological disorder that has significant gynecologic, cutaneous, and metabolic implications that require multidisciplinary collaboration and care. Reproductive goals should be thoroughly discussed prior to starting any treatment, as PCOS is the most common cause of infertility among reproductively-aged women.

Using artificial intelligence to avoid human error in identifying embryos: a retrospective cohort study

PURPOSE

To determine whether convolutional neural networks (CNN) can be used to accurately ascertain the patient identity (ID) of cleavage and blastocyst stage embryos based on image data alone.

METHODS

A CNN model was trained and validated over three replicates on a retrospective cohort of 4889 time-lapse embryo images. The algorithm processed embryo images for each patient and produced a unique identification key that was associated with the patient ID at a timepoint on day 3 (~ 65 hours post-insemination (hpi)) and day 5 (~ 105 hpi) forming our data library. When the algorithm evaluated embryos at a later timepoint on day 3 (~ 70 hpi) and day 5 (~ 110 hpi), it generates another key that was matched with the patient's unique key available in the library. This approach was tested using 400 patient embryo cohorts on day 3 and day 5 and number of correct embryo identifications with the CNN algorithm was measured.

RESULTS

CNN technology matched the patient identification within random pools of 8 patient embryo cohorts on day 3 with 100% accuracy (n = 400 patients; 3 replicates). For day 5 embryo cohorts, the accuracy within random pools of 8 patients was 100% (n = 400 patients; 3 replicates).

CONCLUSIONS

This study describes an artificial intelligence-based approach for embryo identification. This technology offers a robust witnessing step based on unique morphological features of each embryo. This technology can be integrated with existing imaging systems and laboratory protocols to improve specimen tracking.

Thyroid hormone levels associate with exposure to polychlorinated biphenyls and polybrominated biphenyls in adults exposed as children

BACKGROUND

Michigan residents were directly exposed to endocrine-disrupting compounds, polybrominated biphenyl (PBB) and polychlorinated biphenyl (PCB). A growing body of evidence suggests that exposure to certain endocrine-disrupting compounds may affect thyroid function, especially in people exposed as children, but there are conflicting observations. In this study, we extend previous work by examining age of exposure's effect on the relationship between PBB exposure and thyroid function in a large group of individuals exposed to PBB.

METHODS

Linear regression models were used to test the association between serum measures of thyroid function (total thyroxine (T4), total triiodothyronine (T3), free T4, free T3, thyroid stimulating hormone (TSH), and free T3: free T4 ratio) and serum PBB and PCB levels in a cross-sectional analysis of 715 participants in the Michigan PBB Registry.

RESULTS

Higher PBB levels were associated with many thyroid hormones measures, including higher free T3 (p = 0.002), lower free T4 (p = 0.01), and higher free T3: free T4 ratio (p = 0.0001). Higher PCB levels were associated with higher free T4 (p = 0.0002), and higher free T3: free T4 ratio (p = 0.002). Importantly, the association between PBB and thyroid hormones was dependent on age at exposure. Among people exposed before age 16 (N = 446), higher PBB exposure was associated with higher total T3 (p = 0.01) and free T3 (p = 0.0003), lower free T4 (p = 0.04), and higher free T3: free T4 ratio (p = 0.0001). No significant associations were found among participants who were exposed after age 16. No significant associations were found between TSH and PBB or PCB in any of the analyses conducted.

CONCLUSIONS

This suggests that both PBB and PCB are associated with thyroid function, particularly among those who were exposed as children or prenatally.