Time To Say Goodbye to the Traditional BMI?

In the 1800s, a Belgian mathematician by the name of Adolphe Quetelet sought to define the characteristics of a “normal man” after his involvement in a number of human population studies. He devised the Quetelet index, a formula that calculates the ratio of weight (kg) over height (in meters) squared. The simple calculation would later become known as the “body mass index”, a measure for determining a “healthy” body weight.

BMI is adopted across the medical community for determining disease risk and for informing public health and insurance policies. A person might even be refused surgery if their BMI is above a surgeon’s “cut-off” point. Today, a “normal” BMI is between 18.5–24.9. A person with a BMI below 18.5 is classified as “underweight”, and anyone with a BMI above 30 is categorized as “obese”.

Despite its widespread use, there are issues associated with the traditional BMI formula that have led to its criticism and calls for a novel approach for measuring metabolic health.

Now, researchers from the Institute for Systems Biology (ISB) have created a biological BMI, published in the journal Nature Medicine. The team, led by senior research scientist Dr. Noa Rappaport, conducted multiomics profiling on blood samples from 1,000 individuals enrolled in the now closed Arivale wellness program. Using machine learning models, the researchers generated molecular BMI scores, such as a metabolomics- or proteomics-based BMI.

Rappaport and colleagues found that when individuals made “positive” lifestyle changes, the biological BMI calculations were more responsive than the traditional BMI. Among other insights, the team discovered that individuals with a higher biological BMI, but a normal traditional BMI, were less healthy overall but could lose weight more easily following a lifestyle intervention.

Technology Networks interviewed Rappaport and Dr. Kengo Watanabe, molecular and systems biologist at ISB and first author of the paper, to understand how the biological BMI is calculated and why it could be useful as an alternative measure of health. 

Molly Campbell (MC): Can you explain why the traditional BMI measure is sub-optimal? Why has it been continuously used, despite these issues?

Noa Rappaport (NR): The traditional BMI measure is sub-optimal because it fails to accurately classify about 30% of individuals, so some individuals may appear to have a normal weight but have disrupted metabolic health. Part of the reason for this is that traditional BMI does not account for factors like muscle mass, bone density or fat distribution, which are all relevant to health. Despite these issues, the traditional BMI measure has been continuously used because it is a simple and cost-effective measure of obesity, and it correlates with a number of chronic diseases and all-cause mortality.

MC: What is a biological BMI?

Kengo Watanabe (KW): Biological BMI is a multi-dimensional molecular measure of BMI calculated from blood measurements of proteins, metabolites or clinical labs. It is a more comprehensive and accurate measure of metabolic health compared to the traditional BMI measure, which only considers height and weight. Unlike traditional BMI, biological BMI can identify misclassified individuals with a normal weight but disrupted metabolic health, who may not be currently monitored or treated.

MC: Why did you adopt a multiomic profiling approach in this study?

NR: We adopted a multiomic profiling approach in our study to gather a wealth of health information from blood analysis. This approach allowed us to cast a wider net in analyzing more than 1,100 blood analytes, including proteins, metabolites and clinical lab assays, in conjunction with genetic risk scores and gut microbiome composition, to assess metabolic health status and develop biological BMI scores.

Through analyzing a large number of blood analytes, we were able to identify a broader range of potential biomarkers that may be amenable to intervention through lifestyle changes or other interventions. This approach also provides a more detailed understanding of the complex connections between different molecules and how they influence metabolic health, which can help guide future research and interventions.

MC: Can you explain how machine learning models were used to predict variations of a biological BMI?

KW: We used machine learning algorithms to model the traditional BMI scores by constructing a signature of linear combinations of blood measurements. This way, we created several types of biological BMI scores, such as metabolomics-based BMI, or proteomics-based BMI. These scores were able to account for up to 78% of the variability in BMI, which is twice that of what previous studies were able to achieve using different methods and different features.

MC: You found that when people made positive lifestyle changes, the biological BMI was a more “responsive measure” than traditional BMI. Can you describe some of these lifestyle changes and the results you obtained?

NR: We found that metabolomics-based BMI was a more responsive measure than traditional BMI when people made positive lifestyle changes such as lifestyle coaching, regardless of weight loss. Individuals classified as metabolically unhealthy through metabolomics (but that had a normal traditional BMI) improved their metabolic state more than those who were classified as metabolically healthy. People in the (now closed) Arivale program received personalized coaching from clinical dietitians, which was based on their multiomic data with the goal of optimizing wellness through targeting cardiovascular health, nutrition, gut microbiome health etc.

MC: Is it feasible that the approach used in this study could be widely adopted as an alternative BMI measure?

KW: We see deep molecular profiling as the future of precision medicine, as technological advancements continue to make it more cost-effective and useful for capturing the complex nature of health. While we believe that our approach to developing biological BMI scores could potentially be widely adopted as an alternative to traditional BMI measures, more research is needed to fully validate its effectiveness and ensure its applicability across different populations.

MC: Are there any limitations to the study that you wish to highlight?

NR: Yes, for example, our study looked at only a limited number of biomolecules in the blood, and the findings are restricted to this set. Additionally, the study was not designed as a controlled trial, so it is difficult to say how effective the lifestyle interventions were. Finally, the findings may not apply to other populations as the study focused mostly on white individuals.

MC: What are your next steps in this research space?

KW: Moving forward, our research aims to expand on the predictive power of biological BMI for long-term health outcomes. This presents a challenge due to the limited availability of datasets in this emerging field, but we hope to collaborate with other researchers and institutions to address this issue. Additionally, we plan to investigate how we can further personalize interventions based on a person's biological age, in order to optimize their impact on health outcomes. Overall, we see our study as an important step toward developing more effective strategies for preventing and managing chronic diseases associated with obesity and metabolic health.

Dr. Noa Rappaport and Dr. Kengo Watanabe were speaking to Molly Campbell, Senior Science Writer for Technology Networks.

Reference: Watanabe K, Wilmanski T, Diener C, et al. Multiomic signatures of body mass index identify heterogeneous health phenotypes and responses to a lifestyle intervention. Nat Med. 2023. doi:10.1038/s41591-023-02248-0.