Safety in Numbers: Understanding the Effectiveness of Herbals Through Better Science

Owing to my Indian roots, herbal medicine has been a part of my everyday life in the form of ayurveda. As a child, I remember my mum making a concoction of holy-basil, mint and liquorice whenever I had a bad case of sore throat. It worked wonders and the awesome taste meant total ”patient compliance” from my younger self. This early exposure to herbal medicine shaped my doctoral thesis topic that aimed at assessing the impact of dietary and herbal supplements on prescription/OTC drugs. In recent times, herbal medicine is rapidly gaining momentum compared to allopathy (western medicine), because of the comparatively lower costs and the promise of minimal side-effects owing to their natural origins. The global herbal drug market is estimated to generate a whopping USD $111 billion revenue by 2023. Amongst the key segments of the global market, North America dominates in terms of market share and demand.

As a believer in the ancient branches of herbal medicine, their wide-spread use was encouraging. However, as a researcher, I noticed that despite having a huge global impact on public health, herbal products are subjected to minimal regulatory scrutiny. This is especially concerning in the light of increasing legally permissible use of herbal products for recreation such as those derived from cannabis, salvia divinorum, kava and numerous others. Herbal medicinal products are regulated in the United States by the FDA as “foods”, rather than “drugs”, and are categorized as dietary supplements. The manufacturers can sell their products without first or continually seeking approval from the FDA.  The manufacturers are required to ensure that the dose of active components listed on the label match the actual composition in the drug product. However, FDA has found innumerable incidences of contamination in labeled-as-natural products with toxic and prescription-only drugs (e.g. warfarin, phenytoin) to controlled or untested substances, and biological impurities (e.g. E-coli infected marijuana) such as bacterial or viral contaminants. These adulterants can lead to potentially life-threatening adverse events. The FDA has banned the use of supposedly herbal “skinny pills” for weight loss that contained sibutramine – a drug that poses a risk of heart attack and stroke. These are just a few of the plethora of cases of dubious content in modern herbal products

With a simple “FDA has not evaluated this claim” statement (often ignored by most people as it is written in a tiny font size), herbal companies put boisterous statements of being beneficial in treating disorders or deficiencies. While direct claims are not permissible, deceiving general statements about the utility of the herbal products can certainly fool an unaware, trusting consumer. According to a survey conducted by AARP foundation, nearly 49% of adults in the US,  believe that the FDA evaluates the safety and efficacy of dietary supplements before they are sold, while about 36% believe that herbal supplements are subjected to the same level of scrutiny as prescription and over-the-counter medications.

The FDA has the right to monitor dietary supplements and take actions such as issuing warning or recall and discontinuation of the product. However, the regulatory stance on dietary supplements is “considered safe until proven otherwise”, in stark contrast to the “considered unsafe until proven safe” position on medicines.  Thus, much of the regulatory control is exercised too late and in response to incidences of severe adverse events affecting a user base that is already large, and grows exponentially over time.

While the consumption of so called ”dietary”  products is gaining momentum, studies assessing the safety, efficacy and potential to interact with other medications are being conducted at a much slower pace than where the demand for safety should be today. A key limitation of this research is the chemical complexity of the herbal compounds.. A single herbal product can have a multitude of phytoconstituents (plant-originating molecules) with diverse chemical and biological properties that change in combination, at different amounts, and often with unknown or untested levels of understanding regarding toxicity. Depending on the strain of the plant and method of extraction, the relative ratio of concentrations of these active moieties can drastically vary. Reliable detection of these phytoconstituents requires use of  sophisticated bioanalytical instruments and validated methods. Analytical standards are often unavailable or unused by manufacturers. The variability introduced by manufacturing techniques, different in vitro test methods and bioanalytical methods lead to creation of isolated datasets that cannot be used for making meaningful conclusions on the behavior of the herbal drug in the human and animal bodies meant to ingest or otherwise dose these products. Much of the available herbal drug data is collected using animal models, which show poor clinical translatability due to the fundamental fact that animals and humans are not the same.  The meagre human data available is either riddled with variability or conflicting due to unstandardized or incorrect study design and uncontrolled trials. Often these human studies are performed in standardized populations (single age, ethnicity, gender) that require further questioning as to their relevance in understanding efficacy and safety for vulnerable groups (pregnant females, genetically variant ethnicities, pediatric/geriatirc subjects). Ultimately, all of the above factors make it impossible to perform a meta-analysis of the available data to make robust inferences on the true therapeutic potential of these herbal drugs.

To address these issues, computational models are being increasingly used for better clinical interpretation of in vitro and animal data obtained. However, in most cases the model design is very restrictive, in that the predictions are applicable only for a given herbal constituent and potentially its major metabolite. Often, researchers are required to make assumptions about the “major” active component(s) for simplification but this approach fails when a compound present in minor amounts contributes to toxicity or unintended drug-drug interaction. In vitro to in vivo extrapolation using data from high throughput in vitro screenings is helpful in quick phenotyping of the metabolic pathways and potential interaction targets. However, such studies are performed only on single compounds in lieu of testing the actual combination of phytoconstituents as they exist in the final product. Impact of binding to microsomal/cellular protein and labware are seldom considered resulting in significant underestimation of their interaction potential.

In the recent years, significant efforts are being taken to build a standardized framework for computational modeling of herb-drug interactions. One of these remarkable endeavors is the new reality for in silico models through the application of BIOiSIM – VeriSIM Life’s revolutionary platform that integrates the power of machine learning with physiological modeling for faster and better predictions. I could not be more excited to be part of the VeriSIM to develop BIOiSIM that is designed for predicting drug-to-drug and drug-to-herb interactions more accurately and intuitively through the use of models integrated with complex machine algorithms. We train our ML algorithms on biological outputs using a comprehensive database of physicochemically diverse proprietary and non-proprietary compounds. The ML feature is vital when dealing with incomplete datasets, as is often observed. BIOiSIM is ideally suited to perform simulations on combinations of phytoconstituents rather than doing one-off predictions under the assumption of one or two major active components. With minimal in vitro testing, BIOiSIM can provide mechanistic insights on the interaction liability of new drugs.    Additionally, transfer learning modules can prove vital in creating a holistic dataset obtained from multiple sources and using different experimental protocols. BIOiSIM can prove extremely useful not just for pharmaceutical development but also to medical practitioners and pharmacists to recommend appropriate dosing and educate consumers on the safe use of herbal drugs with or without prescription/OTC medication.

   To conclude, I believe there are bad and good actors making these herbal medicines, which take so many forms and are used by countless numbers of people all over the world. The good actors are confused by what they know and don’t know, the bad ones willfully ignore safety and human health while the regulatory bodies face a challenging task of straddling both.  BIOiSIM can help course correct the current herbal pharma landscape, to better understand the relationship between traditional medicine and supplements. VSL’s tech can help in applying the ancient knowledge of herbal medicine to develop safe and effective drug products.