Formulation: Still Challenging Vaccine Developers

While drug discovery gets the headlines, the buzz, and the mega-investments, development is where products actually succeed or fail. Among the several parallel-tracked development activities following a Biologics License Agreement (BLA), formulation presents the greatest number of variables and therefore, at some level, the most potential roadblocks.

With the final goal of turning a molecule into a medicine (or a “drug substance” into a “drug product”), formulation seeks to create a dosage form that preserves or accentuates the positives while minimizing the effects of toxicity and poor bioavailability—and all within the restrictions of the intended dosage form.

Modern vaccines fall into four major categories: live-attenuated, inactivated, toxoid, and a fourth grouping comprising subunit, recombinant, conjugate, and polysaccharide products. Due to space concerns this article only considers traditional antigen-based as opposed to gene-altering products, which carry additional concerns surrounding the vector. Although very little overlap is evident among these categories from the perspective of active ingredient, the similarities are relevant. All vaccine types involve a biological agent, so assuring the product’s stability under conditions likely to be encountered during storage and distribution should be the first goal.

Normally, stability assessment of multi-component vaccines involves the individual assay of all components for which stability data is unavailable, in addition to testing the formulated product. Thus, the immune modulating ingredients (adjuvant plus native or modified antigen), stabilizers, buffers, and excipients all come under scrutiny. Due to the time component of stability testing, few shortcuts are available, so this phase of development often extends into or even beyond phase 1.

Software- and AI-based formulation

Traditional formulation software has long been applied for keeping track of studies employing a design-of-experiment (DOE) approach, and the many variables and variable combinations that apply. These programs, along with statistical and predictive add-ons, offer some guidance and trend elucidation in addition to their main job of maintaining records of hundreds of related experiments. A recent article in Nature Communications described an approach, based on artificial intelligence (AI), for formulating long-lasting injectable-erodible drug formulations. This work, from the University of Toronto, claims the distinction of being the first drug formulation application for AI.

Computer-aided formulation serves to narrow the reasonable available options in a drug formulation exercise, a function in which AI serves admirably. Although it reduces the overall number of experiments and compresses development timelines, it is still an iterative process. Depending on the type of formulation, developers can use as few as two initial, investigatory studies to build their preliminary Bayesian model, which through uncertainty calculations narrows the number of options in each successive step.

According to ChemIntelligence, a drug development software firm, the secrets to success are:

• Apply AI methods early, even when extensive experimental data is unavailable
• Embrace data diversity, that is widely varying concentration values for each ingredient
• Integrate as much “domain knowledge” as possible, for example choice of ingredients, process variables and constraints, and choice of formulation objectives

Drying—vaccines’ “forever” formulation?

The three main formulation modalities for vaccines are frozen, dried (including lyophilized, spray- or vacuum-dried, etc.), and refrigerated, the choice coming down to the most convenient format to support the intended delivery system and anticipated storage requirements. Developers usually arrive at a final formulation empirically, through design-of-experiment exercises that balance optimization of quality attributes with the product’s specific delivery requirements.

Compared with liquid or frozen vaccines, dried formulations have the advantages of longer shelf-life, lower susceptibility to temperature-based degradation, and the potential to support injectable, pulmonary, nasal, oral, or intradermal delivery.

Because the active ingredients vary chemically among vaccine types, they have widely different sensitivities to heat or freezing. Inactivated viruses are the most sensitive since they contain whole organisms, while toxoids and subunit vaccines show the least susceptibility to degradation. Since these are not immutable laws but guidelines, establishing reliable stability data through DOE-type studies is unavoidable.

Vaccine drying methods include spray drying, freeze-drying (lyophilization), spray freeze-drying, vacuum, or air-drying. Drying imparts stability for long-term storage, stockpiling, and transport, but just as significantly (at the expense of an additional reconstitution step) provides opportunities for traditional parenteral delivery as well as less common methods (i.e., nasal, buccal/oral, topical, or inhalable dosage forms). Needle-less delivery of injectable drugs, particularly vaccines and insulin, has at various times been deemed a significant unmet need due to “needle phobia” and the skill set required to administer an injectable drug.

A properly dried vaccine retains the liquid-formulated product’s potency and safety, but getting there involves some work. Assuming all excipients, buffers, and adjuvants experience no unexpected excursions from their advertised quality, developers must consider the antigen’s susceptibility to the several steps involved in drying, e.g., exposure to heat, cold, shear forces, and dehydration itself. For example, the first step in lyophilization is freezing.

When stability studies uncover antigen degradation, sponsors can select from among “biological” stabilizers (e.g., L-leucine, lactose/trehalose, mannitol/dextran). No hard-and-fast rules apply however, and little may be gleaned from structurally or biologically related systems. The only way to tell is to perform plain-vanilla stability studies using the applicable matrix of stabilizers, excipients, buffers, and drug substance concentrations.

For example, for lipid membrane enveloped viruses, trehalose worked best due to increased hydrogen bonding between stabilizer and virus. For non-enveloped adenovirus, mannitol is the stabilizer of choice due to its smaller size and greater efficiency in replacing water as the predominant hydrogen-bonding species. Testing antigen, adjuvants, stabilizers, and excipients in all reasonable combinations takes time, which is why formulation sometimes stretches beyond phase 1 testing.

Conclusion

Successful vaccine formulation demands that developers manage many factors and requirements simultaneously—a task confounded by the increasing complexity of these products. Novel technologies designed to mitigate specific issues are emerging, but lurking in the background is industry’s fear of being first in line to take on additional regulatory risk.

For example nanotechnology—the lipid nanoparticles used in many drugs—may be applied to stabilize active ingredients through their journey from vial to patient. Adjuvant selection is another strategy for optimizing a vaccine’s effectiveness by way of dose reduction or lowering the cost of goods. Both are good ideas but come at a cost in development time and resources, and with the risk that regulators will shoot them down.

So, for the time being, developers will rely on tools that improve existing approaches to formulation and software products that trim the formulation decision tree somewhat.