Quantifying the value of orally delivered biologic therapies: A cost-effectiveness analysis of oral semaglutide

Abstract

Oral semaglutide, currently in phase three clinical trials, represents the first case of an oral biologic medication for type 2 diabetes in the form of a daily capsule. It provides similar efficacy compared to its weekly injection counterpart, but it demands a dose 280 times as high and requires more frequent administration. We perform a cost effectiveness analysis using a first and second order Monte Carlo simulation to estimate quality-adjusted life expectancies associated with an oral daily capsule, oral weekly capsule, daily injection, and weekly injection of semaglutide. We conclude that the additional costs incurred to produce extra semaglutide for the oral formulation are cost effective, given the greater quality of life experienced when taking a capsule over a weekly injection. We also demonstrate that the potency of semaglutide allows the formulation to be cost effective, and less potent drugs will require increased oral bioavailability to make a cost effective oral formulation.

Introduction

Glucagon-like peptide-1 (GLP-1) receptor agonists represent a major therapeutic advance in the treatment of type 2 diabetes (T2D). These agents are among the most effective options available for glycemic improvement ^1–3. They also offer the unique advantage of significant weight reduction, and they do not cause hypoglycemia. Moreover, the major goal of T2D treatment is reduction of cardiovascular disease, and two GLP-1 receptor agonists have been shown to lower major adverse cardiovascular events in large clinical trials ^4,5.

Proteolytic degradation and poor absorption in the gastrointestinal tract presently necessitate delivery of GLP-1 receptor agonists via subcutaneous injections ^6,7, yet patients prefer oral medications over injectable ones ⁸. Physicians prescribe GLP-1 receptor agonists substantially less often than oral hypoglycemic alternatives, even if these are less effective for glycemic improvement ⁹. Sitagliptin, the oral dipeptidyl peptidase-4 (DPP-4) inhibitor, outsold both exenatide and liraglutide, injectable GLP-1 receptor agonists, by billions of dollars in 2017 according to public sales data; however, these drugs act on the same biologic pathway, and GLP1 receptor agonists result in superior glycemic control and weight loss benefits ^10,11. Dosing frequency is another key factor in patient preference and adherence. Patients demonstrate greater adherence to weekly dosed medications compared to daily dosed ones ^12,13. Therapeutic benefits only manifest in full when drugs are used regularly.

Recent scientific advancements have made possible orally bioavailable biologic drugs, and some may even be close to receiving FDA approval. One example is an oral once daily version of the GLP-1 receptor agonist semaglutide, which has been studied in the phase 3 PIONEER clinical trials ¹⁴. Published clinical trials comparing the oral version of semaglutide to the subcutaneously dosed version demonstrate that a 1 mg weekly subcutaneous injection of semaglutide provides the same health outcomes as a 40 mg daily capsule ¹⁵. Lower doses of oral semaglutide such as 14 mg daily oral capsules also provided positive health outcomes, but these health outcomes did not equal ones experienced from the subcutaneous injection. The oral formulation of semaglutide utilizes the absorption enhancer SNAC (Sodium N-[8-(2-hydroxybenzoyl)amino] caprylate) to achieve gastrointestinal uptake ¹⁵. Orally dosed pills containing a ratio of 10 mg of semaglutide to 300 mg of SNAC have been shown to locally increase the absorption of semaglutide in the stomach within the area close to the tablet surface and have been shown to be safe in human and multiple animal models^15–17. SNAC protects the semaglutide from enzymatic degradation by buffering the area close to the tablet and momentarily increases absorption. Oral semaglutide has been shown to achieve a bioavailability of 1.22 ± 0.25% in fasted dogs ¹⁷. This compares to a bioavailability of 89% for subcutaneous injections in humans ¹⁸. In the human clinical trials for oral semaglutide, although a bioavailability was not explicitly calculated, the subcutaneous dose required 0.36% as much drug as the orally dosed formulation to achieve an equivalent health outcome. In addition to capsules utilizing absorption enhancers, preclinical studies utilizing microneedles, ultrasound, nanotechnology, nanoparticles and mucoadhesive patches show promise for further increasing oral bioavailability of therapeutic agents ^{15,19,28,29,20–27}.

Developing an oral formulation of a biologic drug currently requires an increased amount of active pharmaceutical ingredient (API) per dose, due to lower bioavailability. Because biologic APIs have high raw material costs (on the order of $50-$100/g), increasing the drug load significantly increases the cost of the final product ^30,31. An oral formulation or device must provide enough systemic uptake from the gastrointestinal tract to demonstrate acceptable value over competing injectable products. For example, an oral insulin product demonstrated efficacy in phase two clinical trials, but the project was cancelled because the required quantities of the active pharmaceutical ingredient were too high and use of the product was deemed not commercially viable ³².

This study seeks to understand how orally bioavailable GLP-1 receptor agonists and dosing intervals will change quality adjusted life years (QALYs) gained in patients with T2D. We present a cost effectiveness analysis using a mathematical simulation model to compare four possible delivery methods for GLP-1 receptor agonists. The simulation projects and compares the quality-adjusted life expectancies of patients prescribed once daily capsules ^15,19,33, once daily injections ⁴, or once weekly injections ¹³. It also predicts the benefits of once weekly capsules if this type of dosage form for biologics were to come to fruition ^34–37.

Methods

We developed a first-order Monte Carlo microsimulation model to examine GLP-1 receptor agonist therapy delivery options in adults with T2D. The model compared the lifetime incremental cost-effectiveness of oral and injectable formulations prescribed with once daily or once weekly dosing. In the model, both injectable and oral doses provided comparable HbA1C effects, as the patients were dosed with 1 mg/week of drug in the injectable group and 280 mg/week of drug in the pill group. The oral dose used in this model also possessed 2100 mg/week of SNAC. These oral doses have been shown to be safe in phase 2 clinical trials¹⁵. We compared strategies using hypothetical cohorts of 50 year-old US males and females with Hemoglobin A1c (HbA1c) values of 8.0%. Other age groups were also analyzed, and results from these other studies are presented in the supplementary text and Figure S1. The model covered a lifetime horizon, utilized a payer perspective, and applied an annual 3% discount rate for costs and benefits. The discount rate weights each subsequent year with a 3% lower value compared to the previous year. This value is the standard discount rate used in cost-effectiveness models for health and medicine ³⁸. We modeled the patient’s quality of life using health state utility weights, where 0 signified death and 1 represented perfect health. This is the standard method of determining a patient’s quality of life in cost-effectiveness models for health and medicine ³⁸. An otherwise healthy individual with T2D with an HbA1c level of 6.0% had a health state utility of 0.85 in the model ³⁹. A patient’s total quality adjusted life years equaled the sum of their yearly health state utilities over their lifetime. We utilized several variables which tracked the individual-level heterogeneity in the model. Each subject possessed changing HbA1c levels, side effects, and adherence to the medication. These factors impacted the health state utility of the induvial over time. We provide the values and probabilities utilized in the model in Table 1.

Table 1:

Model variables with Base-Case Values and Ranges used in a Probabilistic Sensitivity analysis. Numbers marked with a (*) are estimates.

Parameter	Base case (SD)	Distribution	Source
Disutility from Administration
Daily Pill	−0.007 (0.001)	Normal	56
Daily Injection	−0.061 (0.013)	Normal	8
Weekly Pill	−0.002* (0.001)	Normal	56
Weekly Injection	−0.037 (0.013)	Normal	8
Proportion of Days Covered Taking Medication
Daily Pill	0.721 (0.277)	Beta	44
Daily Injection	Table	Table	12
Weekly Pill	Table	Table	36
Weekly Injection	0.71 (0.28)	Beta	13
Discontinuation of medication
Daily Pill	0.42 (0.03*)	Normal	49
Daily Injection	0.36 (0.03*)	Normal	13
Weekly Pill	0.15 (0.03*)	Normal	37,57,58
Weekly Injection	0.28 (0.03*)	Normal	13
Discontinuation rate for year 2 as a ratio to year 1	0.8* (0.03*)	Normal	49
Side Effect Disutility
5% lower weight and nausea	−0.01 (0.04)	Normal	43
5% lower weight and no nausea	0.03 (0.07)	Normal	43
Side Effect Rate
Oral Administration	0.61 (0.05*)	Normal	15
Injectable Administration	0.54 (0.05*)	Normal	15
HbA1c Effects
HbA1c Initial drop from medication	−1.9 (0.1)	Normal	15
HbA1c Annual Change (Fully Adherent)	0.15 (0.15)	Normal	4
HbA1c Annual Change (No Medication)	0.47 (0.47)	Normal	42
HbA1c All Cause Mortality Hazard Ratio		Table	40
<6.0%	1.1
6.0%−6.5%	1.3
6.5%−7.0%	1.6
7.0%−.9%	1.6
8.0%−8.9%	2.2
>9.0%	2.6
Quality adjusted life year at a given HbA1c		Table	39
<6.0%	0.85
7.0%	0.817
8.0%	0.784
9.0%	0.751
>10.0%	0.18
Costs
Semaglutide base Cost	$8818	Known Value	45
Additional drug cost for oral formulation	$100/g	$50-$100	30,31

During each stage, the patient transitioned through the model as described in Figure 1. Every year patients either remained in the alive state, or they died from background mortality or diabetes-related complications. We calculated the patient’s mortality rate via a published algorithm that defined a hazard ratio from their HbA1c levels ⁴⁰. We used U.S. 2014 life tables to estimate background mortality ⁴¹. Throughout the model, we tracked HbA1c variation in the simulated patients and changed the value based on medication compliance. We examined the phase 2 clinical trial performed with oral semaglutide to estimate the initial HbA1c drop after using the medication for the first year ¹⁵. We drew data for long term GLP1 receptor agonist effects on HbA1c from a multi-year follow-up study on patients taking the GLP-1 receptor agonist liraglutide ⁴. The United Kingdom Prospective Diabetes Study (UKPDS) analysis on the rise in HbA1c levels over time in patients with T2D provided HbA1c trends for non-adhering patients ⁴². The subcutaneously administered semaglutide clinical trial data also provided information on the prevalence of side effects. The disutility of these side effects had been previous studied in the diabetes literature ⁴³.

Fig 1:

Structure of the dosing simulation model. (a) The microsimulation model used to estimate the life expectancy and quality-adjusted life years of patients prescribed semaglutide analyzes four possible administration methods differing in dosing system and frequency. (b) Patients prescribed the medication can die during each yearly cycle due to background mortality or diabetes related complications. Their hazard ratio is tied to their HbA1c level, which is tracked throughout the simulation. Patients will discontinue the drug and experience side effects at a rate consistent with currently marketed GLP-1 receptor agonists. Their quality of life during each cycle is determined based on their HbA1c level, side effects, and choice to continue the drug. (c) Patients can move from the drug administration state to the discontinued state, or they can exit the model due to death.

Medication adherence rates in the model mimicked those of medications currently on the market that act on the same biologic pathway as semaglutide. We used the compliance distribution for sitagliptin to estimate the percent of medication taken for a patient prescribed a daily GLP-1 receptor agonist pill ⁴⁴. We used adherence rates for daily liraglutide and weekly dulaglutide to evaluate adherence to daily and weekly injections respectively ^12,13. We assigned individual patients a random adherence based on these distributions. Although recent developments have reported capsules with the potential to delivery drugs for one week ^34,35, no weekly diabetes capsule exists. We utilized meta-analyses as well as information on weekly bisphosphonate adherence to estimate the adherence to a weekly pill ^36,37. The compliance rate for weekly oral bisphosphonates may not correlate perfectly with oral semaglutide because their administration method is not exactly the same; for example, oral bisphosphonates require that a patient stand for 30 minutes after taking the medication, and oral semaglutide must be taken in a fasted state. However, we perform sensitivity analyses (Figure 2) that demonstrate how the base case is affected by this and several other important assumptions. In this model, a patient was considered adherent to the medication, and received the full effect of the medication, if they took more than 80% of their dose. Patients who took fewer than 80% of their medication received both proportional costs and drug effects depending on the amount of medication they consumed. Their compliance also dictated their disutility from administration. The model varied this disutility value by up to 40% of the base case depending on the adherence of the individual. After the first and second year, a percentage of the patients in the model discontinued their medications. This was defined by follow-up studies of GLP1 receptor agonists and DPP4 inhibitors. Once the patient discontinued the medication, they could not be prescribed the medication again, and all patients remained in their assigned group after the third year.

Fig 2:

Sensitivity analysis around administration disutility and rates of discontinuation for oral and injectable formulations. Preferred administration is determined by QALYs gained. Using the base case for the model with values derived from literature, both the (a) daily and (b) weekly pills are preferred over any type of injection. At a discontinuation rate of around 80% for pills, the weekly injection becomes more favorable. The high injection disutility and low oral capsule disutility defined in the base case demonstrate a preference for the (c) daily and (d) weekly oral formulation over injection. At high pill administration disutility and low injection administration disutility, the injection formulation is preferred.

Each of the parameters in the model was subjected to variation using a probabilistic sensitivity analysis. If the standard deviation was not provided in the acquired data, then additional sensitivity analyses were performed around these values. We used probabilistic sampling on the individual patient level to model heterogeneity in side effects disutility, administration disutility, as well as HbA1c effects. Sampling occurred for side effect rate and medication discontinuation rate on the cohort level. In addition, we performed two-way sensitivity analyses on discontinuation rates and administration related disutility. We ran the model through 1000 unique samples for each method of delivery, and each sample contained a cohort of 10000 patients.

Because the daily oral formulation of semaglutide has not reached the market, and a weekly oral formulation of semaglutide does not exist, we made an assumption about their costs. All drug formulations received a base cost of $8,818 per year, the price for the injectable semaglutide drug in the United States ⁴⁵. We assumed that the lowest possible cost increase for an oral dosage form over an injectable form would directly correlate to the additional costs of goods from increased active pharmaceutical ingredient. We added the additional cost of the raw material to the total cost of the pharmaceutical. Many biologic drugs cost on the order of $50-$100/gram to produce ^30,31. We performed a sensitivity analysis around drug costs within this range.

Additionally, we considered that as the manufacturing process is scaled up, other formulation or manufacturing changes are made, and new breakthroughs are made in biologic drug production the manufacturing cost might change significantly. In supplementary figure S2, we analyze how drug manufacturing costs on the order of $1, $10, and $100/g change the cost of producing the active pharmaceutical ingredient for an oral biologic drug relative to its reimbursement rate. While parenteral drugs have monetary costs associated with them that oral tablets do not, such as producing an aseptic subcutaneous formulation and an injection pen as well as the cost of training a patient to perform injections, these were not considered in the model. Additionally, the cost of potential excipients added to the oral formulation are not incorporated in the model costs. Instead, they are factored into the sensitivity analysis.

We validated the model internally and externally quantitatively by verifying that the adherence, HbA1c levels, and disutility of patient cohorts passing through the model matched the data used to estimate the model parameters. We cross-validated the model by comparing the life expectancy of a 55 year-old male patient in the model to that of 55 year-old male patients in the UKPDS model. Our model predicted a life expectancy of 18.9 years compared to 19.1 years in the UKPDS model ⁴⁶. In addition, we cross-validated by comparing the differences in discounted quality-adjusted life years of a 55 year-old patient prescribed sitagliptin or liraglutide in our model, without accounting for the disutility from the treatment method. Both models predicted a gain of 0.4 quality adjusted life years ⁴⁷.

Results

Overall, a male patient gained an average of 0.17 and 0.38 discounted quality-adjusted life years over their lifetime when prescribed a daily pill compared to a weekly injection or a daily injection respectively (Figure 3a). A female patient gained 0.05 fewer quality adjusted life years than a male patient for either case which can be attributed to the different life expectancy tables for males and females. A male or female patient gained on average an additional 0.2 discounted QALYs from increased compliance when prescribed a weekly compared to a daily pill (Figure 3a). The calculations below were performed for male patients but the conclusions were validated for female patients as well.

Fig 3:

Cost effectiveness analysis of drug administration methods. (a) The average discounted quality adjusted life years and (b) life expectancy for a 50 year-old man with an HbA1c level of 8% prescribed the GLP-1 receptor agonist semaglutide. The weekly formulations show higher life expectancies due to increased compliance, and the capsule formulation shows higher quality adjusted life years experienced due to the disutility associated with injection. (c) Discounted lifetime costs associated with each formulation prescription. (d) The probability of cost-effectiveness for semaglutide formulations. In this model, an injection is dosed at 1 mg/week, and a pill is dosed at 280 mg/week.

Due to the fact that patients demonstrate greater compliance to weekly dosed medications over daily dosed ones, the life expectancy of a patient went up with weekly compared to daily formulations (Figure 3b). For example, using a weekly injection compared to a daily pill accounted for an average gain of 0.06 years of life expectancy. Daily oral capsules and injections showed somewhat different compliance curves and discontinuation rates; however, the life expectancy of a patient on either formulation was equal. A weekly pill provided a higher life expectancy than all of the other formulations because of its high compliance and low discontinuation rate.

It was possible to estimate the potential costs of the active pharmaceutical ingredient required for the oral capsules and subcutaneous injections based the doses required for each method to achieve comparable health outcomes. The subcutaneous injection of semaglutide is dosed at 0.51 mg per week. At a cost of $100/gram, this amounted to a cost of $2.60-$5.20 per year. The oral version of semaglutide required a dose of 21–40 mg per day to replicate the HbA1c effects of the respective injected doses ¹⁵. Therefore, the oral formulation used approximately 280 times the amount of active pharmaceutical ingredient (API). It costs $770-$1460 per year to produce the necessary API for the oral dosage form. Phase three clinical trials for oral semaglutide only tested 14 mg doses of oral semaglutide. Although they did not show comparable effects to the subcutaneous injections, they still demonstrated positive health outcomes. It costs $510 per year to produce the necessary API for this dose size. Figure 3c shows the lifetime costs associated with the prescription of each formulation and Figure 3d details the probability of cost effectiveness for each formulation strategy at varying willingness-to-pay thresholds. Because the cost to produce semaglutide might change as the process is scaled up to produce additional drug, or our estimation of the cost may be incorrect, we performed a sensitivity analysis around this value. Supplementary figure S2 shows the relative cost of producing the active pharmaceutical ingredient semaglutide compared to the annual drug cost for patients and insurers for drug production costs ranging from $1-$100/g.

With the additional expense, the oral semaglutide costs $10,000 more in raw materials over the course of a lifetime when comparing daily dosed pills to weekly injections (taking into account adherence, drop-outs, and discount rates). Still, the QALYs gained from a pill allowed the oral formulation to provide greater QALYs at a lower cost-effectiveness ratio. Both the daily and weekly oral dosage forms were cost-effective in 50% of cases at a cost-effectiveness ratio of $85,000 per QALY. A cost-effectiveness table provides the incremental cost-effectiveness ratios (ICER) (Table 2). This compares to a cost effectiveness ratio of $110,000 per QALY gained for a weekly semaglutide injection. Changing the raw material drug cost to $50/g lowered the cost-effectiveness ratio of an oral pill by $5,000 per QALY.

Table 2:

Lifetime Direct Medical Costs, Effectiveness, and Incremental Cost-Effectiveness for different drug delivery strategies for the GLP-1 Receptor Agonist Semaglutide based on Costs of Goods. Does not include dominated strategies. This table uses 1 mg and 40 mg doses for injected and oral formulations. It includes the retail cost of semaglutide as a baseline cost. Oral formulations receive an increased cost based on the added pharmaceutical ingredient required. An analysis of all possible drug strategies including the less cost effective daily and weekly injections is found in Table S1.

Strategy	Costs (Discounted)	Effects (Discounted Quality Adjusted Life Years)	Incremental Costs	Incremental Effects	Incremental Cost Effectiveness Ratio
Male
No Prescription	$0	10.76	N/A	N/A	N/A
Daily Oral Capsule	$70,000	11.58	$70,000	0.82	$85,000
Weekly Oral Capsule	$87,000	11.78	$17,000	0.2	$85,000
Female
No Prescription	$0	12.17	N/A	N/A	N/A
Daily Oral Capsule	$69,000	12.92	$69,000	0.75	$92,000
Weekly Oral Capsule	$103,000	13.13	$33,000	0.21	$157,000

A sensitivity analysis around disutility from administration and medication discontinuation rate demonstrated that oral delivery provided greater QALYs over injectable delivery for a broad set of input values (Figure 2). The base-case for the model favored daily capsules over both daily and weekly injections. This occurred regardless of the greater adherence to weekly formulations. A weekly injection would require a disutility from administration of less than −0.02 QALYs per year to become preferred over a daily pill. Additionally, patients would require a dropout rate 2–3 times higher for a pill compared to an injection to favor injections over daily capsules. These values fall outside of the variability ascribed in literature (Table 1).

Figure 4 details the effects of dose size on cost effectiveness. Semaglutide requires a dose of only 1 mg per week at maximum dose. At this dose level, daily oral capsules remained more cost-effective than weekly injections. The cost increase from adding extra drug to the formulation did not affect the total cost significantly due to the nature of the drug’s potency. This would change if semaglutide required a larger dose such as 5 mg/week. At this point, the ICER for switching from a weekly injection to a daily pill would exceed $100,000 per QALY gained. A formulation with an increased bioavailability would be required to keep costs down when reformulating a less potent drug.

Fig 4:

The incremental cost-effectiveness ratio in dollars per quality adjusted life year for switching from a weekly semaglutide injection to a daily pill. If the required dose increased, the cost of using an injection versus an oral formulation would also increase. The relative oral dose needed for an equivalent health outcome, which directly relates to the oral bioavailability of the formulation, affected the results more significantly at a higher dose.

Discussion

GLP-1 receptor agonists offer unique benefits, including reductions in adverse cardiovascular outcomes, which make these agents highly desirable choices for the treatment of T2D. To date, only injectable formulations have been approved by regulatory agencies. Currently, an oral once daily version of semaglutide is in phase 3 clinical trials, so oral GLP-1 receptor agonists may soon be a clinical reality. We found a daily oral formulation of semaglutide to be cost-effective over a weekly injection. A lower medication compliance associated with a daily oral formulation was offset by a higher disutility from injection based administration. Semaglutide, due to its potency, provides a path towards oral formulations for biologic diabetes medications that can be cost-effective.

While semaglutide is highly potent, other GLP-1 receptor agonists require more API to achieve similar drug effects. For example, parenterally dosed liraglutide requires a dose of 1.8 mg/day and has an absolute bioavailability of 55% ^4,48. If an oral version of liraglutide utilized SNAC and required the same relative increased oral dose as semaglutide to achieve comparable effects to its injection based formulation, then a patient would require a pill which dosed the drug at 260 mg/day. With SNAC, this would amount to a dose of over 8,000 mg of material. This dose would not be practical for a patient to take every day, and it could result in decreased compliance. Additionally, at a manufacturing cost of $100/g, oral liragutide would not be cost effective according to our model.

In this model, we made an assumption on the drug continuation rate in the case of the pill form of semaglutide since no oral GLP-1 receptor agonist is currently on the market. The base case that was assumed for the semaglutide pill continuation rate was the continuation rate for sitagliptin. However, it is documented in the prescribing information that patients treated with GLP-1 receptor agonists report gastrointestinal adverse events at a significantly higher rate than patients treated with a placebo. Additionally, GLP-1 receptor agonists have exhibited a different adverse event profile than sitagliptin as shown in the prescribing information. The frequency of gastrointestinal side effects may affect the drug continuation rate. To ensure that our assumption did not affect the conclusions generated by this model, we performed a sensitivity analysis on drug continuation rate and compared it to the assumed base case (Figure 2 a,b). The sensitivity analysis showed us that even if the continuation rate was almost ½ of the assumed continuation rate, then a daily pill would still be preferred over a weekly injection. Furthermore, the discontinuation rate of subcutaneously injected GLP-1 receptor agonists is lower than sitagliptin^13,49. Because no oral GLP-1 receptor agonist currently exists, the assumption was made to allow for the production of this model, and the sensitivity analysis enables testing of the assumption on the model outcomes.

Importantly, this cost-effectiveness analysis only included a price increase based on the cost of goods for the API. No additional profit was included in the overall cost. Future studies will need to assess cost-effectiveness after the market price of oral semaglutide is released. Moreover, more efficient and inexpensive manufacturing of biologic APIs could also expand their availability as cost effective orally delivered interventions ^50,51.

Another factor that plays into the GLP-1 receptor agonist market is the presence of a competing oral drug, sitagliptin. Because semaglutide provides more significant health benefits compared to sitagliptin, the new oral formulation provides the potential to overtake the current market for sitagliptin. Several cost-effectiveness analyses examine the QALYs gained by switching from sitagliptin to a GLP-1 receptor agonist. All of the studies demonstrate that prescribing a GLP-1 receptor agonist is cost-effective ^47,52–54. Despite this, sitagliptin currently outsells individual GLP-1 receptor agonists, which supports the possibility that an oral semaglutide formulation would reach a wider population regardless of its cost-effectiveness ⁵⁵. Increased sales numbers from a new oral formulation of semaglutide may offset lower profit margins due to increased production costs.

An oral formulation of a GLP-1 receptor agonist will be an exciting first step towards providing injectable medications in oral forms. Our analysis demonstrates that the increased costs required to switch semaglutide to an oral formulation can be offset by the utility gained. However, our analysis suggests this would not be true for drugs which require a larger weekly dose size. At a manufacturing cost of $100/g, a drug would need to require a weekly dose 1 mg or less to be cost effective at under $100,000/QALY gained. According to our sensitivity analysis, the maximum weekly drug dose would expand to 10 mg/week if the oral bioavailability was increased by 10 fold or the manufacturing cost was reduced by 10 fold. Medications that require these higher doses cannot currently be switched to an oral formulation using the same absorption enhancer strategy in a cost effective manner. New methods for drug delivery utilizing devices ^19,33 or alternative formulations will be required to create oral formulations of other medications. Examples of drugs that might have a great clinical impact as oral formulations but would require doses higher than semaglutide include: insulin, filgastrim, somatropin, adalimumab, infliximab, and etanercept. More work is required to understand the drug potency, variability, toxicity, and long-term side effects of oral formulations of other currently injectable drugs. Additionally, while a daily oral formulation of semaglutide is now in clinical trials, no scientific articles have demonstrated efficacy with a weekly oral formulation for biologic medications.

With more than 30 million patients affected by T2D in the United States alone and more than 400 million patients with T2D worldwide, oral formulations for GLP1 receptor agonists would generate immense value in terms of QALYs gained compared to injectable formulations. Up until now the cost associated with oral formulations remained too high due to the increased costs for producing the additional required drug. Semaglutide’s potency, however, has allowed for the creation of a cost-effective oral biologic medication. Future work surrounding oral biologic drug delivery is still necessary to increase oral bioavailability and allow other less potent drugs to be administered orally as well. Multiple preclinical studies with promising results demonstrate the possibility of a future where many more medications will receive an oral formulation and patients will no longer have to experience the disutility associated with subcutaneous injections.