Abstract
Introduction
Diabetes mellitus is a complex, chronic and progressive disease [1]. In 2013, 9.3% of the United States (US) population had diabetes [2]. The prevalence of diabetes increased from 5.5 million in 1980 to 22 million in 2014 [3]. It is expected that one in three adults in the US will have diabetes in 2050 [2]. Diabetes is the seventh cause of death in the US [2], a major cause of heart disease and stroke, and the leading cause of macro-vascular complications [4].
The US Food and Drug Administration (FDA) has approved several antidiabetic drugs with different mechanisms of action for marketing in the US. In 2015, antidiabetic products, including insulins, were the first largest class of prescription drug products by sales and the sixth largest class by the number of prescriptions in the US (IMS Institute for Healthcare Informatics, 2016). The diabetes market represented 4.8% of the total number of prescriptions dispensed in the US in that year 2015 [5].
Concerns about the safety and complexity of diabetes treatments have increased overtime particularly for most recently approved antidiabetic drugs [6]. These safety related concerns led to delays in the approval of new antidiabetic drugs and prompted several FDA regulatory actions. The FDA delayed the approval of alogliptin (2013) [7] and dapagliflozin (2014) [8, 9] to gather more safety data. Vildagliptin and insulin degludec were approved in Europe in November 2008 and January 2013, respectively. Vildagliptin has not been approved yet in the US [10, 11]. The FDA approved the insulin degludec in September 2015 after requesting more safety data [12].
Safety concerns also led to FDA regulatory actions. In March 21, 2000, the FDA withdrew Troglitazone from the US market due to its high liver toxicity [13]. In September 2010, the FDA restricted the use of rosiglitazone for two years due to safety concerns [14, 15], and it is evaluating incretin mimetic drugs for potential risk of pancreatitis and pancreatic cancer [16]. In addition, cardiovascular risk associated with new antidiabetic drugs [17, 18] triggered the FDA to implement new guidance requesting sponsor companies to provide evidence that new antidiabetic treatments do not result in an increased cardiovascular risk [19]. In April 2016, the FDA added a new warning about the risk of heart failure to the label of antidiabetic drugs containing saxagliptin and alogliptin active ingredients [20].
Long-term safety and effectiveness are major concerns in the treatment of chronic conditions such as diabetes. The FDA releases safety actions to effectively inform patients and physicians when there is a risk of long-term safety either at the drug approval stage or anytime a new safety concern arises in post-marketing surveillance. The FDA’s safety actions may be a boxed warning, a risk evaluation and mitigation strategy (REMS), a medication guide, and a safety communication. The FDA requires a boxed warning to be added to the drug label to alert prescribers of serious adverse events associated with the drug. A boxed warning contains emerging or known adverse events or clinically significant drug-drug interactions [21]. The Food and Drug Administration Amendment Act of 2007 (FDAAA) enabled the FDA to require a REMS for any pharmaceutical product to ensure that a drug’s benefits outweigh its risks [22]. REMS are strategies to manage a known or potential serious risk associated with drugs and biologics. REMS may consist of a medication guides intended to provide information for the safe and effective use of a drug, a patient package insert, and a FDA safety communication to health care providers (e.g., web-based educational materials and presentations to health care professionals by medical science liaisons) [23, 24].
There is a dearth of literature assessing long-term safety and effectiveness of FDA approved antidiabetic drugs marketed in the US. To our knowledge, all prior work in this area described trends in antidiabetic drug utilization [25–28] and focused on the comparison of the effectiveness and safety of metformin, second-generation sulfonylureas, thiazolidinediones, meglitinides, dipeptidyl peptidase-4 inhibitors, and glucagon-like peptide-1 receptor agonists to treat adults with type 2 diabetes [29–31].
In contrast, we comprehensively examine secular trends in all FDA approved antidiabetic drugs during an extended period of time. In addition, this study goes beyond previous work by examining market discontinuations and FDA safety actions of all antidiabetic drugs approved by the FDA and marketed in the US in the period 1980–2015. Specifically, we examine whether there was a differential increase overtime in the boxed warnings, REMS, medication guides, and safety communications to health care providers in insulins and analogues and non-insulin antidiabetic medications. To the authors’ knowledge, no studies have assessed antidiabetic drugs approved by the FDA for marketing in the US, discontinuations and its safety profile. Results from this analysis have the potential to better inform regulators, providers, and patients on the long-term safety of antidiabetic drugs and to improve our understanding of the FDA regulatory actions and effects in the utilization of antidiabetic drugs. In addition, this work highlights the importance of systematic post-marketing studies assessing the long-term effectiveness and safety of antidiabetic drugs to improve the health outcomes of patients with diabetes.
Methods
Regulatory information about FDA-approved pharmaceutical products for diabetes treatment was collected from the FDA databases, Approved Drug Products with Therapeutic Equivalence Evaluations (Orange Book) [32], Drugs@FDA [33], and other FDA documents related to antidiabetic drug approvals, discontinuations, and safety communications.
The regulatory information collected included the FDA application number, generic and brand name, company name, type of submission (i.e., New Drug Application – NDA- or Biologics License Application – BLA-), approval date, patent and exclusivity expiration date, Abbreviated New Drug Application (ANDA) approval date, route and dosage form, chemical type, FDA approval review process, and marketing status. The FDA approval review process information included standard review and priority review. A priority review drug is defined by the FDA as a product that is a significant improvement compared to already marketed products or provides safe and effective therapy where no satisfactory alternative therapy exists. All non-priority review drugs were considered standard review drugs. An antidiabetic drug was considered to be discontinued from the US market as of December 31, 2015 if it was no longer listed in the Orange Book, it was listed in the discontinued section of the Orange Book, and/or it was listed as discontinued at Drugs@FDA.
The FDA safety related information was extracted from the FDA REMS and Drugs@FDA databases [33–35]. Safety data included boxed warnings, REMS approval, release and modification dates, medication guides, and FDA safety communications.
Antidiabetic drugs were classified using level 3 (pharmacological subgroup) and level 4 (chemical subgroup) of the anatomical therapeutic chemical classification system (ATC class) from the World Health Organization Collaborating Centre for Drug Statistics Methodology [36].
The study included all NDAs and BLAs approved by the FDA for treatment of diabetes and marketed in the US in the period 1980–2015. The units of analysis were the pharmaceutical product (i.e., active ingredient or combination of active ingredients of chemical entities and biologics) and FDA marketing application number (i.e., NDA, ANDA or BLA). The analysis included the first NDA and BLA approved by the FDA for each active ingredient, fixed-dose combination drug (i.e., formulations of two or more active ingredients in a single tablet), and new formulations. All data were updated through December 31, 2015.
Descriptive statistics were performed to describe trends in approvals, discontinuations, and safety actions of antidiabetic drugs by year and ATC class. Regulatory data were assessed by year of FDA first approval and by ATC class to evaluate trends in approvals and changes within therapeutic classes, respectively. Safety actions data were analyzed at the levels of single active ingredient, fixed dose combination, and dosage form.
Results
In the period 1980–2015, there were 92 antidiabetic drugs (including 90 NDAs and 2 BLAs) marketed in the US (Table 1). Tolbutamide (approved in May 1957) was the first FDA-approved antidiabetic drug and it was still in the US market as of December 31, 2015. The latest approval was the fixed-dose combination of empagliflozin and metformin (approved in August 2015). Insulins and analogues, glucagon-like peptide-1 (GLP-1) receptor agonists, and amylin mimetics are biological products but were approved by the FDA using the NDA system with the exception of albiglutide and dulaglutide. Of the 92 antidiabetic drugs marketed, there were 39 insulins and analogues and 53 blood glucose lowering drugs, excluding insulins marketed in the US (Table 2).
FDA approved antidiabetic drugs by therapeutic class, 1980–2015
FDA approved antidiabetic drugs by therapeutic class, 1980–2015
Antidiabetic drug approvals, discontinuations and safety actions, 1980–2015
* % of antidiabetic drugs that have boxed warning, REMS, medication guide, safety communication, or sulfonylureas cardiovascular special warning.
Most insulins and analogues marketed in the US were approved by the FDA in the 1980s. Insulins and analogues can be grouped into categories based on the pharmacodynamics of the products (Table 2). Over 30% (n = 12) of insulins were approved before 1980. They were from animal origin and were discontinued after approval of human insulins. As of December 2015, all insulins and analogues marketed in the US were from human origin. The number of insulin and analogue approvals declined after the 1980s (Fig. 1). The first inhaled insulin was introduced into the US market in 2006. The drug was discontinued due to business reasons in 2007 [37]. The FDA approved a new inhaled insulin in 2014.
The FDA approved sulfonylureas (SU), the first non-insulin antidiabetic drug class marketed in the US, in 1957 (Table 1). The first of the biguanides and alpha-glucosidase inhibitors were approved in 1995, thiazolidinediones (TZD) in 1997, GLP-1 receptor agonists in 2005, and dipeptidyl peptidase-4 (DPP-4) inhibitors in 2006. The latest approved drug class was sodium glucose cotransporter-2 (SGLT-2) inhibitors in 2013. As of December 2015, non-insulin antidiabetic drugs were all oral formulations with the exception of GLP-1 agonists and amylin mimetics. Non-insulin antidiabetic drugs were approved after 1980s with the exception of four products (tolbutamide approved in 1957, chlorpropamide in 1958, acetohexamide in 1964, and tolazamide in 1966). The number of FDA approved non-insulin antidiabetic drugs increased after 1995 (Fig. 1).
Oral fixed-dose antidiabetic drug combinations represented 42.9%, 54.5%, 47.1%, and 100% of the non-insulin antidiabetic drugs approved by the FDA in the periods 2000–2004, 2005–2009, 2010–2014, and 2015 respectively (Table 1). Fourteen out of the nineteen fixed-dose combinations included metformin. The second largest drug class by number of fixed-dose combinations was DPP-4 inhibitors (n = 8), followed by TZDs (n = 6). To date, all FDA approved fixed-dose combination antidiabetic drugs are oral forms.

Antidiabetic drugs marketed in the US in the period 1980–2015 by year of first approval.
The time lag from FDA approval of single active ingredients to the approval of fixed-dose combination drugs containing the same single active ingredient varied over the study period (Table 1). The first fixed-dose combination antidiabetic drug (glyburide/metformin hydrochloride) was approved five years after the first approval of metformin in 1995. Fixed-dose combination drugs including DPP-4 or SGLT-2 inhibitors were approved one year after approval of these active ingredients. Two fixed-dose combinations including alogliptin, alogliptin/pioglitazone and alogliptin/metformin, were approved by the FDA the same day of the first alogliptin approval as a single active ingredient.
The FDA review process used for antidiabetic drugs varied by therapeutic class. During the study period, insulins and analogues were approved by the FDA using the standard review process. Among non-insulin antidiabetic drugs, five products (i.e., metformin, troglitazone, repaglinide, rosiglitazone, and pioglitazone) were approved by the FDA using the expedited priority review process used by the FDA for drugs that are considered an improvement over already marketed alternatives at the moment of approval. Pioglitazone (approved in 1999) was the last FDA-approved antidiabetic drug using the priority review process (Table 1).
Overall, 59.0% of insulins and analogues were discontinued during the study period (Table 2). These discontinuations included 19 animal source insulins and 4 human recombinant insulins (Table 1). The development of human recombinant insulins resulted in the discontinuation of animal source insulins. While there are no reported cases of insulin and analogue withdrawals for safety related reasons [38], higher antibody development associated with use of animal insulins may increase the incidence of adverse drug reactions compared to human recombinant insulins [39]. No animal based insulins have been marketed in the US since 2006 [38].
As of December 31 2015, only three (5.7%) non-insulin antidiabetic drugs (acetohexamide, troglitazone, and simvastatin/sitagliptin) were discontinued from the US market. The FDA determined the withdrawal of troglitazone as a safety withdrawal due to hepatotoxicity [13]. According to the sponsor company, simvastatin/sitagliptin was discontinued from the market for business reasons [40]. No information was available about the reason for discontinuation of acetohexamide.
The FDA released at least one safety action, either boxed warnings, REMS, medication guides, safety communications, or SU cardiovascular special warning for 7.7% of insulins and analogues (3 out of 39) and 88.7% of non-insulin antidiabetic drugs (47 out of 53). Excluding the 6 combined insulins and the 19 non-insulin fixed-dose combination drugs, 9.1% of the insulins and analogues (3 out of 33) and 82.4% of the non-insulin antidiabetic drugs (28 out of 34) had at least one FDA safety regulatory action. Likewise, excluding fixed-dose combination drugs and drugs with the same single active ingredients but with different formulations (none of insulins and analogues and 5 non-insulin antidiabetic drugs; extended release, micronized, or solution), 9.1% of the insulins and analogues (3 out of 33) and 79.3% of the non-insulin antidiabetic drugs (23 out of 29) had at least one FDA safety regulatory action during the study period.
Further, the FDA-approved label of 5.1% of all insulins and analogues (2 out of 39) and 52.8% of non-insulin antidiabetic drugs (28 out of 53) had a boxed warning by December 31, 2015. After excluding fixed-dose combination drugs, 6.1% of insulins and analogues (2 out of 33) and 32.4% of non-insulin antidiabetic drugs (11 out of 34) had a boxed warning on their labels. After excluding combination drugs and drugs with the same single ingredients but with different formulations, the percentage of drugs with a boxed warning on their labels remained unchanged for insulins and analogues and decreased to 31.0% of non-insulin antidiabetic drugs (9 out of 29).
In the study period, there were also several safety concerns related with the boxed warnings. Inhaled insulins had a boxed warning due to risk of acute bronchospasm. Lactic acidosis was the boxed warning reason for biguanides. In the case of TZDs, different safety concerns were described in the boxed warnings even within the same drug class. The safety concern of troglitazone was its hepatotoxicity; whereas, in the cases of rosiglitazone and pioglitazone, it was the risk of congestive heart failure. Only TZDs had boxed warnings in their labels after their approvals. Troglitazone was approved in January 1997 and had a boxed warning in June 1999. Rosiglitazone and pioglitazone were approved in May 1999 and July 1999, respectively and had a boxed warning in August 2007. The twice-daily form exenatide did not have the boxed warning required by the FDA for the extended release form of the drug (risk of thyroid C-cell tumors).
The FDA required the implementation of REMS for 2.6% of all insulins and analogues (1 out of 39) and 28.3% of non-insulin antidiabetic drugs (15 out of 53). When excluding fixed-dose combinations, the FDA required the implementation of REMS for 3.0% of insulins and analogues (1 out of 33) and 26.5% of non-insulin antidiabetic drugs (9 out of 34). After excluding combination drugs and drugs with the same single active ingredients but with different formulations, the percentage of REMS required by the FDA remained the same for insulins and analogues and increased to 27.6% for non-insulin antidiabetic drugs (8 out of 29).
The FDA required medication guides for 5.1% of all insulins and analogues (2 out of 39) and 56.6% of non-insulin antidiabetic drugs (30 out of 53). After excluding fixed-dose combination drugs, the FDA required medication guides for 6.1% of insulins and analogues (2 out of 33) and 41.2 % of the non-insulin antidiabetic drugs (14 out of 34). Likewise, after excluding fixed-dose combination drugs and drugs with the same single active ingredients but with different formulations, the percentage of FDA required medication guides remained the same for insulins and analogues and increased to 44.8% for the non-insulin antidiabetic drugs (13 out of 29).
The FDA released safety communications for 2.6% of all insulins and analogues (1 out of 39) and 52.8% of non-insulin antidiabetic drugs (28 out of 53). After excluding fixed-dose combinations, the FDA released safety communications for 3.0% of insulins and analogues (1 out of 33) and 35.3% of the non-insulin antidiabetic drugs (12 out of 34). After excluding fixed-dose combination drugs and drugs with the same single active ingredients but with different formulations, the percentage of FDA released safety communications remained the same for insulins and analogues and increased to 37.9% for non-insulin antidiabetic drugs (11 out of 29).
The number of antidiabetic drugs with FDA safety actions increased overtime (Fig. 2). Antidiabetic drugs approved before 1995 did not have FDA safety actions. Overall, 40% (4 out of 10) of antidiabetic drugs approved in the period 1995–1999, 45.5% (5 out of 11) approved in 2000–2004, 61.5% (8 out of 13) approved in 2005–2009 and 66.7% (12 out of 18) approved in 2010–2014 had a boxed warning. Likewise, 82.9% of antidiabetic drugs approved after 2005 had medication guides (29 out of 35) and 71.4% had safety communication (25 out of 35).

Antidiabetic drugs marketed in the US in the period 1980–2015 with boxed warnings, REMS, medication guides and FDA safety communications by five-year periods of FDA first approval.
All patents and exclusivities for 66.7% of insulins and analogues and 26.4% of non-insulin antidiabetic drugs expired by December 31, 2015. In December 2015, insulin glargine biosimilar formulation was approved. As of December 2015, 23 (43.4%) FDA-approved non-insulin antidiabetic drugs had generic competitors in the US market and 5 (9.4%) non-insulin antidiabetic drugs had tentatively approved generic oral antidiabetic drugs which are expected to enter the US market once all patents and exclusivities expire. Generic non-insulin antidiabetic drugs were approved on average 10.6±2.3 years after approval of the corresponding NDA.
Insulins have been marketed worldwide since the early 1920s [1]. The first animal insulin marketed in the US was approved by the FDA in 1966 and the first recombinant human insulin was approved in 1982. Several insulins from animal sources were approved before 1980. However, the utilization of animal source insulins declined after the introduction of human recombinant products due to safety concerns, such as prion disease and development of insulin antibodies [39, 41]. Beef insulins for human use were discontinued from the US market in 1998 and pork insulins were discontinued in 2006 [38].
The development of recombinant human insulin in 1982 was a major step toward strengthening the safety of insulin treatment for patients with diabetes. Study results show a reduction in the FDA-approved insulins and analogues since the 1980s. The reasons behind this decrease in approvals may be related with the increase in approvals of non-insulin antidiabetic drugs and with changes in prescribing patterns [28].
Unlike non-insulin antidiabetic drugs that have generic competition, one biosimilar insulin was approved in December 2015 in the US. In Europe, the first biosimilar insulin (i.e., insulin glargine) was approved in September 2014 [42].
Study results show that approval and discontinuation rates of non-insulin antidiabetic drugs are different from insulins and analogues. Prior to 1995, there was only one drug class, sulfonylureas. Since 1995, when metformin was approved, the approvals of non-insulin antidiabetic drugs and therapeutic classes significantly increased. Developing new drugs with various mechanisms of action is beneficial to the management of the disease because patients often require individualized diabetes management depending on co-occurring medical conditions, drug contraindications, and diabetic complications. However, the variety of non-insulin antidiabetic drugs available in the US market also generates safety concerns and increases complexity in the management of diabetes [6]. Currently, there are twelve non-insulin antidiabetic drug subclasses based on their mechanisms of action.
Diverse mechanisms of action have raised safety concerns. In response to these safety concerns, the FDA released several safety actions, including boxed warnings, REMS, medication guides, and safety communications. This study shows that the number of antidiabetic drugs which had FDA safety actions significantly increased after 2005. The number of FDA safety communications in particular, significantly increased in antidiabetic drugs approved after 2005. Overall, safety communications were released based on new information emerged in post-marketing studies.
The FDA releases safety actions for fixed-dose combination drugs if any of the single active ingredients included in the combination has safety actions. Therefore, the percentage of antidiabetic drugs with safety actions varied when analyzing data after excluding fixed-dose combination drugs. For insulins and analogues, the percentage of antidiabetic drugs with safety actions increased because the total number of drugs with safety actions decreased; whereas, for non-insulin antidiabetic drugs, this percentage decreased.
Further, novel study findings show that the percentage of non-insulin antidiabetic drugs that have boxed warnings and REMS is higher than the percentage of these same safety regulatory actions for all FDA approved drugs. A previous study examined boxed warnings of all NMEs approved by the FDA in the period 1975–2009 [43]. Authors assessed the number of all FDA approved NMEs which had boxed warnings as of December 2010 -BLAs were not included in the study. Authors estimated that 15.2% of all NMEs approved in the study period had a boxed warning. This percentage is lower than the 31.0% of non-insulin single active ingredient antidiabetic drugs having a boxed warning found in this study.
Another study assessed secular trends in REMS released by FDA in the period January 2008 through May 2012 for all NMEs approved by the FDA as of May 2012 and found that 9.2% of all active ingredients (NDAs and BLAs) had approved REMS [44]. This is also lower than the 27.6% of single active ingredient non-insulin antidiabetic drugs with REMS found in this study. These studies evidenced the importance of systematic post-marketing studies assessing the long-term effectiveness and safety after drug approvals.
One of the main concerns related to the challenges of the FDA drug approval process is the short term nature of the clinical trials required for drug approval and the lack of sufficient evidence to assess the long-term safety and efficacy of pharmaceuticals [45]. Thus, systematic post-marketing studies remain critically important in order to continuously assess the risks and benefits of drugs after its approval [6]. Newly marketed antidiabetic drugs have being adopted rapidly in the US raising concerns about necessary evidence of its long-term safety [26]. However, the time period needed for assessing the long-term safety profile of antidiabetic drugs is difficult to establish. This study found that in the period 1980–2015, only one antidiabetic drug, troglitazone, was withdrawn from the US market for safety reasons. The withdrawal of troglitazone occurred two years after its FDA approval due to the hepatotoxicity [46]. This study also found that half of the antidiabetic fixed-dose combination drugs approved during the study period were approved immediately after the FDA approval of the single active ingredients included in the combination. Hence, fixed-dose combination drugs were approved before evidence of the long-term post-marketing safety of their individual active ingredients was confirmed.
This study has some limitations. The analysis of the FDA safety actions of antidiabetic drugs focused on five specific FDA regulatory actions; market discontinuations, boxed warnings, REMS, medication guides, and safety communications. While these safety actions comprise major safety issues, other important safety problems contained in FDA warnings, contraindications, and other FDA-approved label sections were not assessed in this study.
Also, this study evaluated FDA approvals, discontinuations, and safety actions related to antidiabetic drugs but did not assess trends in the utilization of those drugs in clinical practice, specifically how the FDA safety regulatory actions may affect drug selection and utilization. Future research should address these limitations.
Conclusions
There has been a significant increase in the number of FDA approved antidiabetic drugs marketed in the US in the period 1980–2015. Insulins and analogues were the therapeutic classes with the highest number of approvals during the 1980s. Six in ten insulins and analogues approved in the study period were discontinued.
The approval of non-insulin antidiabetic medications significantly increased over the study period. Non-insulin antidiabetic drugs represent most of the antidiabetic drugs currently available in the US market. The increase in the number of non-insulin antidiabetic drug classes with different mechanisms of action has expanded the therapeutic alternatives for patients with type 2 diabetes. However, newly approved drugs have also raised safety concerns and led to FDA safety regulatory actions including boxed warnings, REMS, medication guides, and safety communications to health care providers. Non-insulin antidiabetic drugs have more boxed warnings and REMS than any other therapeutic class. There is a need for systematic post-marketing studies assessing the long-term effectiveness and safety of antidiabetic drugs to improve the health outcomes of patients with diabetes.
Conflict of interest
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
