Antibiotics Approved for Marketing in Populations Specifically Excluded From Premarketing Trials, 1999–2018

Abstract

Approval by the US Food and Drug Administration (FDA) of a drug for a given indication is thought to reassure clinicians, other health care providers, and patients that substantial evidence of effectiveness exists for specific indicated populations (patients and diseases). This study examines whether FDA approval of certain antibiotics should be so reassuring for all patient populations identified in the FDA-approved labels. Specifically, this study compared patient populations covered by FDA-approved labels for 21 novel antibiotics approved between 1999 and 2018 to the patient exclusion and inclusion criteria of pivotal trials that supported those approvals. We found that every FDA-approved label for these antibiotics included at least one identifiable patient population that was explicitly excluded from enrolling in the supporting pivotal trials. Two antibiotics, bedaquiline and ceftazidime-avibactam, were approved for use in populations that were fully excluded from enrolling in registration trials.

US law requires substantial evidence of population-specific effectiveness from adequate and well-controlled studies be reflected in drug labels. Evidence in specific patient populations is especially important for antibiotics, where injudicious prescribing can lead to treatment failure, antibiotic resistance, and increased adverse events.^1,2 Although clinical trials are criticized for not including a range of patient populations due to lack of enrollment (but not exclusion), in some instances drugs have been approved for populations specifically excluded from registration trials.³ Such actions allow companies to legally market their products for use in populations that were specifically excluded from clinical trials.

For all antibiotics approved since 1999, we systematically assessed the types of populations in which the drugs were approved for use despite being explicitly excluded from registration trials. We also determined whether these exclusions were noted in product labeling.

METHODS

All new molecular entity antibiotics initially approved by the US Food and Drug Administration (FDA) between October 1999 and December 2018 were screened for inclusion. For each indication, FDA-approved populations were characterized based on the “indications and usage,” the “contraindications,” and the “use in specific populations” sections of original approval labels.

To determine populations studied, we examined the inclusion/exclusion criteria of all pivotal pre-marketing trials (generally phase 3 trials intended to establish drug efficacy), as presented in FDA medical officer reports on the Drugs@FDA database upon initial approval.⁴ Indications without at least one sentence or phrase regarding trial inclusion or exclusion criteria in FDA documentation were excluded from the analysis.

Cross-referencing FDA-approved populations listed in the initial label with populations studied, we catalogued and categorized the types of patients in whom drugs could be marketed legally even though those types of patients had been excluded from participating in pivotal trials. We also recorded whether the lack of study for specified excluded populations was mentioned in the label. Exclusion criteria such as that which served to help identify individuals with the target infection or previous recent effective antibiotic therapy in the setting of noninferiority trials that were considered reasonable in assessing the efficacy of treatment were not counted as unstudied but indicated populations.

All data were extracted by one author and checked by a second, with disagreements resolved through discussion, and are publicly available.⁵

RESULTS

Between 1999 and 2018, the FDA approved 26 novel antibiotics for a total of 47 indications: bedaquiline (Sirturo), besifloxacffi (Besivance), ceftaroline (Teflaro), ceftazidime-avibactam (Avycaz), dalbavancffi (Dalvance), daptomycin (Cubicin), delafloxacin (Baxdela), eravacycline (Xerava), ertapenem (Invanz), fidaxomicin (Dificid), finafloxacin (Xtoro), linezolid (Zyvox), meropenem-vaborbactam (Vabomere), miltefosine (Impavido), moxifloxacin (Avelox), omadacycline (Nuzyra), oritavancin (Orbactiv), secnidazole (Solosec), sarecycline (Seysara), tedizolid (Sivextro), telavancin (Vibativ), telithromycin (Ketek), tigecycline (Tygacil). Five antibiotics (telithromycin, gemifloxacin, cefditoren, omadacycline and plazomicin, representing eight indications) were excluded from further analysis because trial inclusion and exclusion criteria for these indications were not listed in the publicly available FDA medical officer’s review. Approximately half of the antibiotics were approved before 2009 and half since, with the majority indicated for respiratory (12 of 47; 25.5%) and skin (14 of 47; 29.8%) infections (Table 1). Whereas a majority of labels either listed age groups or had age-specific sections in the label (36 of 47; 76.6%), less than half discussed the indicated severity of disease (20 of 47; 42.6%).

TABLE 1.

Characteristics of Novel Antibiotics and Associated Indications Approved by the FDA Between December 1999 and October 2018^a,b

	All indications approved within study period	Indications included in analysis
Novel indications	47	39
Approval year range
1999–2003	20 (42.6)	14 (35.9)
2004–2008	4 (8.5)	4 (10.3)
2009–2013	6 (12.8)	6 (15.4)
2014–2018	17 (36.2)	15 (38.5)
Infection type
Target a single-pathogen	6 (12.8)	6 (15.4)
Skinc	14 (29.8)	13 (33.3)
Respiratoryd	12 (55.3)	7 (17.9)
Intra-abdominal	4 (8.5)	4 (10.3)
Genitourinarye	6 (12.8)	5 (12.8)
Ear, eye, nose, throat and sinusesf	5 (10.6)	4 (10.3)
Indication characteristics
Age-group	36 (76.6)	34 (87.2)
Body site or name of infection	42 (89.4)	39 (100.0)
Infecting organisms	40 (85.1)	37 (94.9)
Disease severity	20 (42.6)	18 (46.2)
Use when limited or no treatment options available	2 (4.3)	2 (5.1)

^aFDA = US Food and Drug Administration.

^bValues shown are n (%) unless otherwise stated.

^cSkin infections include both acute bacterial skin and soft tissue infections (n=13) and acne vulgaris (n=1).

^dRespiratory infections include community acquired pneumonia (n=7), nosocomial pneumonia (n=1), and acute bacterial exacerbation of bronchitis (n=4).

^eGenitourinary infections include bacterial vaginosis (n=1), urinary tract infections (n=4), and acute pelvic infections (n=1).

^fEar, eye, nose, throat, and sinus infections included acute bacterial sinusitis (n=2), bacterial conjunctivitis (n=1), and acute otitis edema (n=1) and pharyngitis/tonsillitis (n=1).

All antibiotics were approved for use in at least one identifiable population that was excluded from pivotal trials. The most common population (35 of 39; 89.7%) excluded from pivotal trials consisted of individuals whose infection had specific characteristics that may have required another treatment strategy or was more serious (Table 2). Individuals with immunosuppressive conditions or immunosuppressive treatments were often (25 of 39; 64.1%) excluded without mention in the label. Although only 5 of 39 antibiotic labels recommended against use in individuals with renal impairment, a majority (23 of 39; 59.0%) excluded individuals with some form of renal disorder or impairment from participating in their study (Table 2). Hepatic impairment was also a common exclusion criteria in trials, but when this occurred, the lack of study was mentioned in labels approximately two-thirds of the time (13 of 20; 65.0%). Although lack of study in pregnant and nursing mothers was always mentioned in the label (20 of 20; 100.0%), excluded patients who failed previous antibiotic therapy were rarely mentioned (1 of 13; 7.7%).

TABLE 2.

Types of Individuals Included in Labeled Indication but Excluded From Pivotal Trials Supporting FDA Approval Among 39 Analyzed Indications^a

Trial exclusion criteria	Total indications	Label mentions absence of study in given populationb
Infection with specific characteristics	35 (89.7)	2 (5.7)
Immunosuppressive condition or immunosuppressive treatment	27 (69.2)	2 (7.4)
Renal impairment or disorder	23 (58.9)	5 (21.7)
Hepatic impairment or disorder	20 (51.3)	13 (65.0)
Pregnant or nursing	20 (51.3)	20 (100.0)
Demographics	14 (35.9)	4 (28.6)
Age (year ranges and menopause)	8 (20.5)	4 (50.0)
Weight	6 (15.4)	0 (0)
Drinking alcohol during study period	1 (2.6)	0 (0)
Sexual activity during study period	1 (2.6)	0 (0)
“Good general health”	1 (2.6)	0 (0)
Hospitalized patients	2 (5.1)	0 (0)
Patient’s unwilling to forgo blood donation for 3 months	1 (2.6)	0 (0)
Failed previous antibiotic therapyc	13 (33.3)	1(7.7)

^aValues shown are n (%).

^bExample: “There are no adequate and well-controlled trials in pregnant women. ORBACTIV should be used during pregnancy only if the potential benefit justifies the potential risk to the fetus”.

^cPrevious antibiotic therapy that did not fail was considered a reasonable exclusion criteria and not in conflict with the indication. This type of exclusion was not counted in this category.

Two antibiotics, bedaquiline and ceftazidime-avibactam, were approved for patients with limited or no alternative treatment options. However, pivotal trials for these drugs specifically excluded patients with no alternative treatment options and control groups were given current effective standard of care.

DISCUSSION

US law and regulation requires that marketing approval for a given indication should inform clinicians and patients about which uses of the drug, for which types of patients, are supported by substantial evidence of safety and efficacy. To obtain regulatory approval, a drug must be safe and effective for the conditions of use listed in the label. Although labels did provide information regarding certain excluded populations (eg, all labels approving use in pregnant women also explicitly stated that trials excluded pregnant women), many populations excluded from trials were not mentioned in labels. Examples of excluded populations not mentioned in labels included burn victims, diabetics, older adults, those with more severe infections, those with immunosuppressive diseases, and patients with no effective alternatives.

The exclusions from clinical trials have important clinical, scientific, and ethical consequences. First, the types of patients excluded are often those patients who would most likely be the target population in clinical practice. Namely, patients excluded are those in whom disease due to antibiotic-resistant pathogens are often diagnosed.⁶ This is discordant with aspirations of developing real-world evidence for more effective therapies in antibiotic-resistant disease in patients with no options as it precludes the ability to evaluate drug efficacy in these vulnerable populations. Although some have emphasized the challenge of enrolling sufficient participants with multidrug-resistant disease (at least before epidemic spread),⁷ excluding clinically relevant patients makes completion of trials even more challenging. Ensuring such patients enroll in trials is the only way to develop a robust evidence base for the drugs’ effectiveness in these populations. Restricting marketing approval to only studied populations might incentivize needed studies in populations with unmet needs.

Second, clinicians and patients may assume that efficacy and harms have been evaluated in these patients when in fact they have been specifically excluded. The safety and efficacy of these drugs in unstudied populations are often not predictable; large differences in treatment effects of drugs often depend on patient factors. For example, the absolute effects of penicillin in the 1930s and 1940s on improving survival in pneumococcal pneumonia were much greater in older, sicker patients than they were in younger, healthier patients.⁸ More recently, a meta-analysis of clinical trials evaluating the effects of tigecycline compared with older effective therapies showed numerically worse survival in patients from trials with disease due to resistant pathogens than in trials with disease due to susceptible pathogens.⁹ Clinicians cannot rely on in vitro susceptibility testing to assume the benefits of new drugs in unstudied types of patients because these do not account for differences in patient factors and clinical practice. A number of new drugs with promising in vitro and animal data have shown worse outcomes in adequate and well-controlled clinical trials of new antibiotics.^10,11

Third, these exclusions raise ethical issues, as trial participants who do not have unmet medical needs are asked to bear the risks of evaluating new therapies even though they already have existing, often cheaper,¹¹ effective therapies.¹⁰ Most antibiotics evaluated in this study occurred before the FDA’s release of draft guidance¹² regarding the limited population pathway for antibacterial and antifungal drugs established by section 3042 of The 21st Century Cures Act (2016).¹³ The guidance decreases the required length, size, and number of clinical trials necessary for approval but does not require evaluation of added benefits in patients with unmet needs. This could increase the number and types of populations unstudied or excluded from preclinical trials and thus exacerbate the problem.

Study Limitations

Our study was limited to those indications that we could examine. Five antibiotics (telithromycin, gemifloxacin, cefditoren, omadacycline, and plazomicin, representing eight approved indications) could not be examined as the FDA scientific reviews either lacked any information regarding the inclusion and exclusion criteria of pivotal studies or had redacted such information.

CONCLUSION

Approving antibiotics for populations in which they were never studied may lead clinicians and patients to assume benefits for patient groups that have not been demonstrated, not address clinically relevant questions for patients with unmet needs, and increase costs for unclear benefits. Labels should be revised to limit approved uses to only those that are supported by evidence as studied in pivotal trials that form the basis for approval.