Developing antibiotics for children: There are no easy answers

Aside: If you’ve enjoyed the prior discussions of movies to inspire antibiotic R&D and very apropos given the theme of today’s newsletter, please check out the newly released 4-minute YouTube discussion of a scene from Master and Commander in which antibiotics could have saved a young man’s arm!

Dear All (Wonkish alert! There’s a lot of detail here … please refill your coffee and settle in!),

HL Mencken famously noted that “For every complex problem there is an answer that is clear, simple, and wrong.” This comment is very apropos the problem of pediatric drug development, an arena where good intentions collide with practical reality in sometimes surprising ways. I gave an initial tour of the current guidance on pediatrics in this 7 July 2020 newsletter and you might want to (re)read it in parallel with today’s note for which we have several really interesting elements to juxtapose. Addendum: There is also a 27 May 2022 newsletter entitled “Antibacterial Guidance (Including Pediatrics): Parallel EMA+FDA Updates” that provides a further build on this discussion.

To keep you oriented as you read this relatively complex newsletter, here’s a preview of the ideas:

1. Pediatric labeling is often delayed by years
2. Tick-tock! This is a double whammy for kids as a new antibiotic may have a limited life-span!
3. There is no single, simple cause of the lag and hence no single, simple fix
4. Creative solutions addressing even part of the problem should be pursued and implemented

First up on our tour, and related to the ideas in the recent newsletter about Stewardship and Access plans for new antibiotics, the team at Access to Medicine Foundation (ATMF) have released a report entitled “Closing gaps in access to medicine for children: How R&D and delivery efforts can be ramped up” that has this headline message:

• “Out of more than a thousand R&D projects assessed in the 2021 Access to Medicine Index, only 69 (7%) are for children under 12.”
• And on the theme of access, ATMF found that of the 56 of those pediatric projects that are now in Phase 2+, 31/56 (55%) have an access plan.

Although that 7% may look low, ATMF is counting here projects that are focused solely on children. The total number of R&D projects that have (or will have) a pediatric R&D element must be much higher given that essentially all projects reaching Phase 3 will/must have at least a basic pediatric plan given that (i) FDA’s PREA (Pediatric Research Equity Act, passed in 2003) and (ii) EMA’s Pediatric Regulation (passed in 2007) effectively mean that any drug that is potentially useful for children must have a pediatric development plan. For details, see these comments from FDA (slide deck, webpage) and these materials from EMA. 

Despite these positive signs that both pediatric R&D and access planning are happening, delays in labeling of drugs for children (and hence availability/access in any part of the world!) remain the norm. Average lags from adult to pediatric labeling of about a decade are typical (see Noel et al. J Peds ID Soc  2019 and Noel et al. Ther Innov Reg Sci 2021). A report by the EMA released in 2017 (year 10 of the Pediatric Regulation) and a summary from a related 2018 workshop note delayed completion of pediatric studies is common due to factors that range from impact on quality of life for the family supporting the trial to difficulties with accurately predicting patient availability. A further report from 2018 noted that only 38% of pediatric trials for 122 medicines authorized during 2010-2014 had been completed after a median of 7 years!

And given that antibiotics may have a limited lifespan before (new) resistance develops, these lags are in effect a double whammy for children: Guidance is not available during the drug’s most useful initial years … and by the time the guidance is available, the drug may have lost some of its utility! 

So, what is the source of these lags? Well, the blunt truth is the pediatric development is often delayed by factors that you’d never expect. Although ATMF optimistically notes that, “Companies can accelerate paediatric trials where appropriate to close this gap by, for example, simultaneous enrolment of adults and children in clinical trials and invest in innovative technologies that may be applied to more than one dosage form type and/or route of administration,” the reality on the ground is simply much harder. Let’s take a look.

—
For our second stop on this tour, consider the recent commentary in Lancet ID by Williams et al. entitled “Harmonising regulatory approval for antibiotics in children.” This paper specifically focuses on a difference between FDA and EMA for safety data. This requires some picking apart, so let’s start with text from the key documents. Here’s what EMA says:

• Key document: EMA’s 2018 draft guidance entitled “Addendum to the guideline on the evaluation of medicinal products indicated for treatment of bacterial infections to address paediatric-specific clinical data requirements”:
  • Note: This document is marked as applying to anti-bacterial agents, but one would hope that similar principles would also be applied for other anti-infective agents.
  • Efficacy: “An extrapolation of the efficacy of an antibacterial agent across age ranges is considered possible for the majority of infectious diseases…”
  • Safety: “For the most part, a similar safety profile is expected in adults and paediatric subjects when the paediatric dose regimens achieve similar systemic exposures to those in adults, supporting an extrapolation of safety from adults to paediatric subjects. Safety data may need to be generated in the paediatric population or in specific age subsets if there are emerging concerns…”
  • PK (1): “In many circumstances, only paediatric pharmacokinetic studies will be needed. Depending on the study population, the test antibacterial agent could be administered alone, as adjunctive therapy or on top of an effective antibacterial regimen.”
  • PK (2): “The paediatric pharmacokinetic data may be obtained in patients with one or a limited range of the infectious diseases for which use of the antibacterial agent is proposed, taking into account whether pharmacokinetic differences were observed in adults depending on the site of the infection.”
• Summary by Williams et al.: For EMA, “both efficacy and safety can be bridged from adult studies for well established classes of antibiotics and children can be recruited into pharmacokinetic studies with any relevant clinical infection“

And here’s what FDA says:

• Key document: FDA’s Draft 2020 guidance entitled “Development of Anti-Infective Drug Products for the Pediatric Population”:
  • Note: This document is marked as applying to anti-infectives in general, not just antibacterial agents.
  • Efficacy: “Efficacy results from adequate and well-controlled clinical trials in adult patients can be extrapolated to pediatric patients if … the course of the infectious disease is similar in adult and the pediatric populations AND … effects of the drug product are sufficiently similar in adult and pediatric populations.”
  • Safety: “… however, pediatric data will be needed to assess the safety … of the drug product.“
  • PK (1): “pediatric data will be needed to assess the … pharmacokinetics of the drug product. The sponsor should demonstrate that the proposed dosing regimen results in similar exposures in pediatric patients as in adult patients.”
  • PK (2, with some safety thrown in)  Although the guidance lacks a statement of the form “… and safety and PK data must be obtained in a controlled study of the drug product in the indication(s) for which the drug is / will be indicated in adults,” the text implies this throughout by speaking of flexibility in RCT design and choice of comparator. There may be some wiggle on PK but extrapolating safety does not appear possible. Let’s look at an example…
  • PK/Safety Example: The FDA NDA approval letter (16 Jul 2019) and sNDA approval letter (4 Jun 2020) for imipenem-cilastatin-relebactam taken together show that this program was agreed for children from birth to less than 18 years of age:
    •  An open-label, single-dose study of pharmacokinetics, safety and tolerability in children with proven or suspected Gram-negative infections
    • A randomized, open-label, active controlled trial evaluating safety and tolerability in children with any of the adult indications (complicated urinary tract infections, complicated intra-abdominal infections and hospital-acquired bacterial pneumonia or ventilator-associated bacterial pneumonia)
• Summary by Williams et al.: “Efficacy can be extrapolated from adult studies … but draft FDA guidance continues to require de novo generation of primary safety data in children obtained in randomized trials … of children with the clinical indication for which the agent is licensed in adults.“

Summarizing, both agencies agree that efficacy can (usually) be extrapolated if you get the PK right. But, for safety we have a small shift with big implications given that new Gram-negative agents often have an initial indication focused on (and limited to!) cUTI:

• EMA: Multi-dose PK with any relevant infection may be enough to allow extrapolation of safety and efficacy
• FDA: Single-dose PK can be studied in any infection but multi-dose PK/safety must be in the infection(s) for which the drugs is indicated in adults

Given that we do expect PK to predict efficacy, the most fundamental pivot here to my eye is one’s view on the meaning of safety data in the small numbers of subjects that can be studied in a pediatric trial. Remembering the statistical rule of 3 (converting 3 ÷ study size to % gives the approximate 97.5% upper bound confidence interval estimate of the least frequent events you are likely to detect), a pediatric study of (say) 60 children would reliably detect adverse events occurring at a rate of at least 5% (3 ÷ 60  = 1/20 = 0.05 = 5%). EMA is (effectively) saying that this is too insensitive to be helpful, whereas FDA is saying let’s get at least some data. 

While I see the merits of both perspectives, enrolling even such a small number of children into a clinical program is tough, expensive, and further requirements really need to be minimized! And, it’s not like there are easy ways to resolve this: asking for more data always leads to slower results and hence contributes to those 10-year lags we saw in pediatric labeling relative to adult labeling and the above-noted double whammy for pediatric use of new antibiotics. As a start on agreeing the amount of data needed for a global program, the team at PENTA-ID wrote in 2019 with ideas for standardized inclusions, exclusions, and sample sizes that could be used for harmonized study designs. In particular, and reflecting back to the prior paragraph on safety and sample size, this paper suggests that reliably detecting events that occur at a 5% rate is a good general goal … hence 60 subjects is a good general sample size.

Another challenge lies in creating and maintaining a form (or forms) of the product that can be given in children. Fundamentally, every variant presentation requires work to create and validate even if it is only a small unit dose for IV administration. All of these different presentations have to be manufactured, distributed, and discarded when they exceed shelf-life. And while it’s obvious that a smaller dose will probably be needed for a child, how much smaller? If dosing as a liquid or as an IV is possible, then there is a lot of inherent flexibility, but solid forms require you to pick a unit dose (that is, a specific size such as 1 mg) and stick with it. And unless you’re willing to just work with a fraction of an adult unit dose, even the IV/liquid presentations require some guesswork about size. 

Aside: If you are interested in pediatric formulation work, WHO and Unitaid released a very comprehensive survey in 2020 entitled “Innovative Delivery Systems for Paedatric Medicines: Technology Landscape.” It surveys everything from minitabs to needle-free injections. Check it out!

—
So for our 3rd stop on our tour, let’s loop back and reconsider ATMF’s comment that trials can be accelerated by “simultaneous enrolment of adults and children in clinical trials and investment in innovative technologies that may be applied to more than one dosage form type and/or route of administration.” Perhaps true if things break favorably with no toxicities in adults of note (and hence support for exposing children early in the program), a ready way to adjust the dose for children of different ages, and no great anxiety over differences in clearance in children (kidney function in particular is different in young people). But, it can also be a difficult project that requires substantial effort and $.

Ultimately it’s all only time and money, but the money side of it is of course something in painfully short supply for antibacterial developers! As noted by Noel et al. 2021, pediatric studies require more resources than adult studies. At just the most basic level, suitable patients are less frequent and simply keeping study sites open is costly. In the EU, completion of the PIP can add up to 2 years to the marketing exclusivity period, but this is only valuable if the compound is actually generating sales-related income … and that’s of course the problem as new antimicrobials are placed on reserve and generate little immediate income for the developer. If all of this is new to you, the video and related materials cited in this 10 Feb 2021 newsletter are a good introduction to the overall problem. 

This can be seen to be having direct impacts even now — the withdrawal of plazomicin’s EU marketing authorization was directly linked in part to the cost of pediatric development: “The conditions and work and cost expected to be required for approval and post-approval, including the PIP and terminal sterilization processing, were evaluated and assessed to result in the product being financially and commercially unviable with the limited indication that was to be accepted.”

One bright spot here is that Noel et al. 2021 points to a growing push to enroll adolescents (ages 12+) in Phase 3 trials with adults. As this paper notes, dosing is similar in these groups about 95% of the time with adjustments readily predicted by allometric scaling. Especially when a new drug may be lifesaving or may address severe morbidity, the idea of generating data on 1/3rd of the pediatric population along with data on adults is appealing. There are some issues to address (e.g., operational issues in that sites that can enroll adults are not always able to handle children and vice versa) but taking this step aligns nicely with the vision by both ATMF and Williams et al. to shorten the delay to pediatric approval. Where there’s a will, I’d like to think that there’s a way.

—
So this brings us to our 4th and final stop in which we consider the 2021 paper in CID by Taormina et al. (an FDA-based team) entitled “A Regulatory Review Approach for Evaluation of Micafungin for Treatment of Neonatal Candidiasis.” This fascinating paper reviews the FDA’s efforts to address a difficult problem. In brief, data on micafungin, a therapy for candidiasis in children, strongly suggested that notably higher doses than approved for use in adults (about 2.5x higher: 10 mg/kg/day vs. 4 mg/kg/day) were required when this fungus spread to the brain in neonates.

But, “adequate and well-controlled trials evaluating the efficacy and safety of drugs treating conditions unique to neonates are challenging due to difficulty enrolling patients.” Indeed, that is the entire problem of pediatric development in a nutshell — the data you’d like to have simply cannot be acquired in less than geologic time. What to do?

In a strategy that Solomon would have applauded, FDA added a Limitation of Use statement stating that a micafungin dose regimen of 4 mg/kg QD may be insufficient to treat neonatal brain infection. The clinician still has to do a bit of thinking to understand what this means, but the information is in the FDA label in a way that manages the FDA’s requirement to support all labeling with clinical data. Here’s the key text:

• Section 1 (Indications and Usage): In a subsection labeled “Limitations and Use,” we are told “The safety and effectiveness of MYCAMINE (micafungin) have not been established for the treatment of candidemia with meningoencephalitis and/or ocular dissemination in pediatric patients younger than 4 months of age as a higher dose may be needed.”
• Section 8.4: “Although the dosage for the treatment of candidemia with meningoencephalitis has not been established, antifungal activity in various CNS compartments in the rabbit HCME model (a rabbit model of Candida brain infection) and limited clinical trial data suggest that in patients younger than 4 months of age, dose regimens of 10 mg/kg once daily or higher may be necessary for the treatment of candidemia with meningoencephalitis. Safety data from clinical studies for MYCAMINE at dose regimens of 10 to 15 mg/kg once daily in pediatric patients younger than 4 months of age did not reveal new safety signals.”
• In short, Section 1 says 4 mg/kg looks inadequate but in Section 8.4 we learn that 10 mg/kg looks like a better guess (Hint, hint! A word to the wise is sufficient!).

Because of differences in legal frameworks, EMA’s assessment report label gets to the same place more quickly and simply states that “a dose of 10 mg/kg continues to be recommended for treatment of neonatal candidiasis in order to maximize the proportion of infants who would be expected to achieve the efficacious exposure.”

The point here is that even well-tested regulatory strategies can run afoul of clinical reality. FDA’s ruleset makes good sense and can be paraphrased to say that everything in the label is supposed to tie back to something from a clinical study that is relevant to the impact of the drug on something a patient can perceive. The hard thing is the additional layer that requires all of this to be linked to a specific indication (e.g., cUTI) and hence (usually for anti-infectives) a specific body site. This is the place where I think we need to find nuanced but firmly articulated ways to work with the fact that infections can and do occur in body sites that we will NEVER be able to study in the level of detail required to put them in the product label. This is true not just of pediatrics but indeed of all uses of anti-infectives.

—
Whew! That’s a lot to take in! Let’s close then by returning to these two quotes from the paper by Williams et al.:

• “We suggest there is a clear need to harmonise regulatory approval requirements for new antibiotics in children.”
• “A clear focus on reducing the time between new antibiotics being licensed for use in adults and children is necessary. An achievable goal of pediatric licenses being issued within 5 years of the adult license should be established.”

Indeed, indeed. Regulatory requirements are driven by all of us and reflect a range of views about the needed amounts of safety and efficacy data. Stated differently, ensuring that children have prompt access to new antibiotics will require community-level pragmatic choices between the perfect and the merely good. In particular, we need to ensure that a single pediatric program will be acceptable globally. In theory there is no difference between theory and practice but in practice there is!

Post-newsletter addendum: WHO convened a discussion of Pediatric Drug Optimization (PADO) for Antibiotics during Nov-Dec 2022 and released a report on gaps and strategies for antibiotics on 24 Mar 2023. WHO has a generalized PADO strategy and has published results of similar exercises for HIV and tuberculosis.

All best wishes, –jr

John H. Rex, MD | Chief Medical Officer, F2G Ltd. | Operating Partner, Advent Life Sciences. Follow me on Twitter: @JohnRex_NewAbx. See past newsletters and subscribe for the future: https://amr.solutions/blog/. All opinions are my own.