Case studyJanssen Pharmaceutical - Bedaquiline

Clinical Development On-campus

OpeningDeveloping a new treatment for tuberculosis for patients with limited current

treatment options

On Monday, November 24, 2004 Koen Andries, a senior discovery scientist at the

pharmaceutical company Janssen Pharmaceutica in Beerse, Belgium (a subsidiary of

the American multinational pharmaceutical company Johnson & Johnson) received

word from the editor of the scientific journal Science that his manuscript reporting

the discovery and development of the new antibiotic R20791001 for treating tuber-

culosis (TB) had been accepted for publication. This was exciting news and was in

stark contrast to a previous message regarding the very same manuscript a few weeks

earlier in which the editor stated that the manuscript had been rejected and would

not be sent for review (Exhibit 1). After a second, closer look, the editor reconsidered

their earlier decision and sent the manuscript for review after all. In their assessment,

both reviewers were highly positive and advised Science to publish the manuscript;

as one reviewer wrote, “given the healthcare crisis as a result of TB-HIV co-infec-

tion and rising resistance to existing TB drugs, this is an extraordinarily important

study and… of broad interest to the scientific and healthcare community” (Exhibit

2). The reviewer’s statement that TB is a current health crisis clearly fits the current

situation in developing countries. Each year, approximately 2 million people die of

TB, which corresponds to more than 5,000 patients each day! A full one-third of the

world's population is estimated to carry a latent M. tuberculosis infection, and 10% of

these individuals will eventually develop TB. Because it was such a landmark study

and was highly relevant to the field of TB, Science’s editor informed Andries that the

paper would be published online on December 9, 2004 and would appear in print on

January 14, 2005.2

Publication of their paper in the prestigious journal Science was a triumph for Andries

and his colleagues. The research that led to the discovery of these new anti-TB phar-

maceuticals began seven years previously and had not been easy. Along the way,

numerous challenges arose, many issues had to be resolved, and internal committees

had to be convinced of the potential of R207910.

1 Since the founding of Janssen Pharmaceutica in 1953, all compounds synthesized at the company were assigned a distinct R number. In 1955, R5 (ambucetamide) was the first Janssen Pharmaceutica drug approved by health authorities for the relief of menstrual pain. R5 was discovered in 1953 by the company’s founder, Dr. Paul Janssen.2 Andries K, Verhasselt P, Guillemont J, Göhlmann HWH, Neefs J-M, Winkler H, Van Gestel J, Timmerman P, Min Zhu, Lee E, Williams P, de Chaffoy D, Huitric E, Hoffner S, Cambau E, Truffot-Pernot C, Lounis N, Jarlier V. A diarylquinoline drug active on the ATP synthase of Mycobacterium tuberculosis. Science 2005 Jan 14; 307(5707): 223-7. Epub 2004 Dec 9.

Finally, in the middle of 2004, Andries and his colleagues (both within and outside the

company) were able to demonstrate that R207910 was a promising new antibiotic for

treating TB.

The results described in their 2005 Science paper were impressive. They reported

that R207910 binds a completely novel target, the first new target to emerge after

four decades of research in the field of TB. Andries’ team had also elucidated the

compound’s mechanism of action and found that the compound has potent in vitro

activity against both replicating and non-replicating bacilli. Andries’ team also showed

that R207910 was much more effective than reference compounds in a murine model

of TB infection, exerting significant bactericidal and sterilizing activity. In addition, the

compound was found to be equally active against drug sensitive (DS), multi-drug-re-

sistant (MDR), pre-extensively drug-resistant (Pre-XDR), and extensively drug-resis-

tant (XDR) strains of M. tuberculosis, and the compound’s specific target and mode

of action eliminated the issue of cross-resistance against other TB drugs. Lastly, early

data obtained from a clinical trial with healthy volunteers showed good safety and

tolerability. Of course, additional clinical research was required before they could

conclude that R207910 was indeed both safe and effective for use in treating patients

with TB; however, Andries was confident that R207910 would be approved by the

regulatory agencies once successful clinical trials were completed. Thus, years of hard

work overcoming many problems and challenges seemed to be over, and a phase IIa

7-days early bacterial activity (EBA) trial in TB patients in South Africa was already

under way. The future for R207910 looked bright, and a novel treatment for TB seemed

within reach for TB patients (Figure 1).

Figure 1. Classification of various tuberculosis strains and anti-TB antibiotics. DS: drug-susceptible tuberculosis, MDR: multi-drug-resistant, Pre-XDR: pre-extensively drug-resistant, and XDR: drug-re-sistant strains. The asterisks * indicate that Pre-XDR and XDR are subsets of MDR. (Source: Janssen Pharmaceutics)

Despite its early promise, just a few months later a dark cloud settled over the TB

project. First, Andries received the initial data from the phase IIa clinical trial with TB

patients. The results showed that although R207910 (or TMC207, as the compound

was also known) had some therapeutic effect at the highest dose tested, unex-

pectedly—and in sharp contrast with the mouse studies—R207910 seemed to be

less effective than the reference compounds. The second, however, was far more

devastating to the project: the follow-up data from the phase I trial with healthy

volunteers revealed that R207910 had a terminal half-life of approximately 5 months.

Andries knew that this would be considered as an undesirable attribute for a drug that

was also a neglected disease medicine that would likely fail to produce a profit. The

disappointing phase IIa results, together with the long terminal half-life, were shared

and discussed by top management while checking in at an airport in December 2015.

Karel de Beule, the product development team leader at the time, was informed that

the company would no longer continue to develop R207910.

Andries immediately realized that fast action was needed if he was to rescue

R207910. After all, R207910 was not simply a “me-too” drug or a lifestyle drug. Here

was a compound that could potentially save the lives of the increasing numbers of

patients with MDR-TB and XDR-TB strains of TB. Although Andries had established

an impressive track record as a successful drug-discovery scientist for more than 22

years, he knew that his reputation with the company would not be sufficient; the team

would have to come up with solid, convincing data. De Beule’s team would need to

pore through all of the results obtained in the past seven years in an attempt to find

data that would convince top management the reverse their decision to terminate the

TB project. As Andries was thinking of the best way to tackle this new challenge, his

computer beeped with an incoming email. Immediately upon reading the email, his

mouth went dry and his eyes went moist. The message came from a medical doctor in

the Ukraine who apparently had read his Science paper on R207910.

“Dear Sir,

[….] I am 39 now and I am dead-sick: I suffer from polyresistant tubercu-

losis of the spinal cord and brain ………the disease comes treacherously

towards my eyes. I called the Ukrainian embassy in Brussels and our dip-

lomats are ready to assist in delivery your medicine. Maybe …… you can

treat me as a volunteer? […..]

Please help me!

DO NOT TURN ME DOWN!”

This heartbreaking message hit Andries hard. This was not the type of email he

usually received from fellow researchers who wanted to exchange scientific ideas

and data; this was a desperate cry for help from a distressed patient with TB, a

terminal patient with no treatment options. He realized that—of course—there was

no way his company would be able to provide the compound to this patient, as its

clinical development was still in the early stages and the project had recently been

terminated. Even if top management would decide to reconsider their decision and

he could manage to deliver R207910 to this patient, his company would be heavily

criticized for providing an experimental, potentially dangerous compound to a

desperate, vulnerable patient. Andries became even more determined to find sufficient

arguments to convince the internal review committees and top management that the

development of R207910 should continue. He discussed his thoughts with De Beule

and other colleagues who also believed in the compound, and as a starting point they

decided to re-evaluate the pre-clinical and clinical data, summarized in the Investiga-

tor’s Brochure. They realized that in order to present their findings at the next meeting

of the three internal review committees, there was no time to lose.

Background

The early years at the University of Ghent

When Koen Andries was 18 years old, he needed to choose between his two interests:

medicine and veterinary science. He decided to study veterinary science at the

University of Ghent in Belgium, primarily because the program would take only six

years compared to seven years for medical school. During his studies, he initially

considered becoming a practicing veterinarian; but towards the end of his final year,

he had developed an interest in research and wrote a scientific paper. His professor

in veterinary virology was impressed by the paper and offered Andries a job as his

assistant at the university. Andries accepted and began immediately after graduation

in the summer of 1975.

Andries worked as a virologist at the University of Ghent for seven years. During that

time, he studied veterinary vaccines as well as the pathogenesis of vomiting and

wasting disease, a viral disease in pigs caused by the porcine hemagglutinating en-

cephalomyelitis (PHE) virus. Andries used immunofluorescence to understand the

pathogenesis of PHE by sacrificing experimentally infected piglets at several time

points and studying their organs.3 Using this method, which was considered

cutting-edge at the time,

3 Andries K, Pensaert MB. Immunofluorescence studies on the pathogenesis of hemagglutinating en-cephalomyelitis virus (HEV) in pigs after oronasal inoculation. Am J Vet Res 1980; 41: 1372-1378.

Andries and his colleagues demonstrated that the PHE virus first replicates in the

gastrointestinal tract and then migrates along the vagus nerve to the brain stem, where

it replicates in the vomiting center, ultimately producing the disease’s characteristic

clinical signs, vomiting and wasting. This discovery was Andries’ first major scientific

achievement, and his results were published in several scientific papers. When he

presented his work on vomiting and wasting disease at a conference on coronaviruses

in Munich in 1979, he apparently caught the eye of Dr. D.A.J. Tyrrell, who published a

report of the conference in Nature.F4 That positive report would help Andries get a job

at the University of Antwerp a few years later.

Applying for a job at Janssen Pharmaceutica

In 1982, the University of Ghent announced a hiring freeze, and Andries had to leave

the faculty. He applied for a part-time professorship position at the University of

Antwerp. At about the same time, he applied for a research position at Janssen Phar-

maceutica, and he was interviewed by several staff members. At Janssen Pharma-

ceutica, two scientists wanted to hire Andries for two different research projects. One

scientist wanted him to assess the in-licensing of a vaccine against foot-and-mouth

disease, a job for which Andries was well-suited based on his prior experience with

veterinary vaccines. The other scientist was doing preliminary antiviral research and

had identified a compound that was active against herpes simplex virus in vitro. He

was looking for a researcher to study this compound in further detail. During his

interview, Andries met with three or four staff members throughout the company’s

hierarchy. Eventually—and before he realized it—he was introduced to Dr. Paul

Janssen (or “Dr. Paul”, as people used to call him), the company’s legendary founder

himself. Dr. Paul was interested in Andries’ previous achievements and was becoming

interested in antiviral compounds. At around that time, the pharmaceutical company

Burroughs Wellcome had published the first positive antiviral results using acyclovir, an

antiviral against the herpes simplex virus. That paper had caught Dr. Paul’s attention,

making him realize that medicines could be used to cure viral diseases. When Dr.

Paul asked Andries what he thought about developing drugs for viral diseases, the

reply was brutally honest: “I don’t really know much about it. I do know that viruses

are replicated by the host cells, and that it’s not easy to interfere with that replication

without disrupting the host cell’s metabolism. But some antivirals have selective

activity, so I think it can be done. I’d certainly like to try.” Dr. Paul apparently liked

this frank answer and hired Andries, even though he admitted knowing little about

antivirals.

4 Tyrrell DAJ. Viral pathogenesis and virulence. Nature1979; 280: 194.

During his interview at Janssen Pharmaceutica, Andries mentioned that he had also

applied for a part-time teaching position at the University of Antwerp. The company

had no objections, and when Antwerp offered him the part-time teaching job, Andries

successfully combined his research at Janssen Pharmaceutica with his teaching re-

sponsibilities.

Andries’ first research projects in the company

Andries first worked on the compound with in vitro activity against herpes simplex

virus. His task was to characterize the compound’s activity and establish in vivo models

that could be used to confirm its activity. Andries tested the compound both on mice

and guinea pigs, but was unable to replicate the in vitro activity in vivo. This was a

disappointing finding, particularly because he was unable to explain it. Andries then

decided to test the compound in other cell lines, as it had only been tested previous in

one cell line. The results were immediately clear: the compound was only active in that

one cell line, and not in any other cell lines tested. This striking result likely explained

why he saw no activity in vivo. Although the scientist in him wanted to investigate why

this drug was so cell type-specific, the drug developer in him realized that it would not

likely lead to a viable new drug, so he abandoned this line of research.

In the meantime, he had identified another promising compound that was effective

against human rhinoviruses (the family of viruses responsible for the common cold),

and this compound formed the basis of Andries’ first major research project, a project

that would last more than a decade. Initially, Andries and his colleagues identified

compounds that were active against only a few of the more than 100 rhinovirus

serotypes.5 Later, they discovered another group of compounds that was active against

other serotypes. By combining these two chemical scaffolds into hybrid structures,

they created broad-spectrum antiviral compounds.6 In addition, they deciphered the

mechanism of action: the compounds bind to the virus’ capsid, preventing uncoating

of the viral particles in the host cell. Interestingly, these broad-spectrum uncoating

inhibitors were also effective against polioviruses and other picornaviruses. It was a

very exciting time in the antiviral research world, and the team had several scientific

breakthroughs.

For example, by screening a panel of capsid-binding compounds against all known

rhinovirus serotypes, they were able to decipher the phylogeny of the rhinovirus

5 Andries K, Dewindt B, De Brabander M, Stokbroeckx R, Janssen PAJ. In vitro activity of R 61837, a new antirhinovirus compound. Arch Virol; 1988, 101: 155-167.6 Andries K, Dewindt B, Snoeks J, Willebrords R, Van Eemeren K, Stokbroekx R, Janssen, PAJ. In Vitro Activity of Pirodavir (R 77975), a Substituted Phenoxy-Pyridazinamine with Broad-Spectrum Antipicor-naviral Activity. Antimicrobl Agents and Chemother 1992; 36: 100-107.

group.7 Several compounds moved into phase I and phase IIa trials with human

subjects. The synthetic antiviral R61837 became the first compound with demon-

strated clinical effect in human volunteers infected with rhinoviruses.8 R77975 (i.e.,

pirodavir) became the first broad-spectrum compound effective against rhinoviruses.

Unfortunately, however, and as is often the case, the drug’s ADME profile9 was the fly

in the ointment. The team tried to develop orally bioavailable alternatives, but that

was problematic as well—because the compounds were esters, esterase activity in the

plasma rendered them inactive upon absorption. The results were disappointing, and

the project was terminated.

Exciting years in HIV research

In 1986, Andries began another large research project that focused on developing

antivirals against HIV. At the time, Dr. Paul had become highly interested in the

agent that causes AIDS. Science and Nature published several papers written by the

scientist Peter Duesberg, who was quite skeptical of the idea that AIDS was caused

by a virus. Andries and Dr. Paul had many lively discussions about the cause of AIDS

and whether drugs against HIV might be used to treat the disease. In the end, they

became convinced that HIV was indeed the cause of AIDS, and they decided to

initiate a project to develop antivirals that would be effective against HIV. However,

this was not a trivial undertaking. In those early days, only a few labs in the world—

including the group of Luc Montagnier and Françoise Barré-Sinoussi and the group

of Robert Gallo—had samples, and initially no one had been able to culture the virus.

So Dr. Paul was pushing Andries to find a solution: “What can we do? What can we

do?” But Andries was busy with his anti-rhinovirus research and did not know how

he could undertake another major research project at Janssen Pharmaceutica. In

addition, Dr. Paul had concerns about bringing HIV to the company. In those days,

nearly everybody was afraid of the virus, a situation comparable with today’s fears

surrounding the Ebola virus. Dr. Paul feared a negative response from labor unions

if they brought HIV to the company. Thus, they had to overcome this obstacle if they

were to develop antivirals for treating AIDS.

7 Andries K, Dewindt B, Snoeks J, Wouters L, Moereels H, Lewi PJ, Janssen PAJ. Two groups of rhi-noviruses revealed by a panel of antiviral compounds present sequence divergence and differential pathogenicity. J Virology 1990; 64: 117-1123.8 Al-Nakib W, Higgins PG, Barrow GI, Tyrrell DAJ, Andries K, Vanden Bussche G, Taylor N, Janssen PAJ. Suppression of colds in human volunteers challenged with rhinovirus by a new synthetic drug (R61837). Antimicrob Agents and Chemother 1989; 33(4): 522-525.9 ADME is a pharmacokinetics and pharmacology term meaning absorption, distribution, metabolism, and excretion. A compound’s ADME profile describes its disposition within an organism. These four factors influence the drug levels and the kinetics of drug exposure in the tissues and therefore influence the compound’s performance and pharmacological activity.

In 1986, Andries attended a meeting on antiviral research in Rotterdam, the

Netherlands, where he met Dr. Rudi Pauwels, a young researcher working in Professor

Eric de Clercq’s group at the University of Leuven, Belgium. Professor Eric de Clercq

had obtained an HIV sample from Japanese virologist Hiroaki Mitsuya and was able

to culture the virus in MT4 cells in vitro. Pauwels was using phenotypic screening to

identify antiviral compounds. In this method, the virus is produced by cells in culture,

and test compounds are added to the cells. The antiviral activity of the compounds

is then measured by quantifying the inhibition of virus production. Andries was quite

pleased to find a researcher who was culturing the virus and screening antivirals using

an approach that he felt was likely to succeed. Andries then discussed this promising

meeting with Dr. Paul, who immediately agreed to collaborate with Pauwels to

develop new drugs against HIV. In this collaboration, Janssen Pharmaceutica would

provide the compounds for screening, and the viral testing would be done at the

University of Leuven.

Once Janssen Pharmaceutica and the University of Leuven agreed on the specifics

of their collaboration, Dr. Paul and Andries selected a list of compounds for testing.

At that time, Janssen Pharmaceutica had nearly 100,000 compounds in its chemical

library, which was far too many to screen using Pauwels’ assay. Rather than choosing

compounds at random, Dr. Paul and Andries selected representative compounds from

a variety of chemical classes and scaffolds. The final “short” list contained approxi-

mately 600 compounds, which Pauwels then tested using his in vitro assay. Within six

months, he identified a compound—called R14458—that had in vitro anti-HIV activity

when applied at a concentration of 14 µg/ml. Importantly, R14458 had a selectivity

index of 10, meaning that the cytotoxic concentration was ten-fold higher than its

antiviral concentration, suggesting that the compound had high antiviral specificity.

R14458 was the first member of what would eventually become the tetrahydro-imid-

azobenzodiazepine (TIBO) class of compounds. After the discovery of R14458 in the

late 1980s, medical chemists were recruited to the project. Thus, Dr. Paul asked Mike

Kukla at Janssen Pharmaceutical’s Spring House, Pennsylvania division to synthesize

new analogs of R14458.

An interesting finding was that R14458 represented a subclass of compounds in their

extensive library. Once they learned that R14458 had anti-HIV activity, Andries and

his colleagues immediately looked at the other compounds in this subclass that

were structurally related to R14458 but were not included in the initial list. However,

although approximately 50 other compounds were chemically similar to R14458,

R14458 was the only member of the subclass with antiviral activity for HIV. Thus, their

initial choice of R14458 was extremely fortuitous.

Dr. Kukla synthesized several analogs of R14458 that were active against HIV, and he

increased the antiviral activity considerably by making various changes to the chemical

structure. Ultimately, the research team had several compounds with robust anti-HIV

activity in vitro. They had also elucidated the mechanism of action—compounds

appeared to act by inhibiting the virus’ reverse transcriptase via a mechanism that

differed from existing, nucleoside-based inhibitors of reverse transcriptase. Thus,

Janssen Pharmaceutica and their collaborators had developed the first non-nucle-

oside-based reverse transcriptase inhibitors (or NNRTIs). These exciting new results

were published in Nature.10

Unfortunately, once again the ADME profile was a fly in the ointment, as the

compounds were absorbed poorly when taken orally. In the UK, Prof. Brian Gazzard

tested the compounds in AIDS patients by administering the drugs intravenous-

ly. However, the results were disappointing, as they had no effect on the CD4

count, which was used as a readout of antiviral activity (in those days, researchers

were unable to measure viral loads). Ultimately, the TIBO group of compounds

was abandoned; in the meantime, the research team had discovered several other

chemical scaffolds with anti-HIV activity, some of which were bioavailable when taken

orally.

At about the same time, competitors had successfully developed their own NNRTIs.

For example, Boehringer Ingelheim and DuPont Pharmaceuticals developed

nevirapine and sustiva, respectively. So even though Andries, Dr. Paul and Pauwels

first discovered the NNRTI class of compounds, two rival companies had more success

in terms of developing orally bioavailable, clinically active NNRTIs.

In the early ninethies, the first NNRTI-resistant HIV strains were reported in patients

who had been treated with competitors’ NNRTIs. In 1993, Andries realized that

Janssen Pharmaceutica was too far behind its competitors to develop simply another

class of first-generation NNRTIs. Therefore, he suggested to aim for second-genera-

tion NNRTIs that would hopefully be effective against these new resistant strains of

HIV.

10 Pauwels R, Andries K, DeSmyter J, Schols D, Kukla MJ, Breslin HJ, Raeymaeckers A, Van Gelder J, Woestenborghs R, Heykants J, Schellekens K, Janssen MA, De Clercq E, Janssen PAJ. Potent and selective inhibition of HIV-1 replication in vitro by a novel series of TIBO derivatives. Nature 1990 (01 February 1990); 343: 470 – 474; doi:10.1038/343470a0

Also in 1993, it was discovered that HIV isolates carrying identical mutations in reverse

transcriptase did not have necessarily have the same IC50 values.11

This was a critical observation, as it suggested to Andries it would need a panel

of recombinant viruses, each with different point mutations in the same genetic

background if they were to develop drugs that would be effective against viruses with

first-line NNRTI mutations. A list of the most promising clinical mutants was quickly

prepared and discussed with Pauwels. To optimize the lead structures, Andries, Dr.

Paul and Pauwels decided to study the effects of individual chemical changes on

individual point mutations. In this respect, it was extremely important that the medical

chemist had a clear idea of what would happen if he modified the compound (in other

words, which changes to the compound would yield good activity on one HIV mutant

but not on another HIV mutant).

In 1995, Pauwels and colleagues—now at Tibotec—created the panel of NNRTI-resis-

tant HIV strains and began screening the chemical scaffolds again, this time against

both the original wild-type HIV virus and the new panel of resistant strains.

11 The half-maximal inhibitory concentration (IC50) is a measure of a compound’s ability to inhibit a specific biological and/or biochemical function. This quantitative measure indicates the concentration of a given drug is needed to inhibit 50% of a given biological process (or component in a process such as an enzyme, cell, cellular receptor, or microorganism). According to the FDA, the IC50 is defined as the concentration of a drug that provides 50% inhibition in vitro.

Pauwels did not remain at the University of Leuven. Although he was

a pharmacist by training, he was part of the medical faculty. And as a

trained pharmacist, his upward mobility in the medical faculty was rel-

atively limited. This was a source of frustration to Pauwels, an excellent

scientist who did solid research. He discussed his situation with Dr. Paul,

who offered to support him financially for five years so that he could con-

tinue his HIV research. This was certainly a generous offer, plus it saved

their promising HIV program. Then Dr. Pauwels made a clever move: he

took his agreement with Dr. Paul tofinancial investors. Because he had

secure income for the next five years, the banks were willing to lend him

additional capital. Pauwels now had the money he needed to create his

own biotech start-up company. Andries suggested the name “Tibotec”, a

reference to the TIBO compounds, and in 1994 Pauwels founded Tibotec.

Andries, Dr. Paul and Pauwels soon identified one scaffold that was not only active

against the wild-type strain as well as several—but unfortunately, not all—of the NN-

RTI-resistant strains. They realized that it might be possible to optimize this scaffold in

order to achieve broad-spectrum activity. However, their research was complicated by

the fact that their lead compound was metabolized rapidly by liver enzymes, again an

ADME-issue.

Therefore, Andries developed a bioassay by incubating all of the derivatives with

liver extracts, and Pauwels measured the anti-HIV activity of these derivatives both

before and after incubation, thereby measuring the extent to which the derivatives

were metabolized. Finally, fate smiled on the research team. Using their bioassay,

they were able to identify three separate metabolic “hot spots” in the lead compound,

and chemically modifying these hot spots yielded metabolically stable compounds.

At the same time, medical chemists Mike Kukla and Don Ludovici examined further

the activity against NNRTI-resistant mutants and eventually succeeded at making

chemical modifications that both increased the drug’s anti-HIV activity and increased

metabolic stability. This time-consuming process took nearly three years of intensive

research, but in April 1998, R147681 (which was later renamed TMC120, dapivirine) was

the first compound in the series to be selected by Janssen

Pharmaceutica for testing in humans. Shortly after that, new bioassay results provided

the key for a chemical modification that led to the discovery of R165335 (later renamed

TMC125), which was also known by its brand name, intelence.

In the summer of 1999, Andries recommended that R165335 undergo further

development based on its antiviral activity against NNRTI-resistant strains, its

metabolic stability, and its favorable plasma levels in animals. In December 1999,

Janssen Pharmaceutica’s internal Development Research Committee (DRC) agreed

to take R165335 to clinical development. In March 2000, a licensing agreement was

reached between Janssen Pharmaceutica and Tibotec (which was then a privately

held biotech company) for the development of intelence; in the agreement. Tibotec

did an excellent job developing intelence, which was not an easy task given that the

compound is highly insoluble. It was therefore quite an achievement to develop the

optimal formulation and bring the compound to the clinic. Their success paid off, and

intelence had robust activity against viral load in phase IIa studies. In 2001, Andries,

Dr. Paul and Pauwels had selected a second NNRTI, edurant, which had excellent

anti-HIV activity at extremely low doses, for further testing.

At about the same time, and independent from Janssen Pharmaceutica, Tibotec

acquired the rights to a protease inhibitor and performed additional chemical modifi-

cations, eventually developing the broad-spectrum protease inhibitor darunavir, inde-

pendently from Janssen Pharmaceutica. They applied the same strategy. A panel of

protease-resistant HIV strains was used to screen all three new compounds (intelence,

edurant, anddarunavir) for their activity against both wild-type and resistant strains of

HIV.

In 2002, Johnson & Johnson had the choice between leaving the HIV field altogether,

re-obtaining the licensing rights to intelence, or acquiring Tibotec, thereby obtaining

the rights to both intelence anddarunavir. They decided to acquire Tibotec, and

Tibotec became responsible for the clinical development ofdarunavir, intelence, and

edurant, which were approved by the FDA in June 2006, January 18 2008, and May

2011, respectively. Dapivirine is still being developed as a vaginal microbicide for use in

resource poor countries by International Partnership for Microbicides (IPM), an NGO.

The bedaquiline story

The search for non-azole-based antimycotics…

In the early 90’s, the famous mycologist Dr. Frank Odds joined Janssen Pharma-

ceutica to run the Mycology and Bacteriology department and to discover and

develop non-azole-based antimycotics. Janssen Pharmaceutica had been previously

successful at developing azole-based antimycotics such as ketoconazole and itracon-

azole, and they were now interested in developing non-azole-based antimycotics.

In addition to screening for new antimycotics, the company was also interested in

developing antibiotics against Helicobacter pylori. This was a strategic decision by the

company’s management. In the mid-1990s, Australian researchers discovered that

gastric ulcers were not caused by stress (as previously believed), but by the bacterium

Helicobacter pylori. Antacids, which were commonly used to control the symptoms

associated with gastric ulcers,12 drew large revenues, and this piqued the company’s

interest, as an antibiotic effective against H. pylori could potentially replace antacids

in the market. In his search for such an antibiotic, Odds decided to include Mycobac-

terium tuberculosis, the bacterium that causes TB, in his screen. This decision was

inspired by the World Health Organization’s statement in 1993 that TB had once again

risen to the level of a Global Health Emergency. This new TB pandemic was indirectly

due to the sharp increase in the number of HIV-infected patients: patients with AIDS

have a compromised immune system, which can activate tuberculosis in patients with

a latent M. tuberculosis infection.

12 Antacids are a type of medicine that can control the acid levels in your stomach.

Until that time, the research community had been generally complacent regarding TB,

as the disease was considered to be under control. As a result, very few researchers

were developing new antibiotics for treating TB. Spurred on the WHO statement,

Odds was keenly interested in developing compounds against this deadly bacterium.

However, a major obstacle stood in his way: his department did not have a biosafety

level 3 lab, which was required in order to work with Mycobacterium tuberculosis. He

solved this problem by using Mycobacterium smegmatis, a related—but nonpatho-

genic—mycobacterium.

Adopting this strategy was a bold decision and potentially a long shot, as compounds

that are effective against M. smegmatis are not necessarily effective against M. tuber-

culosis. The decision to initiate TB research at Janssen Pharmaceutica was not made

by upper management, nor was upper management even informed during the early

stages in the project. Indeed, the decision was made solely by Odds. This decision may

seem controversial in retrospect, but it is important to understand that researchers as

Janssen Pharmaceutica were generally allowed to perform small experiments without

consulting upper management, as long it did not involve excess costs or staff, and may

end up in a discovery. In the case of initiating TB research, the costs were relatively low

and only required adding some mycobacteria to a screen that was already planned.

The discovery of an antibiotic effective against H. pylori and

mycobacteria

Odds initially discovered a diarylquinoline compound which was synthesized by

medical chemist Jérôme Guillemont, with activity against H. pylori, and management

was informed. The level of activity was considered sufficient to initiate an optimization

project. Interestingly, the compounds that were found to be active against H. pylori

were also somewhat active against M. smegmatis (although the link had not been

established at that time). A lead optimization program was initiated, and R207910 was

one of the compounds included in the program. Despite his success, Odds’ primary

interest was finding new antimycotics that have a non-azole-based mechanism of

action. Unfortunately, he was not successful at that, and he left the company in 1999.

Taking over the TB project

Dr. Jef van Gestel was Odds’ successor. Van Gestel sent samples of R207910 to four TB

experts for testing against M. tuberculosis, but received somewhat conflicting results.

Two laboratories (one in vitro and one in vivo) reported that the compound was active,

whereas the other two laboratories (again, one in vitro and one in vivo) found no

activity.

In 2001, Janssen Pharmaceutica underwent a major reorganization. As a result, the

Mycology and Bacteriology department—where the TB research project had begun—

was closed, and Van Gestel retired, leaving R207910 sitting on the shelf. When Van

Gestel left the company, Andries took over the R207910 project. Although Andries

had been exclusively researching viruses for more than 25 years, he was very keen

on doing antibacterial research. His theoretical expertise by teaching Microbiology at

the University of Antwerp certainly helped him. He was passionate about developing

a drug for which there was such a high medical need. The challenges he faced were

clear. First, the results obtained from four external labs were conflicting. Moreover,

R207910 was extremely difficult to solubilize, given its apparent cLogP value of 7.2.13

In addition, R207910 was actually a mixture comprised two diastereoisomers with two

enantiomers each, and only one of these four chemical variants was active.

Developing a drug containing several chemical variants is generally not ideal, as

the inactive variants may have unwanted side effects. Moreover, obtaining the pure

enantiomer early in the development process is important from both a production

perspective and a financial perspective. Thus, medical chemists must find the best

strategy for producing the pure compound during a development process.

13 Solubility of a compound in both water and fat is one of the properties that determines its bio-availability, as an orally administered drug must first pass through the intestinal lining, be carried in aqueous blood, and finally penetrate the lipid-based cell membrane to reach the inside of the target cell. A model compound for the lipophilic cellular membrane is 1-octanol (a lipophilic hydrocarbon). The logarithm of the octanol/water partition coefficient (known as the LogP) is used to predict the solubility of an oral drug. If LogP is calculated (rather than measured experimentally), the term is called “cLogP”. A compound with a cLogP value of 7 means that the compound dissolves in organic solvents 107 times better than in aqueous solvents.

The discovery power of phenotypic screening

R207910 was discovered by phenotypic screening using cultured M.

smegmatis. Andries is a strong proponent of phenotypic screening, as it

is more efficient than using a target-based approach for screening new

anti-infection agents. “It is really quite easy to explain,” says Andries.

“With phenotypic screening with cultured mycobacteria, you’re actually

testing every compound against 614 targets at the same time, because

M. tuberculosis contain 614 genes that are essential for bacterial growth.

Thus, phenotypic screening is 614 times more efficient than single tar-

get-based assays. On the other hand, phenotypic screening can be a little

more labor-intensive. With some target-based screens, you can perform

high-throughput screening. With phenotypic screening, you can easily

test 100 compounds in a week or so, which I would call ‘medium-through-

put’ screening. Of course, you have to be smart in your approach. Rather

than randomly choosing compounds, you make strategic choices. This

approach has two key components: you need to select the best screen-

ing strategy, and you need to select the right compounds to screen. One

always talks about finding the needle in the haystack. I say you need to

find the right haystack first. Not every haystack contains a needle, and

some may contain several needles. At Janssen, our haystack—our chemi-

cal library—contains many needles. Why? Because Dr. Paul was an excel-

lent medical chemist, who motivated his medical chemists to synthesize

compounds with drug-like properties. In this respect, Janssen Pharma-

ceutica’s haystack is pure gold. But the sheer size of the library is not the

most important factor; quality, not quantity, is what matters most. You can

screen a million compounds and still not find a drug. I give Dr. Paul a lot

of credit for creating an excellent library of compounds. For example, we

discovered at least ten different scaffolds with activity against HIV. We also

found three or more scaffolds with nanomolar activity against respiratory

syncytial virus, and scaffolds with activity against rhinoviruses and the in-

fluenza virus. For nearly every virus we tested, we found reasonably selec-

tive compounds with moderate activity at µg/ml activity. If you then put a

medical chemist on the project, you can increase that activity by a factor

of 100 to 1000 in just a couple of years. In our hands, it worked every time.

People may say, ‘You were lucky.’ But if you are successful ten times in a

row, then there must be more to it than just luck.”

In 2001, the active enantiomer in R207910 was purified and tested in mice. The results

arrived in 2002 and were far from promising. At high doses, the drug killed the mice,

whereas at lower doses no response was measured. These findings—together with

the conflicting results from the four external labs—opened the project to skepticism at

Janssen Pharmaceutica, where other, commercially more attractive research programs

were competing for resources.

A setback, followed by amazing results

The disappointing initial data with mice notwithstanding, Andries and his colleagues

continued with their TB research, and in 2003 they found that R207910 had a unique

spectrum of antibacterial activity. In vitro tests with approximately 20 mycobacte-

ria, including M. tuberculosis and an MDR strain of TB, revealed that the minimum

inhibitory concentration (MIC)14 of ≤0.060 µg/ml, whereas the MIC for non-myco-

bacteria, including H. pylori, was significantly higher (≥32 µg/ml) (Exhibit 3). In the

meantime, scientists in the company’s formulation department solved the solubility

problem by using hydroxypropyl-beta-cyclodextrine, and Andries then tested the

new formulation in a mouse model of TB, with amazing results. When given as a

monotherapy, R207910 was as effective as—or in some cases, better than—rifampicin

monotherapy and the standard triple-cocktail containing the three frontline drugs

rifampicin, isoniazid, and pyrazinamide (Figure 2).

Figure 2. Efficacy of R207910 (bedaquiline; “B”) at reducing the number of bacterial colony forming units (CFU) in the lungs of an established mouse model of TB infection 4 and 8 weeks after infection. For comparison, mice treated with rifampicin alone (“R”) or with a combination of rifampicin, isoniazid (“H”), and pyrazinamide (“Z”) are also plotted. Data courtesy of K. Andries.

14 In microbiology, the minimum inhibitory concentration (MIC) is the lowest concentration of an antimicrobial that will inhibit the visible growth of a microorganism after overnight incubation. MIC is important in diagnostic laboratories to confirm resistance of microorganisms to an antimicrobial agent and to determine the potency of new antimicrobial agents.

Despite their high promise, these remarkable findings from the mouse studies did little

to silence the critics at Janssen Pharmaceutica. The most pressing questions were

related to the projected return on investment (ROI) for a drug that targets a neglected

disease and—perhaps more importantly—regarding the compound’s mechanism of

action (“If you don’t know the mechanism of action, how can you develop it into a

drug?”). Nevertheless, thanks in large part to support from his supervisor, Didier de

Chaffoy, Andries was able to continue working on the TB project. At the time, R207910

was one of the most promising drugs in De Chaffoy’s discovery pipeline. De Chaffoy

had the authority to approve testing through phase I clinical trials (pharmacokinet-

ics studies in human volunteers). He therefore approved preclinical toxicology testing

of R207910 in animals. Unfortunately, these preclinical toxicology studies gave the

Andries team an unexpected setback: at high doses, the drug caused over propor-

tional tissue accumulation, resulting in adverse effects in animals, and phospholip-

idosis15 due to its cationic amphiphilic properties. Although phospholipidosis can

be associated with myopathy and hepatic toxicity, this finding was not a complete

deal-breaker, as several other drugs on the market also cause phospholipidosis; nev-

ertheless, it was still a red flag. Thus, Andries and his team also attempted to create a

compound with better physicochemical properties; however, after considerable effort,

they found that eliminating the lipophilicity also eliminated the compound’s anti-TB

activity. Ultimately, they decided to continue developing the parent compound,

R207910.

Identifying the target and the mechanism of action

To convince his colleagues in the company, and to silence the skeptics, it became

extremely important to identify R207910’s target and its mechanism of action.

Therefore, Andries and his team isolated strains of M. tuberculosis and M. smegmatis

with in vitro resistance to R207910, and their colleagues in the company’s Functional

Genomics department sequenced the complete genomes of these strains for

comparison with their respective wild-type strains; it should be noted that in those

days, whole-genome sequencing was still quite a difficult process. However, their

efforts were rewarded with an interesting result: sequence analysis revealed that only

one bacterial gene—the atpE gene, which encodes the F0 subunit of ATP synthase—

contained mutations in all three mutant strains.

15 Phospholipidosis is a lysosomal storage disorder characterized by the excess accumulation of phos-pholipids in tissues. Many cationic amphiphilic drugs, including anti-depressants, antianginal, antima-larial, and cholesterol-lowering agents, are reported to cause drug-induced phospholipidosis (DIPL) in animals and humans. The mechanisms of DIPL involve trapping or selective uptake of DIPL drugs within the lysosomes and acidic vesicles of affected cells. (Source: Wikipedia)

The identification of these mutations provided compelling evidence that R207910

inhibits the proton pump function of ATP synthase in M. tuberculosis.

Despite this preliminary evidence, more data was needed in order to reach any firm

conclusions and—more importantly—to convince the research field. After all, during

more than four decades of research, no one had ever suggested that ATP synthase

may be a target for antibiotics. To confirm the novel but potentially controversial

finding that the atpE gene was responsible for developing resistance to R207910,

Andries performed complementation studies by transforming wild-type M. smegmatis

bacteria with a plasmid encoding the mutant F0 protein. The transformed cells were

resistant to R207910, and their MIC was identical to the source M. smegmatis strain.

The plasmid was then isolated back from the transformed bacteria, and the team

found that the mutation was retained. These results confirmed that ATP synthase was

the target for R207910. Identifying the compound’s target and its putative mechanism

of action were important achievements, as they increased the team’s likelihood of

securing internal support in order to continue developing the compound and to

publish the data in a top-tier scientific journal.

More—but manageable—complications

During the development process, Andries and his team discovered that R207910 was

subjected to oxidative metabolism primarily by the oxidizing enzyme CYP3A4, causing

the formation of the N-monodesmethyl (M2) metabolite. In mice, 80% of R207910

is converted to the M2 metabolite, which is both less active and more toxic than the

parent compound. The team then measured whether R207910 is metabolized in other

species, including rats, dogs, monkeys, and guinea pigs. In dogs, approximately 50%

of the drug was converted to the M2 metabolite, and was less toxic than in rodents. In

addition, the team found that virtually all (i.e., 99.9%) of the compound was bound to

proteins; thus, the free fraction was unmeasurable. Most PK/PD specialists consider

protein-bound compounds to be inactive, whereas only the free compound has the

desired activity. To get at this issue in further depth, Andries tested the effect of human

proteins on the MIC of R207910; to his surprise, including human proteins increased

the in vitro MIC only a factor of ten.

The metabolic conversion data and the finding that R207910 is almost exclusively

bound to proteins made it difficult to translate their in vitro and animal data to

the clinic. Moreover, it was difficult to establish doses for human volunteers and

patients in phase I and phase IIa studies, respectively. However, these difficulties

were manageable, and doses were determined for the first phase I study, which was

conducted in healthy male volunteers. In this double-blind, randomized,

placebo-controlled study, safety and tolerability were studied using six cohorts

containing nine subjects each. In each cohort, six subjects received the same single

dose of R207910 (10, 30, 100, 300, 450, or 700 mg), and three subjects received a

placebo. In a second trial with a similar design, three cohorts containing nine healthy

male volunteers received escalating doses (50, 150, and 400 mg) of R207910. In

this second study, each subject received either placebo (three subjects per cohort)

or R207910 (six subjects per cohort) for 14 days. Plasma analysis revealed that only

25-30% of R207910 is converted in humans to its main metabolite M2, and the

compound had an “effective half-life” of 24 hours. Importantly, the compound was

tolerated well in all cohorts and caused no consistent trends in the vital signs, electro-

cardiogram, or laboratory tests.

At this stage, Andries now had all the data he needed to write a paper for submission

to a top-tier scientific journal. Specifically, he had a compound from a new chemical

class, a new target with a novel mechanism of action, excellent results using a mouse

model of TB, and promising safety, tolerability, and pharmacokinetics data from two

independent phase I trials with healthy subjects. Thus, Andries and his colleagues

began writing a paper for submission to Science.

A big splash in the TB research community

Andries’ Science paper took everybody in the TB research community by surprise.

This was Andries’ first paper in the field of bacteriology and nobody in the TB

community had ever heard of him. Although he had published in the field of antivirals,

his research group was completely unknown in the TB field. Admittedly, the paper’s

last four authors—Vincent Jarier and Nacer Lounis from Paris, and Emma Huitric and

Sven Hoffner from Sweden—all of them were already established TB experts, but the

authors from Janssen Pharmaceutica, including the first author, were not. Thus, the TB

field was buzzing with excitement: here was the first new TB drug in four decades, with

a novel target and novel mechanism of action, robust activity, and much better than

the first-line TB drugs. Of course, not everyone in the field was convinced. Andries

later heard through other channels that his paper was discussed widely among the

TB research community, and quite a few researchers in the field were skeptical of the

data, even though it was published in Science.

The review by Science

Andries submitted their manuscript to Science in September 2004. Within

just a week, he received their reply. “To our utter surprise, our manuscript

was rejected. We had worked so hard to put together a complete picture.

We could have published a partial picture much earlier, but we wanted

the full story, and we wanted to include not only the drug’s discovery

and mechanism of action, but also some preliminary safety data and

even human pharmacokinetics. And after all that work, the manuscript

was rejected in a week. I asked a member of our company’s Public Rela-

tions department to contact Science and request a teleconference; not

to challenge their decision, of course, as their decision was made and we

respected that, but rather to understand the rationale behind their deci-

sion. Our plan was to submit an improved version to Nature, so it would

have been helpful to know how the paper could be improved. Shortly

after the PR department sent the email, Science replied. They had taken

a second look at our manuscript and said that if we still wanted to publish

in Science, they would send it out for review. Of course, we agreed imme-

diately. It turned out that the manuscript was originally screened by an

assistant editor because the editor was not available. When they received

our follow-up request, the editor evaluated the paper, realized her assis-

tant had been mistaken, and overruled her original decision.

Science sent the manuscript to two reviewers, and both were very posi-

tive. The paper was published, and we even got an image of the ATP syn-

thase on the front cover of the issue in which it was published. This was an

important lesson. If I would not have persist, and if we had not contacted

the journal, the paper would not have been published in Science.

Initially, despite our preliminary evidence, we were not 100% certain that

we had identified the correct target. Therefore, we opted for a cautious

title: ‘R207910 a diarylquinoline active on a potential new target of Myco-

bacterium tuberculosis’. To our surprise, Science changed the title to one

One of the paper’s co-authors was participating in a discussion with colleagues in the

US. The other participants in the discussion did not realize he was actually a co-author

of the paper, and many stated that they did not believe the results. They felt that the

high cLogP suggested that the compound might actually interact nonspecifically with

the bacterial cell wall. Andries had similar experiences when presenting the paper at

conferences; he often found that quite a few audience members did not believe that

ATP synthase was the target. After all, no one had ever heard of an antibiotic that

targets this protein, and ATP synthase is one of the most fundamental enzymes in

biology. Moreover, the F0 subunit—which was specifically implicated in the Science

paper as the target—is highly conserved across species. Skeptics simply did not

believe that it was possible to develop a drug that interacts specifically with ATP

synthase in mycobacteria without cross-reacting with the patient’s own mitochondrial

ATP synthase.

The Science paper included the relevant sequences of human and mycobacterial

ATP synthases as an explanation of R207910’s specificity. Resistant mutants contain

an alanine (A) to proline (P) substitution at amino acid 63 in the F0 subunit of myco-

bacterial ATP synthase. This single change in the protein increased the compound’s

MIC by a factor of a thousand. The importance of an amino acid substitution at this

position is supported by three mycobacterial species that are naturally resistant to

R207910, including M. xenopi (Figure 3). All three R207910-resistant species lack

an alanine residue at this position, and the MIC values are all 1000-fold higher than

R207910-sensitive (i.e., wild-type) M. tuberculosis. The human homolog also lacks an

alanine at this position, which explains why the patient’s ATP synthase is not affected

by R207910. Despite this evidence, and despite the complementation study, many

researchers in the field still refused to believe the bacterial ATP synthase was the

target. Andries explains his feelings at the time: “As scientists, we are trained to believe

the data. When someone sees a computer-generated 3D structure of a drug binding to

its target, they believe it, even though it may be completely inaccurate. We may like to

think we’re objective, but we’re really not.”

that was stronger: ‘A diarylquinoline drug active on the ATP synthase of

Mycobacterium tuberculosis’. Clearly, they found the data from comple-

mentation assay to be quite convincing.”

Figure 3. Amino acid alignment of ATP synthase F0 subunit from wild-type (R207910-sensitive) M. tuberculosis, R207910-resistant M. tuberculosis, M. xenopi, and H. sapiens. (Source: Janssen Pharma-ceutics with minor modifications).

In 2004, just after Andries discovered the drug’s target and mechanism of action,

Anil Koul joined his research group to examine whether R207910 was active against

dormant bacilli. Anil developed an assay in which the bacteria are inverted, with the

synthase target exposed to the outside. Anil was able to show clearly that R207910

binds to wild-type ATP synthase, but not to mutant F0 subunits. These results finally

convinced the skeptics in the TB community that the F0 subunit of mycobacterial ATP

synthase is the target of R207910.

Further discussion regarding the development of R207910

After completing the phase I studies, bedaquiline (the new name for R207910) went

to full development. At the time, two options were available for developing the drug.

The company had a development group based in the US with experience with the

antibiotic levofloxacin (a fluoroquinolone); however, they had no experience with TB.

Alternatively, they could hand the development over to Tibotec (which had already

been acquired by Johnson & Johnson by then), which was an attractive option, given

that they had been so successful at developing anti-HIV compounds. Because of

Andries’ close ties with Tibotec, he wanted bedaquiline to go with Tibotec, although

Tibotec had little experience outside of virology, and their development pipeline was

Mutations and polymorphism in

resistant mycobacteria

rather full with several promising antivirals, three for HIV and two for Hepatitis C (HCV).

Nevertheless, Andries was able to convince both Janssen Pharmaceutica’s upper

management, and clinical development of bedaquiline was transferred to Tibotec in

the summer of 2004 with De Beule appointed as the compound development team

leader. At Tibotec, scientists designed a one-week efficacy study in drug-susceptible

TB (DS-TB) patients to be performed in South Africa as a proof-of-principle study.

Because of his expertise in TB and microbiology in general, Andries stayed with the

compound’s development (which is rather unusual for a discovery scientist), and

became the development team’s Microbiology leader.

Disappointment from the first clinical trial with TB patients

A few months after the paper appeared in Science, the results from the Early Bacteri-

cidal Activity (EBA) phase IIa study became available. Although technically the study

was a success, the data were far from encouraging. In the trial, 75 treatment-naïve

patients with DS-TB were assigned randomly to receive once-daily monotherapy

with oral bedaquiline (25, 100, or 400 mg), 600 mg rifampin, or 300 mg isoniazid

for seven days. The results showed a statistically significant bactericidal effect (0.8

log10) in patients treated with the highest dose of bedaquiline, but only on the last

day of the experimental treatment (day 7). In contrast, the two reference compounds

rifampicin and isoniazid yielded much better results (Figure 4). This outcome was quite

unexpected, given that bedaquiline monotherapy outperformed all three first-line

TB drugs when tested in mice. Thus, the researchers questioned whether the mouse

data provided an accurate prediction of the compound’s therapeutic value in patients.

Also troubling was the finding that the patients who received bedaquiline had a small

increase in corrected QT interval (an increase of approximately 10 msec); which—

although not necessarily a deal-breaker—was certainly a concern.

Figure 4. Time course of bactericidal activity during 7 days of treatment with 25, 100, or 400 mg TMC207 (bedaquiline), 600 mg rifampin, or 300 mg isoniazid. The values are presented at the mean and 95% confidence interval. Log Fall indicates the change in log10 CFU/ml16 in the sputum from baseline (day 0). Note: at day 8 the log10 CFU counts are effected by standard TB treatment, which was initiated at day 8. (Source: Janssen Pharmaceutica).

The final nail in bedaquiline’s coffin

The final blow to the TB program came shortly after the clinical trial, when

management received the follow-up data from the phase I trial with healthy

volunteers. The data showed that bedaquiline had an effective half-life of approx-

imately 24 hours but also an extremely long terminal half-life, approximately five

months (Figure 6). In pharmacokinetics, the rule of thumb is that a compound remains

in the body for five times its terminal half-life, which translates to a retention time

of two years. To the company, this finding was a serious issue because low levels

of bedaquiline could still be measured in the patient’s plasma months after the last

drug intake and potential delayed toxicity could not completely be ruled out at this

point. The company was quite concerned, and upper management terminated the

development of bedaquiline in 2005.

16 In microbiology, colony-forming units (CFUs) are used to estimate the number of viable bacterial or fungal cells in a sample. In this context, viable is defined as the ability to multiply by binary fission under controlled conditions. Unlike most microscopic measures of cell viability, which count both live and dead cells, using CFUs requires culturing the microbes and therefore measures only viable cells.

Figure 5. Plasma levels of bedaquiline (TMC207) measured in healthy volunteers after the last dose. The black dotted horizontal line indicates the No Adverse Effect Level (NOAEL)17 determined in a 6-month study in dogs. At 6 months, the average concentration measured in the volunteers was 157 ng/ml. Effective half-life and terminal half-life are depicted by Eff.t½ and Term.t½ respectively. Data courtesy of K. Andries.

Andries discussed the situation with his colleagues. The decision had been made, and

it was “Game over” for bedaquiline. Nevertheless, some of his colleagues shared his

frustration. Andries explains the struggle he was facing: “Some experts focus so much

on their own specific expertise that they overlook to broader picture and don’t always

distinguish between development of a life-saving drug and development of a lifestyle

drug. TB is a life-threatening disease, and patients with extensive drug-resistant TB

have very limited treatment options and poor outcomes. Most of these patients simply

die. In such case, you must weigh the benefits for the patients against the potential

disadvantages. It’s a classic risk-benefit analysis.”

17 The No Observed Adverse Effect Level (NOAEL) denotes the level of exposure of an organism, found by experiment or observation, at which there is no biologically or statistically significant (e.g. alteration of morphology, functional capacity, growth, development or life span) increase in the frequency or severity of any adverse effects in the exposed population when compared to its appropriate control. In drug development, NOAEL of a new drug is assessed in toxicology experiments using laboratory animals. In pharmacology in addition to the NOAEL, often the Minimal Anticipated Biological Effect Level (MABEL) is determined to establish a safe clinical starting dose in human trials.

We cannot simply let it go!

After much consideration, Andries and colleagues concluded that they could not

simply let the project end. They wanted to resume development and bring this

compound to the clinic. They realized that they were facing huge challenges and

would need to overcome four major hurdles. They would have to convince (i) the

internal pre-clinical toxicology expert committee, (ii) the ‘First in Human’ committee

and (iii) the Clinical Advisory Board with external TB experts. Andries and colleagues

were well aware of the fact that they could not convince these internal committees

merely by making arguments, no matter how sound those arguments might be;

they needed convincing data. Even if they were able to convince the three internal

committees to re-start the development, subsequently they would be facing a though

meeting with the FDA to discuss the development path forward. The challenges were

enormous indeed.

Andries explains: “Looking back, the most important lesson I learned during my career

is that you need to be passionate about your drug; to convince others, you need to

serve as an advocate for the drug, be a ‘champion’ for your product. And in most cases,

people will be convinced only when presented with solid data. Of course, if you’re not

convinced yourself, you’ll never convince others; so you need to generate solid data

first. Once you’ve convinced yourself, you can attempt to convince your colleagues,

their colleagues in the internal committees, followed by upper management, and

finally the regulatory agencies in the countries where the patients are. Throughout the

entire process, you need to push; if you don’t push, you will lose.”

Readings

Before class, please read the following texts:

1. Tibotec - Investigator’s Brochure TMC207 (Bedaquiline), Edition 3, June 2006

and supplement data.

2. Andries K, et al., A Diarylquinoline Drug Active on the ATP Synthase of Myco-

bacterium tuberculosis Science 2005 Jan 14; 307(5707): 223-7. Epub 2004 Dec

9. http://science.sciencemag.org/content/307/5707/223.long

3. Cohen J. New TB drug promises shorter, simpler treatment. Science 10 Dec

2004; 306 (5703): 1872. DOI: 10.1126/science.306.5703.1872. http://science.

sciencemag.org/content/306/5703/1872

4. Cole ST, Alzari PM. TB - A new target, a new drug. Science 14 Jan 2005;

307(5707): 214-215, DOI: 10.1126/science.1108379. http://science.sciencemag.

org/content/307/5707/214

Additional information (Optional)

5. Ridley DB, et al., Developing drug for developing countries. Health Affairs,

2006; 25(2), 313-324. http://content.healthaffairs.org/content/25/2/313.full.

pdf+html

6. If you want to refresh your memory regarding basic pharmacology, you may

wish to look at the Pharmacology tutorials at http://handwrittentutorials.com/

videos.php. This site also contains other tutorials that you may find helpful.

7. In addition, you may wish to look at the Teaching Resource Centre Pharmacol-

ogy (TCR) database at http://coo.lumc.nl/TRC (you can create a free account at

http://www.medicaleducation.nl). You can download the TRC app for free from

the App Store and iTunes.

Exhibit 1. Rejection of the manuscript by Science

6 October 2004

Dr. K. Andries

Johnson & Johnson Pharmaceutical R & D Beerse

BELGIUM Ref: 1105869

Dear Dr. Andries:

Thank you for submitting your manuscript "R207910, a diarylquinoline active on

a potential new target of Mycobacterium tuberculosis" to Science. Because your

manuscript was not given a high priority rating during the initial screening process, we

will not be able to send it out for in-depth review. Although your analysis is interesting,

we feel that the scope and focus of your paper make it more appropriate for a more

specialized journal. We are therefore notifying you so that you can seek publication

elsewhere.

We now receive many more interesting papers than we can publish. We therefore

send for in-depth review only those papers most likely to be ultimately published in

Science. Papers are selected on the basis of discipline, novelty, and general signifi-

cance, in addition to the usual criteria for publication in specialized journals. Therefore,

our decision is not necessarily a reflection of the quality of your research but rather of

our stringent space limitations.

We wish you every success when you submit the paper elsewhere.

Sincerely,

Senior Editor

Science International Bateman House, 82-88 Hills Road, Cambridge CB2 1LQ, UK

Tel: +44 (0) 1223 326 500 Fax: +44 (0) 1223 326 501 Email: science@science-int.co.uk

Science International is the business name of a U.K. branch of AAAS Science International, a U.S. company.

Corporate Registration No. FC17250 Branch Registration No. BR556 Vat No. 626138545

Ref. 1:

While this is an exciting and important article that merits early publication

in Science, there are certain additions to it that would be helpful. These

are: Some indication of how the molecule was discovered. Clearly, this

was a process that did not start with the mycobacterial genome.

Is there any information on the MIC of M. ulcerans, as the infections with

this organism are currently very difficult to treat?

While much attention is paid to issues of efficacy, less is available on po-

tential toxicity. One needs some reference to screening results for geno-

toxicity and for toxicity tests on the usual two animal species. These

would supplement the rather scanty information from the phase I human

studies. In the description of the results in an established murine infec-

tion (Fig 4b), it would be helpful to note the bacterial load at the time that

treatment was started. This is an indication of the severity of the infection

at that time.

Ref. 2:

This manuscript describes the properties of a diarylquinolone recent-

ly patented and termed R207910 or compound J. This compound has

exquisite selectivity and potency for mycobacterial species, including

M. tuberculosis and the M. avium complex, for which effective drugs are

lacking. Related actinomycetes are only poorly inhibited, while a number

of both Gram-negative and Gram-positive pathogens are essentially un-

affected. The in vitro potency against M. tuberculosis is as good or better

than isoniazid and rifampicin, and the in vivo potency is a log order better

in one mouse model. Perhaps most importantly, the compound retains

activity against M. tuberculosis strains that are singly or multiply resistant

to commonly used anti-tuberculars.

Exhibit 2. Reviewers’ comment Science manuscript

Replacement of either isoniazid, rifampicin or pyrazinamide, a combina-

tion used in all “short course” therapies, with J caused enhanced steril-

ization in an established infection mouse model, arguing that inclusion of

J with one or more of these compounds (the authors correctly argue that

monotherapy would be ill-advised) could potentially further shorten the

course of TB treatment, and be efficacious against strains that were singly

or doubly resistant to isoniazid or rifampicin.

The target for the compound is the proton-translocating F0 subunit of the

ATPase, a novel target that has not previously been the target for anti-

bacterial inhibitor discovery. Strains resistant to J have single amino acid

replacements in this gene, and expression of the gene in multi-copy leads

to resistance. Alignments of the sequence of the gene provide a reason-

able rationale for the selectivity of the compound, and its toxicity profile.

There are numerous other studies reported, including a pharmacokinetic

study and preliminary human tolerance studies, all of which suggest the

compound exhibits desirable properties as an addition to the dwindling

number of effective compounds available to treat TB. As such, and given

the health care crisis as a result of TB-HIV co-infection and rising resis-

tance to extant TB drugs, this is an extraordinarily important study and

series of results of broad interest to the scientific and health care commu-

nity. I personally would be interested in seeing a short paragraph related

to the discovery of the compound class, without the necessity of entering

the patent literature and perhaps structures of other compounds that are

stated as being effective (without a complete SAR1 description). Some of

the detailed pharmacokinetic studies could be included in the supple-

mentary material if space is an issue. Overall, the manuscript is concisely

written and understandable to those outside of the area.

1 The structure–activity relationship (SAR) is the relationship between the chemical or 3D structure of a molecule and its biological activity. The analysis of SAR enables the determination of the chemical groups responsible for evoking a target biological effect in the organism. This allows modification of the effect or the potency of a bioactive compound (typically a drug) by changing its chemical structure. Medical chemists use the techniques of chemical synthesis to insert new chemical groups into the biomedical compound and test the modifications for their biological effects.

Exhibit 3. In vitro activity of bedaquiline against M. tuberculosis, other mycobacteri-

al species, and non-mycobacteria isolates.

OrganismMTB Resistance Subtype N

Bedaquiline MIC (µg/ml)

Range MIC50 MIC90 MIC95

M. tuberculosis All 109 ≤0.008 - 0.12 0.03 0.06 0.06

DS-TB 65 ≤0.008 - 0.12 0.03 0.06 0.06

MDR-TB 44 ≤0.008 - 0.12 0.03 0.06 0.06N = number of isolates; MIC, minimum inhibitory concentration

Table 1. In vitro activity of bedaquiline against M. tuberculosis preclinical isolates. (Source: Janssen Pharmaceutica background document for the FDA’s Advisory Committee meeting)

Mycobacterial species N

Bedaquiline MIC (µg/ml)

Range Median

M. bovis 1 NA 0.003

M. kansasii 1 NA 0.003

M. marinum 1 NA 0.003

M. smegmatis 7 0.003 – 0.010 0.007

M. avium/M. intracellare (MAC) 7 0.003 – 0.010 0.010

M. fortuitum 5 0.007 – 0.010 0.010

M. absessus 1 NA 0.250

M. ulcerans 1 NA 0.500N = number of isolates; MIC, minimum inhibitory concentration; NA, not applicable

Table 2. In vitro activity of bedaquiline against other mycobacterial species. (Source: Janssen Pharma-ceutica background document for the FDA’s Advisory Committee meeting)

Non-mycobacterial organisms N

Bedaquiline MIC (µg/ml)

Range Median

Corynebacterium jeikeium 1 NA 4

Corynebacterium urealyticum 1 NA 4

Helicobacter pylori 20 2 - 4 4

Nocardia asteroids 1 NA >16

Nocardia farcinica 1 NA >16

Escherichia coli 1 NA >32

Haemophilus influenzae 1 NA >32

Streptococcus pneumoniae 10 16 - 32 >32

Staphylococcus aureus 1 NA >32N = number of isolates; MIC, minimum inhibitory concentration; NA, not applicable

Table 3. In vitro activity of bedaquiline against non-mycobacterial isolates. (Source: Janssen Pharma-ceutica background document for the FDA’s Advisory Committee meeting)

On-campusAssignment

You and your group members are co-workers in Andries’ research team. You are

attempting to obtain convincing data that you can present at the meeting with

the three company review committees describe above and upper management

in the hopes that they will reconsider their decision to terminate development of

bedaquiline. Therefore, you will re-evaluate all of the data in the Investigator’s

Brochure and will come up with a solid plan for the restart of the development.

Convincing the internal review committees and upper management will not be an

easy task, given their clear position. Specifically, they concluded that the problems

associated with bedaquiline were serious and further development of the product

for which the company will likely not receive a suitable return on investment would

require strong arguments.

To doYour group will prepare a brief PowerPoint presentation (5 slides

maximum), which will be presented during the meeting with the internal

committees.