Balancing access to BPaLM regimens and risk of resistance

In May, 2022, WHO released a rapid communication
stating that the 6-month all-oral regimen of bedaquiline,
pretomanid, linezolid, and moxifloxacin (BPaLM) “may
be used programmatically for patients (aged ≥15 years)
with rifampicin-resistant tuberculosis not yet exposed
to bedaquiline, pretomanid and linezolid [BPaL]”. The
rapid communication continued that in patients with
fluoroquinolone resistance, a 6-month regimen of BPaL
can be used. WHO concluded that “drug-susceptibility
testing (DST) should not delay treatment initiation and
DST to fluoroquinolones is strongly encouraged but not
required”.1 Although near universal eligibility for short,
all-oral, rifampicin-resistant tuberculosis treatment is
excellent news, data presented at the European Society
for Mycobacteriology conference in Bologna, Italy (June,
2022), suggest the long-term effectiveness of these
regimens might already be at risk.
First, the efficacy of a drug can vary by Mycobacterium
tuberculosis lineage. At the conference, Claudio Köser
presented the results of a study showing an increased
pretomanid minimum inhibitory concentration (MIC)
distribution for wild-type M tuberculosis lineage 1 strains
compared with strains of other lineages.2 Praharshinie
Rupasinghe presented a poster with data from a
multicountry study (Rupasinghe P, unpublished) suppor-
ting this finding. Emmanuel Rivière raised the hypothesis
of a lineage-dependent MIC distribution for bedaquiline,
with an increased proportion of lineage 4 strains with
MICs closer to—albeit below—the epidemiological
cutoff.3 It is unknown if lineage-dependent differences in
MIC distribution are clinically relevant.
Second, different drugs penetrate lung lesions at
different rates. Véronique Dartois showed experimental
data that moxifloxacin, linezolid, and pretomanid rapidly
penetrate the caseum, whereas bedaquiline—even when
using a loading dose—only fully reaches mycobacteria
in the caseum’s centre after 2–5 weeks of treatment.4
The experimental data presented by Sonnenkalb and
colleagues show that bedaquiline exposure at one eighth
of a strain’s MIC results in acquisition of variants in
resistance genes within 20 days of exposure, suggesting
that the slowly increasing bedaquiline concentration in
the caseum of lung lesions could select for resistance. 5
Furthermore, in contrast to moxifloxacin, linezolid,
and pretomanid, bedaquiline has a long half-life,
resulting in long-lasting presence in the caseum and
potential exposure to a single drug in case of treatment
interruption. Third, primary resistance to BPaLM drugs
already exists. Primary resistance was first documented
for bedaquiline in 2016. 6,7 Diana Machado presented the
first case report of primary pretomanid resistance due to
a ddn88* variant in a patient diagnosed in 2008, years
before pretomanid and delamanid introduction.8
Taken together, these issues raise questions regarding
the robustness of the BPaL regimen for patients now
and in the future. Specifically, administration of BPaL
to patients with fluoroquinolone-resistant tuberculosis
might increase the risk of acquisition of resistance to
bedaquiline and linezolid. This risk could be especially
elevated during the first weeks of treatment when the
mycobacterial burden is high and bedaquiline is not
yet fully distributed into lung lesions. Mycobacteria in
the caseum might be temporarily exposed to linezolid
monotherapy if elevated MICs reduce pretomanid’s
effectiveness. If compensatory mutations are acquired,
resistant strains might then transmit at rates similar
to pan-susceptible M tuberculosis, as shown by
Sébastien Gagneux for rifampicin-resistant lineage 2
strains carrying a rpoB S450L and a compensatory rpoC
mutation. This expands on the previously reported
association between compensatory mutations and
transmission.9
These data highlight the need to balance the risks of
emergence of resistance to BPaLM drugs with the risks
to current patients from not gaining access to BPaLM
or BPaL regimens. The potential risk of empirical (ie,
12PraharshinieRupasingheEmmanuel Rivière3Véronique Dartois456,7Diana Machado8Sébastien Gagneux9
Comment1412 www.thelancet.com/infection Vol 22 October 2022
blinded to DST) use of BPaLM and BPaL regimens could be
especially high for rifampicin-resistant tuberculosis caused
by lineage 1 strains, estimated to cause 28% of global
tuberculosis, mainly in Africa and Asia.
Although there are no easy solutions for this complex
issue, some things can be done now. The COVID-19
pandemic has boosted the global next-generation
sequencing (NGS) infrastructure, which should be
harnessed to predict resistance, guide rescue treatment
in case of resistance or toxicity, and reduce knowledge
gaps on resistance-associated mutations. Patients for
whom treatment does not appear to be effective should
have samples assessed by DST at reference centres. As
we move forwards, rapid diagnostic tests, including
targeted NGS, will be important. Data presented by Philip
Supply and Viola Dreyer suggest that targeted NGS can
detect resistance directly from smear-positive sputum 10
and maybe from stool samples. 11 These diagnostic
approaches need validation and WHO endorsement to
facilitate governmental investment and roll-out. We are
fortunate to have more new drugs in the pipeline, but as
a community we should demand that every new drug is
developed alongside a companion diagnostic that that will
ensure the longevity of new regimens well into the future.