1. One-line summary

• Population: MB Hansen's disease at one Brazilian referral centre, 2015-2025.
• Intervention: RIMOXCLAMIN (see the dosing in the Methods).
• Comparator: WHO-MDT for 12 months.
• Outcomes: nerve thickening and pain, symptoms, Semmes-Weinstein monofilament (SWM) altered points, physical disability grade (PDG), adverse events. There was no relapse or bacteriological endpoint.

2. Strengths

• Important question. Dapsone toxicity (haemolysis, hypersensitivity) and clofazimine pigmentation are real problems. Almost no new regimen has been tested in leprosy for decades. Neural outcomes are what patients care about.
• Large for leprosy. 162 patients is substantial for a neglected disease, and it extends the group's earlier pilot.
• Single expert reference centre. Consecutive enrolment and a uniform protocol reduce between-site variability.
• Good neurological phenotyping.
  • Standardised SWM mapping (7 points per hand, 11 per foot), MRC strength, a numerical pain scale, WHO PDG, Doppler nerve ultrasound and electroneuromyography (ENMG).
  • Visits at baseline and at months 3, 6 and 12.
  • Outcomes are patient-relevant and not only laboratory markers.
• Sensible baseline checks. The authors tested whether lesion count tracks neural burden (Table 2) and ran a logistic model for treatment allocation.
• Biomarkers included. PGL-I, RLEP-qPCR and slit-skin smear, plus the finding that ultrasound and ENMG pick up far more neural disease (about 96% and 85%) than laboratory tests (about 24-28%).
• Reasonably candid limitations. They admit non-randomised allocation, no post-treatment follow-up, no relapse data and a single centre, and they ask for controlled trials.
• Plausible biology. Fluoroquinolone, macrolide and tetracycline combinations are bactericidal against M. leprae, and minocycline and macrolides have anti-inflammatory effects. Open access.

3. Drawbacks

A. Design and confounding (the core problem)

1. Non-random, availability-based allocation. Everything else rests on this, and the authors concede it.
2. Drugs were self-funded or donated. The Discussion says RIMOXCLAMIN drugs were "obtained either privately or through donations from the patients themselves due to their cost". Treatment arm is therefore tied to ability to pay.
   • The RIMOXCLAMIN arm had more healthcare professionals (19.4% vs 10.2%), more "high-income intellectual professionals" (21.4% vs 10.2%) and fewer homemakers (2.9% vs 13.6%, P = .01).
   • Income, education, health literacy, adherence, nutrition, access to rehabilitation and early steroid treatment of neuritis all plausibly improve neural recovery independently of the antibiotics.
   • The authors call occupation the "only socioeconomic factor analysed" and say there was no significant socioeconomic difference, yet P = .01 for homemakers. The two statements conflict.
3. The groups differ at baseline in clinically relevant ways.
   • The WHO-MDT arm had more nerve thickening (94.9% vs 80.6%; the authors report this as non-significant, P = .095).
   • It also had more cramps (91.5% vs 63.1%, P < .0001), more tingling (94.9% vs 80.6%, P = .011) and more pain ≥3 (96.6% vs 85.4%).
   • Hand sensation was worse (77.7% vs 93.2% normal).
   • It had more altered SWM points (the authors acknowledge this).
   • It had many more "contact" entries (20 vs 5 in Table 1, as I read it).
   • The RIMOXCLAMIN arm had more new diagnoses but also more insufficient-treatment and recurrence cases, and a longer diagnostic delay (36 vs 24 months).
   • These imbalances run in both directions. That makes the direction of bias unpredictable, and a regression model with R² 0.048 cannot "adjust" for them.
4. Time and era effects. The study spans 10 years, with many patients treated in 2022-2024. It is not stated whether the arms were concurrent. Diagnostic and rehabilitation practice may have changed over that time.
5. Co-interventions are not reported in the main text. Corticosteroids for neuritis and reactions are the main driver of nerve function recovery in leprosy. Neuritis occurred in 26.2% of the RIMOXCLAMIN arm and 35.6% of the WHO-MDT arm (P = .21). Steroid dose, timing and duration, and physiotherapy and self-care, are not in the main text.
6. Unblinded assessment of subjective outcomes. Palpated nerve thickening, SWM, pain scores, MRC and PDG were scored by clinicians who knew the regimen. The group that designed and promotes RIMOXCLAMIN also ran the study. No conflict of interest statement is visible in the extracted text.
7. Possible overlap with the pilot. Does this "confirmation" reuse patients from the earlier pilot (PMID 40373668)? If so, it is not independent confirmation.

B. Statistical analysis

1. "Significant in one arm, not in the other" is not a between-arm comparison. The paper emphasises within-arm trajectories. For example, PDG-2 fell significantly with RIMOXCLAMIN (P < .0001) but not with WHO-MDT (P = .49). The correct test is the group × time interaction, or a direct between-arm comparison of change adjusted for baseline.
2. No adjusted between-arm effect estimates. There are no effect sizes with confidence intervals, no mixed-effects or GEE models, no ANCOVA and no propensity weighting. The authors themselves say "future analyses should incorporate logistic regression models and Kaplan-Meier curves", but those are the analyses this paper needed.
3. Floor and ceiling effects and regression to the mean.
   • The WHO-MDT arm started with more disease, so it had more room to improve.
   • It also started with fewer PDG-2 cases (10.2% vs 23.3%).
   • The smaller baseline PDG-2 count limits how much that arm could show.
4. Multiplicity. There are dozens of outcomes, time points and tests, with no pre-specified primary endpoint, no correction and no sample size calculation.
5. Unit of analysis. Table 8 appears to analyse SWM points as the unit, not patients. Points within the same patient are correlated, which inflates significance unless clustering is handled.
6. Missing data and attrition are not clearly reported. There is no flow diagram in the main text, no handling of dropouts, and no statement of intention-to-treat vs per-protocol. 33.9% of the WHO-MDT arm switched from dapsone. The paper does not say whether they were analysed as WHO-MDT or as something else.
7. The regression is misread. The text says sex, age and duration were "not significantly associated" with treatment, but the sex odds ratio was 2.11 (P = .028) and the overall model χ² gave P = .017. The model has low explanatory power, which means it is poorly fitted. It does not mean the groups are balanced.

C. Outcome definitions and measurement

1. Nerve thickening and nerve pain on palpation are subjective and prone to observer bias. Reduction in palpable thickening is not an established marker of bacteriological or neurological cure.
2. PDG-2 "reversal" is biologically surprising. PDG-2 means visible deformity, such as ulcers, clawing and lagophthalmos. A fall from about 24 to 4 patients within 12 months is striking. Part of it may be ulcer healing and different grading rather than true reversal of fixed deformity. Inter-rater reliability of PDG is not reported.
3. "Cure" and "bactericidal superiority" were not measured. There were no relapse data, no bacillary index trend, no repeat PGL-I or RNA-based viability PCR, and no resistance testing. The Discussion's claim that complete neurological recovery is "unattainable after just 6 months" has no 6-month comparator.
4. Anti-inflammatory vs antimicrobial effect cannot be separated. Minocycline, macrolides and fluoroquinolones modulate inflammation. Faster nerve symptom improvement may reflect that rather than faster killing of bacilli.

D. Population and generalisability

1. Not the typical MB patient.
   • 81.5% of RIMOXCLAMIN patients were borderline and 17.5% pure neural, with 0-1 borderline-lepromatous and no lepromatous cases.
   • Only about 18% were RLEP-qPCR positive, and bacilloscopy was done in only 35 RIMOXCLAMIN patients.
   • These are low bacillary load patients. MDT must perform in high-load lepromatous disease, where relapse and resistance risk are highest, and the study cannot speak to that.
2. Women and older patients. About 65% of the RIMOXCLAMIN arm were women and the median age was 55-59. This is a tertiary referral population with unusually long diagnostic delay (24-36 months) and many with neurological symptoms. It is not typical of Brazilian or global leprosy (poorer, younger, more often male).
3. Circularity in the lesion-count conclusion. The claim that lesion count "fails to reflect neural involvement" comes from a cohort with MB status defined partly by neurological abnormalities, with near-universal neurological symptoms. That is selection, not a test of the operational classification. It is also a cross-sectional baseline analysis in a study designed for treatment comparison.
4. Public health and feasibility are not addressed.
   • WHO-MDT is free, blister-packed and monthly supervised. RIMOXCLAMIN is a four-drug regimen with daily clarithromycin and minocycline, and no cost or adherence data.
   • Mass moxifloxacin use in TB-endemic countries raises concerns about fluoroquinolone resistance in M. tuberculosis.

E. Safety claims

1. "Safe" is based on 103 patients with open-label, unblinded adverse event reporting. Severity grading is not clearly standardised (CTCAE or similar is not named in the main text).
2. The adverse events listed are not trivial.
   • RIMOXCLAMIN: nausea and vomiting 16.5%, bitter taste 14.6%, diarrhoea 12.6%, hyperpigmentation 13.6%, balance change 7.8%, vertigo 3.9%, one rhythm change.
   • The authors acknowledge QT prolongation risk from moxifloxacin plus clarithromycin. The main text does not report ECG monitoring.
3. Fluoroquinolone class risks matter in leprosy. Peripheral neuropathy and tendon toxicity can confound neurological assessment in a neuropathic disease.
4. Drug interaction. Rifampicin induces CYP3A4 and lowers clarithromycin exposure. Is that addressed?
5. Dapsone anaemia in 33.9% looks high. Was G6PD status tested? What were baseline haemoglobin and dapsone dose adjustments? Anaemia at this rate may reflect local practice and age (median about 59), which could inflate the apparent safety gap.
6. Rare events cannot be assessed. Rare serious adverse events (hepatotoxicity, C. difficile, arrhythmia) need much larger samples.

F. Internal inconsistencies to point out

• Abstract vs results on month 3. The abstract says "At month 3, neurological symptoms were greater in the RIMOXCLAMIN group than in the WHO-MDT", but the results show larger reductions in the RIMOXCLAMIN group. This is likely a wording error.
• "Newly diagnosed" vs Table 1. The abstract and Methods describe 162 new cases, but Table 1 includes contacts, insufficient prior treatment and recurrences.
• Laboratory positivity. The text says 28.4% of the RIMOXCLAMIN arm had at least one positive laboratory test. Table 7 shows 23.5% (24/102).
• Typos in Table 5 and the text. For example "P = .009%" and "P = .01%".
• PubMed labels it a "Clinical Trial". The paper itself says "observational cohort".
• The conclusion overreaches. "Solidifies RIMOXCLAMIN's position as safe, effective" and "potential as a future standard of care" sit next to the authors' own call for "controlled, head-to-head studies" to "definitively assess" efficacy.

4. How to improve the study

• A pragmatic randomised trial, multicentre, with blinded outcome assessors (nerve ultrasound and ENMG readers, SWM and MRC examiners). A non-inferiority or superiority design with a pre-registered primary endpoint, for example the proportion with improved PDG or nerve function at 12 months.
• If randomisation is impossible, use concurrent controls, a clear and equal public-funded drug supply, and prospective protocol registration.
• Stratify or enrich by bacillary index (include BL/LL), and report PB, PNL and MB separately.

• Mixed-effects models with group × time interaction, adjusted for baseline neural status, age, sex, SES, diagnostic delay, case category and steroid exposure.
• Propensity weighting or matching, E-values for unmeasured confounding, and sensitivity analyses excluding contacts and recurrences.
• Intention-to-treat and per-protocol analyses, with explicit handling of the dapsone switchers.
• Multiplicity control, effect sizes with confidence intervals, and a sample size calculation.
• Patient-level analysis of SWM, or models that handle clustering of points.
• Follow the STROBE (or TREND) checklist and include a CONSORT-style flow diagram.

• Report steroid dose and duration, neuritis and reaction management, and rehabilitation or self-care.
• Use standardised graded adverse event criteria, and report ECG/QTc and liver tests.
• Measure adherence (pill counts, electronic monitoring) and report G6PD status and haemoglobin.
• Add quality-of-life and participation scales (for example SALSA, stigma scales).
• Add molecular endpoints: repeat PGL-I, RNA-based viability PCR, and resistance mutations (rpoB, gyrA, 23S rRNA).
• Follow patients at least 5 years for relapse, permanent nerve function impairment and late reactions.
• Report the cost per patient.
• Consider comparators beyond WHO-MDT, such as rifampicin-ofloxacin-minocycline (ROM).

5. Questions for the discussion

1. If the drugs were bought by patients, is this a comparison of regimens or of socioeconomic groups?
2. Can within-arm P values support the claim of superiority? What would an interaction test show?
3. Does PDG-2 really "reverse" within 12 months, and how reliable is PDG grading across examiners?
4. Is nerve palpation a valid primary outcome when the examiner knows the regimen?
5. Are anti-inflammatory effects of minocycline and macrolides being credited as antimicrobial superiority?
6. Would you use this in a lepromatous patient with a high bacillary index, based on this evidence? What about in a TB-endemic setting?
7. How much of the WHO-MDT adverse event burden comes from dapsone dosing and G6PD screening practice?
8. Does cross-sectional data in a neurologically selected cohort justify abandoning lesion-count classification?
9. Is the "first-line" and "standard of care" language consistent with the authors' own request for controlled trials?
10. What are the public health costs: resistance, cost, supervision, pill burden?

6. Overall verdict

1. The authors' own earlier work (the most important one to know)

• Pilot study. Frade et al., Braz J Infect Dis 2025 (PMID 40373668, full text in PMC12141834).
  • It included 66 new MB patients (46 on RIMOXCLAMIN and 20 on WHO-MDT) treated between 2015 and 2023.
  • It used the same design, centre, and outcomes (SWM, PDG, nerve palpation, pain).
  • Its conclusion was also that RIMOXCLAMIN gave faster neurological recovery.
• Overlap. The date ranges overlap, so the 162 patients in the paper you are presenting very likely include these 66. This is an expansion of one dataset, not an independent confirmation. Ask the authors in your discussion.
• A point to raise. In the pilot abstract, nerve thickening in the RIMOXCLAMIN arm fell from 65% at baseline to 9% at 12 months. In your paper it falls from 80.6% to 21.4%. The larger cohort looks less dramatic, which is consistent with a pilot overestimating the effect. This is my inference from the abstracts, so check the tables before you say it on a slide.

2. Evidence that the component drugs kill M. leprae

• Moxifloxacin. Pardillo et al., AAC 2008 found strong bactericidal activity in human leprosy.
• Clarithromycin and minocycline. Ji et al. showed bactericidal activity in lepromatous leprosy (JID 1993). A later randomised study tested single-dose clarithromycin plus minocycline, with or without ofloxacin (AAC 1996).
• Rifapentine-moxifloxacin-minocycline (PMM). Ji and Grosset, Lepr Rev 2000 is mouse-model work.
• Mouse post-exposure prophylaxis. Lenz et al., PLoS NTD 2020 tested rifampin, rifapentine, moxifloxacin, minocycline and clarithromycin.

3. Similar alternative regimens (ROM and relatives)

• ROM (rifampicin, ofloxacin, minocycline) meta-analysis. Setia et al., 2011 is a systematic review. The LSHTM commentary on it reports that single-dose ROM was less effective than WHO-MDT in paucibacillary disease (see the Lockwood and Cunha item below).
• Comparative review. Lazo-Porras et al., BMC Infect Dis 2020 is a systematic review and meta-analysis of WHO regimens against other regimens.
• Philippines RCT. It compared 24 monthly doses of ROM with two years of MDT in MB patients (Villahermosa et al., cited in the review linked in the next item). One review reports comparable outcomes. All WHO-MDT patients developed clofazimine pigmentation.
• Lockwood and Cunha, Lepr Rev 2012. The commentary argued for testing 12-month ROM regimens. It also suggested replacing ofloxacin with moxifloxacin. This is the conceptual ancestor of RIMOXCLAMIN.
• Hospital for Tropical Diseases, London. Sivakumaran et al., PLoS NTD 2024 is a retrospective cohort of monthly ROM.
• US case series. Franco-Paredes et al., 2022 used monthly rifampin, moxifloxacin and minocycline (RMM).

4. Trials that are the proper comparison for the claim

• RMC randomised trial (India). The trial protocol describes a multicentre, randomised non-inferiority trial. It compares WHO MB-MDT with monthly rifampicin, moxifloxacin and clarithromycin for 12 months. This is the design RIMOXCLAMIN needs, though it tests a different (monthly) regimen. Check whether results have been published.
• U-MDT/CT-BR (Brazil). Penna et al., PLoS NTD 2017 is an open-label RCT in 613 new MB patients, comparing 6 months with 12 months of WHO-MDT. It found no significant difference in reactions or disability progression. It is a useful benchmark for how a Brazilian MB trial is done, and it makes you ask whether the 12-month WHO-MDT arm in your paper is itself the right comparator.

5. Neurological outcomes after MDT (context for the "recovery" claim)

• Marques, Curr Opin Neurol 2024 is a review of leprous neuropathy by a co-author of your paper.
• Penna et al., Mem Inst Oswaldo Cruz 2023 found progressive neuropathy in lepromatous patients after MDT.
• Somasundaram et al., Trans R Soc Trop Med Hyg 2025 followed subclinical neuropathy in newly diagnosed patients started on MDT.

6. How to use this in the journal club

• Pitch. The mechanism is supported (sections 2 and 3). The clinical neural advantage over WHO-MDT comes from one group's own non-randomised data (sections 1 and the paper itself).
• The gap. Randomised comparisons of alternative regimens exist (ROM, RMC, U-MDT), but none yet exists for RIMOXCLAMIN. Until then the claim is hypothesis-generating.
• Slide idea. A timeline running from the mouse and early human bactericidal work, through the ROM trials and reviews, to the 2025 pilot, this paper, and the registered RMC trial.
• Caveat. I did not read the full texts of most of these papers. Verify any detail you plan to quote before you present it.

Comparison table

What the table shows

1. Evidence grade. ROM is the best-studied (a meta-analysis and some randomised data). The RMC trial will be the first randomised test of a moxifloxacin-containing regimen. RIMOXCLAMIN rests only on its authors' non-randomised cohorts, so it has the weakest design of the three.
2. Different endpoints. RIMOXCLAMIN papers measure neurological function. The RMC trial measures bacterial viability. Neither is yet shown to predict the other.
3. Regimen complexity. RIMOXCLAMIN is the only one with daily drugs after month 2, so its adherence and interaction profile differs from the monthly regimens.
4. No common outcome. No study compares ROM, RMC and RIMOXCLAMIN directly, and none measures the same outcome. Indirect comparison would not be valid.

Caveats

• The RMC row comes from the trial protocol, not results, so the efficacy of RMC is unknown.
• I took ROM randomised-trial details second-hand from reviews. I did not retrieve the original Philippines trial.
• The "daily drugs" and "all monthly" entries come from the protocol and abstract text. Check the full texts if you plan to state them as fact.