Study to Assess the Efficacy and Safety of LYS006 in Patients With Moderate to Severe Inflammatory Acne: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Study to Assess the Efficacy and Safety of LYS006 in Patients With Moderate to Severe Inflammatory Acne (NCT03497897), a randomized phase 2 study. The primary outcome is Total Inflammatory Lesion Count.

2. Clinical Question

The study’s stated purpose was to assess preliminary efficacy and safety and to determine whether LYS006 had an adequate clinical profile for further clinical development. That purpose is distinct from establishing a definitive treatment effect. The posted estimates describe the recorded Week 12 outcome under the registry’s analysis; they do not by themselves establish what would happen in other populations, at other time points, or under different treatment conditions.

3. Trial Design

The registry identifies NCT03497897 as a completed phase 2 study sponsored by Novartis Pharmaceuticals. Its official description specifies a randomized, subject- and investigator-blinded, placebo-controlled, multicenter study with parallel groups. The record’s masking field is double, and its allocation field is randomized. The study began on 2018-09-10, and primary completion was recorded as 2022-02-24.

The planned treatment groups were LYS006 20 mg twice daily, LYS006 2 mg twice daily, and matching placebo twice daily. Each regimen was oral and continued for 12 weeks. The registry records enrollment of 66. The outcome population and safety populations are described separately in the posted results: the primary outcome uses the PD analysis set with valid outcome assessment, while adverse-event counts are reported for participants at risk in each group. These populations should not be assumed to be identical.

The primary outcome was total inflammatory lesion count at Week 12. The registry describes a Bayesian mixed effect model for repeated measurements applied to natural log transformed inflammatory facial lesion counts up to Week 12. It reports model-estimated posterior geometric means for each group and a Bayesian comparison of LYS006 20 mg with placebo. The record does not report a posted analysis comparing LYS006 2 mg with placebo for this outcome.

Further details about randomization procedures, the allocation ratio, the blinding procedures in practice, and the sample-size justification are not reported in the registry record. The record also does not report the planned analysis population definitions in enough detail to reconstruct them. Consequently, the broad design is identifiable, but several operational and analytic details cannot be assessed from this record alone.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Before assignment: There was an initial screening period of up to 4 weeks.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Participants in the PD analysis set who had a valid assessment of the outcome measure.

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

6. Primary Endpoint

The registry lists the primary outcome as Total Inflammatory Lesion Count, measured over Week 12.

The primary endpoint is total inflammatory facial lesion count at Week 12. It includes papules, pustules, and nodules. Counting lesions turns a clinical feature of inflammatory acne into a quantitative outcome that can be compared across randomized groups. This kind of measure can describe the burden of inflammatory lesions at a specified assessment time, while the study’s repeated-measures model uses the recorded counts up to Week 12 to estimate the Week 12 result.

The analysis transformed lesion counts using the natural logarithm and reported posterior geometric means. Geometric means are a way to summarize values on a multiplicative scale after log transformation; they are not arithmetic averages of the raw counts. The registry reports estimated Week 12 geometric means of 18.55 for LYS006 20 mg, 24.51 for LYS006 2 mg, and 20.04 for placebo. The corresponding reported interval for each group is a 90% interval, described in the outcome text as a credible interval.

As a general statistical point, a prespecified endpoint and assessment time help define what a study is evaluating and limit ambiguity in interpreting results. The registry identifies the primary outcome and Week 12 timeframe, but does not report the clinical rationale for selecting this endpoint or explain how a particular difference in lesion counts would be judged clinically important. No such threshold should therefore be inferred from the posted estimates.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary analysis used a Bayesian mixed effect model for repeated measurements. It analyzed natural log transformed inflammatory facial lesion counts collected up to Week 12 and reported model estimates at Week 12. In general, a repeated-measures model uses observations across time rather than treating the final assessment as the only information. A mixed effect model can represent patterns shared across participants while allowing participant-level variation. The registry record does not report the model’s full specification, including its covariates, covariance structure, prior distributions, or handling of incomplete measurements.

Why transform lesion counts?

General statistical explanation: a natural log transformation expresses positive values on a scale where multiplicative differences are often easier to model. When results are returned to the original scale, the summary is commonly expressed as a geometric mean. A geometric mean summarizes the center of values through their multiplicative structure; it should not be mistaken for the arithmetic mean of the raw lesion counts. The record identifies natural log transformation and posterior geometric means, but does not explain why that transformation was selected.

What does the ratio represent?

The posted contrast is the posterior geometric mean ratio for LYS006 20 mg relative to placebo. A ratio of 1 corresponds to equal geometric means on this scale. A ratio below 1 corresponds to a lower geometric mean in the LYS006 20 mg group, while a ratio above 1 corresponds to a higher geometric mean. This is a general interpretation of the ratio’s direction, not a claim about the clinical importance of any particular value. The posted ratio is 0.93, with a 90% interval from 0.64 to 1.34.

How should the interval be read?

In general, an interval communicates uncertainty around an estimate. A Bayesian credible interval is interpreted using the posterior distribution and the model’s assumptions: it gives a range containing the stated share of posterior probability. A frequentist confidence interval has a different repeated-sampling interpretation. The registry’s outcome description calls the group intervals 90% credible intervals, while the comparison is recorded with a 90% interval. Because the comparison is Bayesian and reported as a posterior quantity, its interval should be read in that Bayesian context. The exact interval label in the analysis record is not more specifically explained.

The interval for the ratio includes 1. This means the posted interval does not exclude equality of the group geometric means, and also allows ratios on either side of equality. It does not mean there is no effect, nor does it show that all values within the interval are equally likely. General explanation: interval width reflects uncertainty, which can depend on the amount and variability of data and on model assumptions. The registry result should therefore be read as an estimate with uncertainty rather than as a single settled value.

What do the posterior probabilities add?

The reported probability that the geometric mean ratio is less than 1 is 0.637. The reported probability that it is less than 0.75 is 0.171. These posterior probabilities state how much posterior probability lies below the respective thresholds, conditional on the data and Bayesian model. They are not p-values, and they are not probabilities that a treatment works in a general population. The threshold of 0.75 is a statistical reference in the posted analysis; the record does not report a justification for treating it as a clinically meaningful threshold.

The probability of being below 1 and the probability of being below 0.75 address different levels of reduction on the ratio scale. Neither replaces the interval or the full model context. Bayesian results depend on both the observed data and the prior distributions. The prior distributions are not reported in the registry record, so the contribution of prior assumptions cannot be assessed from the posted information.

Why can over-dispersion matter for counts?

General statistical explanation: count outcomes can vary more than a simple model assumes. Over-dispersion means the observed variability exceeds what a particular count model expects. If it is not accommodated, uncertainty estimates can be too small and comparisons may appear more precise than they are. This analysis is described as a model of log transformed counts using a mixed effect repeated-measures approach; the registry does not report an over-dispersion assessment or a separate count-model specification. It is therefore not possible to determine from the record how this issue was evaluated.

The reported geometric means are estimated from the model, not raw summaries identified as simple averages. The primary analysis population consists of participants in the PD analysis set with a valid assessment. The registry does not provide enough information to independently reproduce the model estimates, assess the consequences of missing outcome measurements, or determine whether alternative modeling choices would materially change the result. General statistical interpretation should remain conditional on those limits.

10. Analyses Reported by the Sponsor

The registry reports the sponsor’s Bayesian analysis for the comparison of LYS006 20 mg with placebo: a posterior geometric mean ratio of 0.93, with a 90% interval from 0.64 to 1.34. It also reports posterior probabilities of 0.637 for a ratio less than 1 and 0.171 for a ratio less than 0.75. These are the posted inferential results and are the appropriate basis for describing the registry’s comparison.

No discrepancy between the sponsor analysis and a recalculation is flagged in the record. A separate recalculated estimate is not reported for this outcome comparison, and the registry does not post a Bayesian comparison of LYS006 2 mg with placebo. The group-level geometric means can be described as reported, but they do not generate a substitute inferential comparison. The sponsor’s analysis should therefore be read as the available comparison rather than as one result among competing recalculated estimates.

The record does not report the prior distributions or enough model detail to reproduce the analysis independently. That limits external statistical verification, but does not alter what the registry states as its result. The posted estimate and interval remain conditional on the sponsor’s model and analysis population.

11. Multiplicity, Missing Data and Analysis Populations

The registry posts one primary outcome, total inflammatory lesion count at Week 12, and three analyses: a Bayesian comparison of LYS006 20 mg with placebo and two posterior probability statements for that same comparison. No secondary outcomes are posted in the record. This does not establish that no other analyses were conducted; it describes what is reported in the registry.

Multiplicity arises when several outcomes, time points, doses, or comparisons are tested or interpreted together. In general, examining many results can increase the chance of finding an apparently notable result by chance, depending on the analysis framework and decision rules. The registry does not report a multiplicity adjustment, a hierarchy of hypotheses, or a prespecified strategy for interpreting the reported posterior probabilities. It is not possible to determine from the record whether multiplicity was addressed elsewhere.

The primary outcome population is described as participants in the PD analysis set who had a valid assessment. The registry does not report a complete account of exclusions, missing assessments, or the method used to handle missing outcome data. The repeated-measures model is described, but the assumptions used for incomplete observations and any sensitivity analyses are not reported in the registry record. These omissions limit evaluation of how robust the Week 12 estimate is to missingness.

The safety results use their own at-risk denominators: 26 for LYS006 20 mg, 11 for LYS006 2 mg, and 29 for placebo. These figures differ from the outcome-analysis counts reported for the groups, so the endpoint and safety summaries should not be treated as if they necessarily use the same analysis population. The registry does not provide further detail on the relationship between these populations.

The sample-size assumptions describe an illustration based on an observed effect, with two-sided significance level 0.05, equal allocation between two arms, and a normal approximation. Such planning illustrations are hypothetical design calculations, not observed power and not evidence about the probability that this completed study detected an effect. The registry record does not report the trial’s own sample-size justification or enough planning detail to assess its original design rationale.

12. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 2% reporting threshold.

Most frequent non-serious adverse event terms

The adverse-event record reports no serious adverse events in any of the three treatment groups and no deaths. The number at risk for serious adverse events was 26 for LYS006 20 mg, 11 for LYS006 2 mg, and 29 for placebo. The registry also reports no serious adverse-event terms. These are the posted counts; they should be read in relation to the reported at-risk groups and the study’s recorded follow-up.

General statistical explanation: observing no events in a study does not prove that the underlying risk is zero. The precision of a safety estimate depends on the amount of information available, including the number of participants and duration of observation, as well as how events are collected and classified. The registry record does not report enough detail to estimate a precise underlying risk or assess rare events.

For other adverse events, the record reports 10 participants affected in the LYS006 20 mg group, 8 in the LYS006 2 mg group, and 16 in the placebo group. These are counts, not evidence on their own that one regimen is safer or less safe. The record does not provide adverse-event terms in this summary, severity, timing, or a comparative safety analysis. The absence of reported serious events should not be expanded into a claim that there were no safety concerns of any kind.

13. What the Result Does, and Does Not, Mean

The results establish what the registry reports for this trial: model-estimated Week 12 inflammatory lesion counts in three groups and a Bayesian comparison of LYS006 20 mg with placebo. The estimated geometric means were 18.55 for LYS006 20 mg, 24.51 for LYS006 2 mg, and 20.04 for placebo. The reported comparison for LYS006 20 mg was a posterior geometric mean ratio of 0.93, with a 90% interval from 0.64 to 1.34. The posterior probability that the ratio was less than 1 was 0.637; the probability that it was less than 0.75 was 0.171.

These estimates do not establish a definitive treatment effect. The ratio interval includes 1, and the registry does not report a conventional p-value. The point estimates are not proof that one group’s outcome is better, and the posterior probabilities are conditional on the Bayesian model. No posted comparison with placebo is available for LYS006 2 mg, so its group estimate cannot be interpreted as a formal treatment contrast.

The findings also do not establish effects beyond the measured Week 12 endpoint, in populations outside the analyzed group, or under treatment schedules not described in the study. The record does not report a threshold for clinical importance, so statistical estimates cannot be translated into a judgment about how meaningful a difference would be to patients. Nor do the safety counts establish that serious or other adverse events cannot occur.

In general, a randomized comparison can support a causal interpretation when its design, conduct, and analysis are sound. The registry identifies randomization and double masking, but does not provide all details needed to evaluate implementation, missing data, or model assumptions. The most defensible reading is therefore limited to the posted estimates and their uncertainty, interpreted in the context of those reporting gaps.

14. Why This Trial Matters Statistically

This trial is statistically informative because it reports an outcome analyzed on a transformed scale and a Bayesian contrast, rather than only listing raw group outcomes. The distinction matters: the estimated geometric means and the posterior ratio express different summaries, and the ratio directly addresses the posted LYS006 20 mg versus placebo comparison. The uncertainty interval and posterior probabilities make uncertainty visible rather than leaving interpretation to the point estimate alone.

The result also illustrates why the numerical direction of a point estimate is not enough. The posterior ratio is 0.93, but its 90% interval extends from 0.64 to 1.34. The interval includes equality and values on either side, so the estimate does not sharply distinguish the groups under the reported analysis. The probability of a ratio below 1 does not answer the same question as the probability of a ratio below 0.75; presenting both helps distinguish any reduction from a more substantial ratio threshold, though the registry does not explain the clinical rationale for that threshold.

The endpoint itself is a count, but the analysis transforms it and reports geometric means from a repeated-measures model. Understanding the scale is necessary to avoid treating the estimates as arithmetic means or interpreting the ratio as a difference in lesion counts. The registry’s limited reporting of model details also demonstrates why a result’s method description matters: full evaluation depends on assumptions, prior distributions, missing-data handling, and population definitions that are not all available in this record.

Finally, the record presents results for two active doses but posts an inferential comparison only for the 20 mg dose. Keeping group summaries separate from formal contrasts avoids overstating what was analyzed. This is a useful example of how to distinguish descriptive estimates, model-based comparisons, and uncertainty in a clinical trial report.

15. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The principal limitation for interpretation is the limited detail available about the Bayesian model. The registry describes natural log transformation, a mixed effect model for repeated measurements, and Week 12 model estimates, but the prior distributions, covariates, covariance structure, and other implementation details are not reported in the registry record. Independent reproduction or sensitivity assessment is therefore not possible from the posted description alone.

The outcome is analyzed in the PD analysis set among participants with a valid assessment. The record does not define that set in detail or explain how participants without valid assessments were handled. It also does not report the pattern or amount of missing outcome data or sensitivity analyses for missingness. The results may therefore be difficult to compare with analyses using a different population or missing-data approach.

The posted inferential comparison covers LYS006 20 mg versus placebo only. Although a group-level estimate is reported for LYS006 2 mg, a corresponding Bayesian comparison with placebo is not reported in the registry record. The group summaries do not supply the uncertainty for that contrast. The record also does not report a multiplicity strategy, clinical importance threshold, or the rationale for the posterior probability threshold of 0.75.

Safety reporting is limited in the record excerpt: serious adverse events and deaths are reported as zero, and other adverse events are given as counts without event terms or detailed comparative analysis. The duration and completeness of safety follow-up are not specified beyond the 12-week treatment regimen. These data do not resolve the possibility of uncommon events.

Finally, enrollment is recorded as 66, but the registry does not report the trial’s sample-size justification. The study is phase 2 and was described as assessing preliminary efficacy and safety. Its results should be interpreted as preliminary evidence from this design and analysis, not as a definitive answer about treatment effects or safety.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Study to Assess the Efficacy and Safety of LYS006 in Patients With Moderate to Severe Inflammatory Acne — NCT03497897.
• Registry record last updated 2023-08-31; status last verified 2023-08.
• Results first posted 2023-03-20.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

19. Record Summary

NCT03497897 is a completed phase 2, randomized, double-blind, parallel-group, placebo-controlled study sponsored by Novartis Pharmaceuticals. It enrolled 66 participants with moderate to severe inflammatory acne. Participants received oral LYS006 20 mg twice daily, LYS006 2 mg twice daily, or matching placebo twice daily for 12 weeks.

The primary outcome was total inflammatory facial lesion count at Week 12, including papules, pustules, and nodules. A Bayesian mixed effect model for repeated measurements analyzed natural log transformed counts and estimated posterior geometric means. The posted comparison of LYS006 20 mg with placebo gave a posterior geometric mean ratio of 0.93, with a 90% interval from 0.64 to 1.34. A comparison for LYS006 2 mg is not reported in the registry record.

The safety record reports no serious adverse events or deaths. Model detail, missing-data handling, multiplicity procedures, and the trial’s sample-size justification are not reported in the registry record.