Pentoxifylline/Nonalcoholic Steatohepatitis (NASH) Study: The Effect of Pentoxifylline on NASH: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Pentoxifylline/Nonalcoholic Steatohepatitis (NASH) Study: The Effect of Pentoxifylline on NASH (NCT00267670), a randomized phase 2, phase 3 study. The primary outcome is The Number of Participants With a 30% Reduction in Alanine Aminotransferase (ALT) Treated With Pentoxifylline (PTX) or Placebo for 12 Months..

2. Clinical Question

The endpoint is framed as a count of participants meeting a threshold, rather than as the average change in ALT. The registry’s description says the primary goal was to determine whether pentoxifylline therapy improved serum ALT by a 30% change from baseline to month 12 compared with placebo. The posted endpoint title specifies a 30% reduction.

The clinical question is therefore about a threshold response over the stated period. It is not, on its own, a question about changes in liver histology, long-term clinical outcomes, or effects on every participant’s ALT value. Those broader outcomes are not reported as the primary endpoint in the registry record.

3. Trial Design

The study is recorded as randomized, double-masked, and parallel-group, with pentoxifylline and placebo as the interventions. It is listed as Phase 2 and Phase 3. Northwestern University is the sponsor. The registry record gives an enrollment of 26 and a status of completed.

The recorded study start is 2005-03, and primary completion is listed as 2009-09. The primary endpoint timeframe is baseline and 12 months. The pentoxifylline arm is described as 400mg PO TID, while the placebo arm is described as 1 pill PO TID. Additional detail about how masking was maintained, the randomization procedure, or the allocation ratio is not reported in the registry record.

The primary analysis is described as intention to treat. The record also says a per-protocol secondary analysis was performed and that there were no differences between the two analyses; intention-to-treat results are reported. Details of the per-protocol definition and the specific comparison used to assess differences between analyses are not reported in the registry record.

The record lists three secondary outcomes: change in tumor necrosis factor (TNF)-α levels, change in serum leptin levels, and change in serum adiponectin levels. Leptin and adiponectin are described as changes over 12 months; a timeframe for the TNF-α outcome is not reported in the registry record. The eligibility criteria, participant characteristics, and details of follow-up beyond the stated endpoint timeframe are also not reported.

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: Subjects were excluded from the study if they had evidence of another form of liver disease or if they were HIV positive, pregnant or had evidence of ongoing alcohol consumption exceeding 20g(males) and 10g(females)daily. Furthermore, subjects were excluded if they were taking drugs known to cause steatohepatitis.

Overall Study

Analysis population for the primary outcome, as defined in the registry: The primary analysis was done as intention to treat. A secondary analysis was performed per protocol and there were no differences between the two analyses. Intention to treat results are reported.

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 The Number of Participants With a 30% Reduction in Alanine Aminotransferase (ALT) Treated With Pentoxifylline (PTX) or Placebo for 12 Months., measured over baseline and 12 months.

The primary endpoint counts participants with a 30% reduction in ALT from baseline to month 12. ALT is a serum measure, and the registry frames the endpoint as whether each participant reaches a specified reduction threshold. This converts a continuous laboratory measurement into a binary outcome: a participant either meets the stated threshold or does not.

General statistical context: a threshold outcome can be easier to interpret as the number of people meeting a prespecified response definition, but it does not preserve all the information in the underlying measurement. Two participants with different ALT changes may fall into the same category if both are on the same side of the threshold. Conversely, changes close to the threshold can result in different classifications even when the underlying values are similar. These are general features of threshold-based outcomes, not additional findings about this trial.

The registry identifies the primary endpoint as a participant count and gives baseline and 12 months as its timeframe. The arm entries list 19 for pentoxifylline and 7 for placebo, matching the listed group sizes; they do not yield an estimable count contrast for independent recalculation. A fuller account of endpoint measurement, handling of values near the threshold, and missing ALT measurements is not reported in the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The statistical information in the registry combines a binary endpoint, a sponsor-reported odds ratio, and a method label of ANOVA. The record does not explain the analysis implementation, so the reported fields should be read as they stand rather than used to infer an unreported model or procedure.

What does a binary endpoint represent?

General explanation: a binary endpoint places each participant into one of two categories. For this study, the intended distinction is whether a participant had a 30% reduction in ALT between baseline and month 12. A group-level comparison can then be expressed using counts or proportions of participants meeting the definition, or using a relative measure such as an odds ratio. The registry identifies the endpoint as a count of participants but does not provide usable event counts for a recalculation.

The entries for the two arms list 19 and 7, and the same numbers are listed as the respective group sizes. Because the registry result does not yield a recalculable binary contrast, those entries cannot be treated as informative response counts for an independent effect calculation. The record does not clarify why the arm entries take these values or provide another set of arm-specific counts for the threshold outcome.

How should the odds ratio be read?

General explanation: an odds ratio compares the odds of an outcome between groups. An OR of 1 is the point of no difference in estimated odds. Values on either side of 1 would indicate different directions of the estimated association, but a point estimate alone does not convey precision. An interval estimate is commonly used to show how uncertain the estimate is. The registry does not report a confidence interval for the OR, so the range of estimates compatible with the sponsor’s analysis cannot be assessed from that result.

The record reports p = 0.46. General explanation: a p-value is calculated under a specified null hypothesis and analysis procedure; it describes how unusual the observed result, or a more extreme one, would be under that setup. It is not the probability that the null is true, nor does it describe the size of a treatment effect. Because the registry does not provide fuller details of the test, the p-value cannot be used to reconstruct the analysis.

What can the method label tell readers?

The registry lists ANOVA as the analysis method and Odds Ratio (OR) as the parameter. ANOVA is commonly associated with comparisons of means, whereas an odds ratio is a relative measure for an outcome represented by event and non-event categories. The registry does not explain how these labels connect in this analysis. It would be speculative to infer a particular model, transformation, correction, or test from the labels alone.

The posted information is not sufficient for an independent calculation of the primary binary comparison. There are no recalculated event counts, effect estimate, or confidence interval to report. The sponsor’s result remains the available reported analysis; its relation to any independently recalculated estimate cannot be assessed because a recalculation is not estimable from the posted summaries.

Why do event counts and uncertainty matter?

General explanation: binary comparisons depend on how many participants meet the endpoint in each group and how many do not. Those counts determine the estimated group contrast and contribute to its uncertainty. Without them, readers cannot reproduce the odds ratio or examine how changes in individual classifications might affect the estimate. A confidence interval would add information about precision, but none is reported for this result.

General explanation: over-dispersion is a concern when observed variability exceeds what a model assumes. It matters when the chosen statistical model relies on a particular variance structure, because underestimated variability can make uncertainty appear smaller than it is. The registry does not report an over-dispersion assessment or enough model detail to determine whether this issue applies to the primary analysis. It would not be appropriate to claim that over-dispersion was present or absent.

How should planning illustrations be distinguished from trial results?

The planning assumptions recorded for sample-size illustrations are a two-sided significance level of 0.05, equal allocation, and a normal approximation; the effect to detect is defined as a stated fraction of the observed effect. These are planning assumptions, not a description of the trial’s original design justification. The record does not report the trial’s own sample-size justification. Any sample-size illustration based on an observed effect is a planning exercise and is not observed power. It should not be interpreted as evidence that the completed study had a particular probability of detecting an effect.

General explanation: planning calculations ask how many participants per arm might be needed under selected assumptions about the effect, significance level, allocation, and statistical approach. Their conclusions depend on those assumptions. They do not change the observed results or replace a confidence interval. The registry does not report whether the original design used these same assumptions, so the listed planning framework should not be presented as the trial’s original statistical plan.

10. Exact Estimates for Each Arm

No two-arm comparison is shown for this trial, but each arm's result can still be summarized with exact interval estimates computed from the posted counts. The Clopper-Pearson interval is exact and conservative; the Wilson interval is a common closer-to-nominal alternative; the Bayesian credible interval uses a uniform Beta(1, 1) prior. For small groups the three can differ noticeably.

R code and output

arm    <- c("Pentoxifylline", "Placebo")
n      <- c(19, 7)
events <- c(19, 7)
for (i in seq_along(n)) {
  ci <- binom.test(events[i], n[i])$conf.int          # Clopper-Pearson exact interval
  cr <- qbeta(c(0.025, 0.975), 1 + events[i], 1 + n[i] - events[i])   # Bayesian, uniform Beta(1,1) prior
  wi <- (function(x, n) { p <- x / n; z <- qnorm(0.975); d <- 1 + z^2 / n
    c((p + z^2 / (2 * n) - z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d, (p + z^2 / (2 * n) + z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d) })(events[i], n[i])
  cat(sprintf("ARM|%s|%d|%d|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f\n", arm[i], n[i], events[i], 100 * events[i] / n[i],
              100 * ci[1], 100 * ci[2], 100 * wi[1], 100 * wi[2], 100 * cr[1], 100 * cr[2]))
}

ARM  Pentoxifylline  19  19  100.0  82.4  100.0  83.2  100.0  83.2  99.9
ARM  Placebo  7  7  100.0  59.0  100.0  64.6  100.0  63.1  99.7

11. Analyses Reported by the Sponsor

The sponsor-reported primary comparison is an OR of 1 and p = 0.46. The registry identifies ANOVA as the analysis method and does not report a confidence interval or explain the method and parameter combination in further detail.

The registry arm entries are 19 for pentoxifylline and 7 for placebo, matching the listed group sizes. They do not support an independent recalculation of the binary endpoint comparison. Accordingly, there is no recalculated estimate or p-value to place alongside the sponsor’s result. The sponsor’s point estimate and p-value are the reported analysis, not an independently verified calculation.

The record does not flag a discrepancy between a recalculation and the sponsor analysis. That does not resolve the limits caused by the lack of usable endpoint counts and a confidence interval. The OR of 1 is the reported point estimate; the p-value does not demonstrate equivalence, and neither statistic alone establishes the precision or clinical importance of the result.

The primary analysis is described as intention to treat, with a per-protocol secondary analysis also performed and no differences between the two analyses reported. Detailed results for that secondary analysis are not included in the record. Consequently, the relationship between the reported sponsor result and the per-protocol analysis cannot be examined further.

12. Multiplicity, Missing Data and Analysis Populations

The registry lists one primary endpoint and three secondary outcomes. The primary endpoint is the number of participants with a 30% reduction in ALT from baseline to month 12. The secondary outcomes concern change in TNF-α, serum leptin, and serum adiponectin. The leptin and adiponectin outcomes are described over 12 months; the TNF-α timeframe is not reported in the registry record.

The record lists four outcome measures and four analyses. It does not report a multiplicity adjustment, a hierarchy among secondary outcomes, or a prespecified strategy for controlling the chance of false-positive findings across multiple comparisons. Without that information, the statistical interpretation of any collection of tests cannot be evaluated for multiplicity. The presence of several outcomes alone does not establish whether or how the investigators adjusted their analyses.

The primary population is described as intention to treat, and the registry states that a per-protocol secondary analysis was performed with no differences between the two analyses. It does not report the per-protocol criteria, the number included in that analysis, or detailed estimates for it. The phrase that there were no differences is not accompanied by numerical results or a definition of what was compared.

Missing-data procedures are not reported in the registry record. In particular, the record does not describe how missing ALT measurements, incomplete follow-up, or missing secondary-outcome measurements were handled. It also does not report the amount or pattern of missing data. These omissions limit assessment of whether the reported analysis reflects all randomized participants and whether the endpoint classifications could have been affected by unavailable measurements.

General statistical context: intention-to-treat analysis generally aims to preserve the original randomized group assignment when comparing groups, while a per-protocol analysis typically focuses on participants meeting specified adherence or eligibility conditions. The exact meaning depends on the study’s definitions and procedures. Here, those details are not reported, so the labels should not be taken as a complete account of how participants or missing outcomes were handled.

13. Safety Results

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

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

The registry reports 0 serious adverse events among the 19 participants listed at risk in the pentoxifylline arm and 0 among the 7 listed at risk in the placebo arm. The record lists no serious adverse-event terms. It also reports 0 other adverse events among the participants listed at risk in each arm.

These figures describe what is recorded in the registry, not a demonstration that serious or other adverse events are impossible. General statistical context: when no event is recorded in a group, the estimate of observed event frequency is zero, but uncertainty about the underlying risk remains, especially when the amount of observed safety information is limited. The event counts alone do not establish a comparative safety profile.

Death counts are not reported in the registry record. The timeframe and collection procedures for the adverse-event summaries are not reported in the registry record either. The primary endpoint timeframe is baseline and 12 months, but the safety summaries are not explicitly tied to that timeframe. These limits mean the adverse-event figures should be interpreted as registry-reported counts, without assuming a specific follow-up period or completeness beyond what is recorded.

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

The registry describes a randomized, double-masked comparison of pentoxifylline and placebo in people with NASH. Its primary question concerns whether participants had a 30% reduction in ALT between baseline and month 12. The sponsor’s reported comparison is an OR of 1 with p = 0.46.

That result gives a point estimate and a p-value, but it does not provide a confidence interval. The arm entries do not permit an independent recalculation of the binary comparison, and the record does not explain how the listed ANOVA method produced the reported OR. As a result, readers cannot assess the precision of the effect estimate or reproduce the comparison from the registry summary.

General statistical context: failure to obtain a small p-value does not establish that treatments are equivalent. Demonstrating equivalence requires a suitable design and a prespecified margin, with an analysis that evaluates whether the uncertainty lies within that margin. Such a margin and an equivalence analysis are not reported in the registry record. Likewise, a point estimate of OR 1 does not prove that the underlying odds are exactly equal.

The primary endpoint is a threshold-based laboratory outcome. It does not, by itself, describe the full distribution of ALT changes, other clinical outcomes, or the secondary biomarker outcomes. The record lists secondary outcomes involving TNF-α, leptin, and adiponectin, but detailed results for those measures are not reported in the information available for this analysis.

The study’s recorded enrollment and group sizes provide context for the extent of the comparison, but the missing endpoint counts and interval estimate limit what can be concluded about precision. The registry results do not establish a broad treatment effect, a lack of effect, equivalence, or a complete safety profile. They describe a specific reported analysis of a threshold ALT outcome, with important limits on independent verification and interpretation.

15. Why This Trial Matters Statistically

This trial illustrates how the statistical meaning of a clinical result depends on the information reported alongside a point estimate and p-value. The primary outcome is a binary threshold response: whether ALT fell by 30% from baseline to month 12. That endpoint definition is more specific than a general claim about change in liver tests, but interpretation still depends on access to the underlying group counts and an estimate of uncertainty.

The sponsor-reported OR is 1, while the record gives p = 0.46. Those values do not answer every statistical question readers may have. Without a confidence interval, it is difficult to assess precision. Without usable arm-specific event counts and further method detail, the comparison cannot be independently reproduced from the registry summary. A familiar effect measure does not remove the need for transparent outcome data.

The record also shows why a result should not be reduced to whether a p-value crosses a threshold. A p-value is tied to an analysis model and does not state the probability that a treatment effect exists. Nor does a point estimate at OR 1 establish equivalence. Those distinctions matter particularly when the registry does not report the interval estimate or explain how the method and parameter labels fit together.

Other statistical questions remain relevant: how missing measurements were handled, how the intention-to-treat and per-protocol analyses were defined, and whether the several outcomes were considered under a multiplicity plan. These details are not reported in the registry record. Their absence does not prove that no procedures were used; it means they cannot be assessed from the record.

The study is therefore useful as an example of separating the stated clinical question from what the reported statistics can establish. The endpoint, design, and sponsor result are identifiable, while precision, reproducibility, and several analysis choices remain unclear. That distinction supports careful reading without turning an incomplete registry summary into a stronger conclusion than it warrants.

16. 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.

Several trial-specific limitations affect interpretation of the primary comparison. The registry arm entries list 19 for pentoxifylline and 7 for placebo, matching the listed group sizes, but do not provide a usable set of binary endpoint counts for recalculation. The sponsor reports an OR of 1 and p = 0.46, yet no confidence interval is reported. The precision of the estimate therefore cannot be assessed from the registry result.

The registry identifies ANOVA as the method and OR as the parameter, but does not explain how those labels relate to the analysis. It does not report further statistical details sufficient to reproduce the primary result. The absence of a recalculated estimate also means there is no independent comparison with the sponsor’s analysis.

The record states that the primary analysis was intention to treat and that a per-protocol secondary analysis found no differences between the two analyses. It does not give the per-protocol definition, participant counts for that analysis, or numerical results. The amount and handling of missing outcome data are not reported. These omissions prevent assessment of how incomplete measurements may have affected the endpoint comparison.

The record lists three secondary outcomes in addition to the primary endpoint, but does not report a multiplicity strategy or detailed results for those outcomes. Participant characteristics, eligibility criteria, allocation procedures, and a sample-size justification are not reported in the registry record. Safety summaries list no adverse events but do not specify their timeframe; death counts are not reported. These are reporting limits, not evidence that the unreported procedures or information did not exist.

17. Related Tutorials

18. Related Statistical Calculators

19. Sources

• ClinicalTrials.gov: Pentoxifylline/Nonalcoholic Steatohepatitis (NASH) Study: The Effect of Pentoxifylline on NASH — NCT00267670.
• Registry record last updated 2014-09-09; status last verified 2014-08.
• Results first posted 2014-09-09.
• 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.

20. Record Summary

NCT00267670 is a completed randomized, double-masked, parallel-group study sponsored by Northwestern University and listed as Phase 2 and Phase 3. The recorded enrollment is 26. The study start is 2005-03, and primary completion is 2009-09.

The study compared pentoxifylline, described as 400mg PO TID, with placebo, described as 1 pill PO TID, in people with NASH. Its primary endpoint was the number of participants with a 30% reduction in ALT from baseline to month 12. The sponsor-reported comparison is OR = 1 with p = 0.46; a confidence interval is not reported, and the binary comparison cannot be independently recalculated from the arm summaries.

The record lists secondary outcomes for TNF-α, leptin, and adiponectin. It reports 0 serious adverse events in each arm and 0 other adverse events in each arm; death counts are not reported. Details about missing data, multiplicity, and the full statistical procedure are not reported in the registry record.