SGI-110 in Combination With Carboplatin in Ovarian Cancer: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for SGI-110 in Combination With Carboplatin in Ovarian Cancer (NCT01696032), a randomized phase 2 study. The primary outcome is Stage 1: Dose Limiting Toxicities.

2. Clinical Question

3. Trial Design

This was a Phase 2, randomized, controlled, open-label, parallel trial with two parts. The registry describes Stage 1 as a safety lead-in with dose escalation. Participants received guadecitabine with carboplatin, and the optimum guadecitabine dose was identified based on safety and efficacy. Stage 1 included a 30 mg/m2 group with 14 participants and a 45 mg/m2 group with 6 participants for the primary endpoint.

Stage 2 was an open-label randomized comparison. Participants were assigned in a 1:1 ratio to guadecitabine plus carboplatin or treatment of choice. The registry describes the Stage 2 combination as guadecitabine 30 mg/m2 subcutaneously once daily on Days 1-5 and carboplatin intravenously at AUC 4 on Day 8 in 28-day cycles. Treatment of choice could be topotecan, pegylated liposomal doxorubicin, paclitaxel, or gemcitabine, at the investigator’s discretion. The Stage 2 groups included 51 participants assigned to the combination and 49 assigned to treatment of choice. A further group is listed as crossover treatment of choice to guadecitabine plus carboplatin.

The distinction between stages matters when interpreting estimates. The Stage 1 dose groups address dose escalation and safety; they were not randomized against each other in the manner described for Stage 2. The primary endpoint is reported for Stage 1, while the registry’s available secondary estimates include both Stage 1 and Stage 2 groups. The listed enrollment is 120, and the trial status is completed. The registry record does not report additional details about the randomization sequence, allocation concealment, or the timing and conduct of any interim analyses.

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.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Safety data set includes data from all participants who received any amount of study drug, including any component of a multi-dose study treatment regimen.

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.

Population: Includes all treated participants

6. Primary Endpoint

The registry lists the primary outcome as Stage 1: Dose Limiting Toxicities, measured over Up to 12 months.

The primary endpoint is the number of Stage 1 participants with dose-limiting toxicities during follow-up of up to 12 months. The registry defines the outcome as a count of participants, rather than a continuous measurement or a time-to-event estimate. The safety data set includes anyone who received any amount of study drug, including any component of a multi-dose study treatment regimen.

A DLT count is used in a dose-escalation setting to describe clinically important toxicity observed while assessing treatment doses. Such counts can inform dose selection, but they are not, by themselves, a complete description of safety. They do not convey the severity, duration, timing, reversibility, or attribution of each event unless those details are separately reported. The registry record does not provide a DLT definition or the rules used to determine whether an event met that definition.

The posted counts are 0 participants among 14 in the 30 mg/m2 Stage 1 group and 4 among 6 in the 45 mg/m2 Stage 1 group. These are descriptive results from small dose groups. The registry does not provide confidence intervals or a formal comparison for this primary endpoint. The counts therefore describe the observed Stage 1 experience; they do not establish a precise estimate of the underlying frequency of DLTs or a randomized treatment effect.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is a count of participants with a DLT, so its basic interpretation is straightforward: the count records how many participants met the endpoint definition. The posted Stage 1 values are 0 of 14 at 30 mg/m2 and 4 of 6 at 45 mg/m2. The endpoint is attached to the safety data set, defined in the registry as participants who received any amount of study drug. That population definition helps identify whose safety experience contributes to the count. It does not clarify the event definition or adjudication process, which are not reported in the registry record.

Small samples make estimates uncertain. When only a few participants are observed, each individual outcome has a comparatively large influence on the group result. A count such as 0 among 14 does not establish that the underlying risk is zero; it means that no DLTs were observed in that group. Similarly, 4 among 6 is a description of the observed group and not a precise estimate of what would occur in a larger population. The registry does not post confidence intervals for the primary count, so uncertainty cannot be quantified from the primary endpoint result alone using a reported interval.

For binary outcomes expressed as percentages, a confidence interval describes a range of values compatible with the observed data under the method used to construct it. It communicates precision, not the probability that a particular treatment effect is true. Wider intervals indicate less precision. In the posted clinical benefit rate and CA-125 estimates, the intervals are wide, consistent with the relatively limited group sizes. The registry record does not identify the interval-construction method, so the particular assumptions behind those intervals are not reported.

The risk differences and related comparisons are labelled as recalculated, unadjusted results. A risk difference expresses the difference between two proportions on an absolute scale. A value below zero indicates a lower observed proportion in the comparison arm under the stated direction of subtraction; a value above zero indicates a higher proportion. An interval that includes zero is compatible with no difference as well as with differences in either direction. These comparisons use the Stage 1 30 mg/m2 group as the reference and Stage 2 groups as the other arms. Since the reference is from Stage 1, the resulting estimates do not retain the protection of a direct randomized Stage 2 comparison against treatment of choice.

The unadjusted p-value measures how unusual the data would be under a specified null hypothesis, according to the test used. It is not the probability that the null hypothesis is true, nor does it measure clinical importance. These p-values are not listed as analyses reported by the registry. They are unadjusted recalculations, and no multiple-comparison correction is described for them. Their interpretation should account for the many outcomes and comparisons that may be considered, as well as the nonconcurrent stage reference.

The planning illustrations use observed effects as inputs and display hypothetical group sizes for specified power levels and fractions of the observed effect. They are not observed power and are not the trial’s original sample-size justification. The registry’s illustration assumptions are a two-sided significance level of 0.05, equal allocation, and a normal approximation; the selected effect is a stated fraction of the observed effect. Because the observed effects can be imprecise and may be affected by the stage mismatch, planning outputs based on them should not be treated as reliable design recommendations. The trial’s own sample-size justification is not reported in the registry record.

Over-dispersion is a concern in count or rate data when variability exceeds what a simple model assumes. General statistical context: a model that underestimates variability can produce intervals that are too narrow and tests that are too optimistic. This trial’s primary endpoint is reported as a participant count, not as a model-based rate, and the registry record does not describe a model or an over-dispersion assessment. It would therefore be inappropriate to infer that such a model was used or that an over-dispersion adjustment was made.

Randomization supports a causal comparison when the randomized groups are analyzed as assigned and when the analysis respects the randomized design. In this record, Stage 2 is described as randomized 1:1, but the posted recalculated comparisons use a Stage 1 group as the reference. The registry record does not report a direct Stage 2 comparative analysis for these secondary endpoints. Nor does it report whether outcomes were analyzed by intention to treat, per protocol, or another analysis population. Those omissions prevent a full assessment of the analysis strategy.

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("Stage 1: Guadecitabine+Carboplatin 30 mg/m2", "Stage 1: Guadecitabine+Carboplatin 45 mg/m2")
n      <- c(14, 6)
events <- c(0, 4)
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  Stage 1: Guadecitabine+Carboplatin 30 mg/m2  14  0  0.0  0.0  23.2  0.0  21.5  0.2  21.8
ARM  Stage 1: Guadecitabine+Carboplatin 45 mg/m2  6  4  66.7  22.3  95.7  30.0  90.3  29.0  90.1

11. Multiplicity, Missing Data and Analysis Populations

The registry lists multiple secondary outcomes, including objective response rate, progression-free survival at 6 months, clinical benefit rate, CA-125 levels, duration of response, overall survival, and Stage 1 pharmacokinetic measures. The available record does not report a prespecified hierarchy among these outcomes, a family-wise error strategy, or another multiplicity adjustment. It also does not explain whether the posted comparisons were selected before data review. Without this information, unadjusted p-values should be interpreted cautiously: they do not account for the number of outcomes or comparisons considered.

The record does not report how missing outcome data were handled, whether participants were censored or excluded from particular analyses, or whether sensitivity analyses assessed the impact of missingness. The analysis population for the secondary outcomes is also not reported in the registry record. Although the primary safety population is defined as participants who received any amount of study drug, that definition should not be assumed to apply to every secondary endpoint.

The registry describes Stage 2 randomization in a 1:1 ratio, but it does not provide a reported analysis of the Stage 2 randomized comparison for the secondary outcomes discussed in the recalculations. Those comparisons use the Stage 1 30 mg/m2 group as a reference. The record does not explain the rationale for that reference in the recalculated estimates or provide an adjustment for differences between trial stages. The randomization method, allocation concealment, and any stratification factors are not reported in the registry record.

The primary endpoint is a Stage 1 safety count, but the registry record does not report its DLT criteria or a formal inferential method. It also does not report a sample-size justification for the primary endpoint, an interim-analysis plan, stopping boundaries, or missing-data sensitivity analyses. These are unknown from the registry record; no particular procedure should be assumed. Together, these omissions limit assessment of the prespecification, completeness, and robustness of the statistical analysis.

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 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: neutropenia (33.3%), vomiting (33.3%), nausea (33.3%), abdominal pain (18.5%), adverse drug reaction (16.7%).

The adverse-event record gives counts of participants at risk and counts affected by serious adverse events and other adverse events, across Stage 1, Stage 2, and the crossover group. For serious adverse events, the listed counts are 8 of 14 in Stage 1 at 30 mg/m2, 4 of 6 in Stage 1 at 45 mg/m2, 26 of 51 in the Stage 2 combination group, 24 of 49 in the Stage 2 treatment-of-choice group, and 12 of 27 in the crossover group. These are descriptive counts, not adjusted comparisons.

Serious adverse events are not interchangeable with DLTs: the primary endpoint is a specifically defined DLT count, while the adverse-event summary describes serious events. The registry record does not report how serious-event status relates to DLT classification, whether events were judged treatment-related, or the detailed event criteria. The record also lists deaths by group, but does not explain their relationship to adverse events or treatment. They should not be interpreted as treatment-caused deaths.

The listed serious-event terms include neutropenia, vomiting, nausea, abdominal pain, adverse drug reaction, and mental status change, with percentages reported for those terms. The record does not provide enough detail to assign these terms to a particular treatment group or to infer causality. Differences across groups should not be interpreted as treatment effects without accounting for the trial stages, group sizes, follow-up, and the method used to collect and summarize events.

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

The primary result describes the Stage 1 experience with guadecitabine plus carboplatin at two dose levels: 0 participants with a DLT among 14 in the 30 mg/m2 group, and 4 among 6 in the 45 mg/m2 group. These observed counts provide a safety description for the dose-escalation stage. They do not estimate the comparative effect of the combination versus treatment of choice, because that randomized comparison occurred in Stage 2 and is not the primary endpoint.

The result does not establish that the lower dose has no DLT risk, or that the higher dose will produce the same frequency in another group. The small numbers and lack of a reported confidence interval for the primary endpoint mean that the precision of the underlying risks is not quantified. The DLT definition and adjudication procedure are not reported in the registry record, adding uncertainty about how the count was formed.

The secondary outcomes provide additional descriptions but do not supply a definitive answer to the randomized comparative question. The available estimates have broad intervals, and recalculated comparisons use a reference group from Stage 1. Those results should not be interpreted as direct randomized evidence for one Stage 2 option over another. Other listed secondary outcome results and analysis details are not reported in the registry record.

These findings are statistical descriptions of the registry record, not a clinical recommendation. They do not establish whether a treatment should be selected for an individual, and the posted results cannot answer questions beyond the endpoints and populations described. Interpretation depends on design, endpoint definitions, follow-up, missing-data handling, and analysis choices; important details for several of these areas are not reported in the registry record.

14. Why This Trial Matters Statistically

This trial illustrates why a multi-stage design requires stage-specific interpretation. The primary endpoint comes from a safety lead-in and describes DLT counts across dose groups. The randomized comparison belongs to Stage 2. Treating the dose-escalation groups as though they were randomized controls would confuse dose selection with comparative treatment evaluation.

It also illustrates how a registry can list a randomized trial while the available posted analyses do not directly answer the randomized comparison. The secondary recalculations use the Stage 1 30 mg/m2 group as a reference for Stage 2 arms. This choice produces numerical contrasts, but it does not make the groups equivalent to randomized arms. The distinction between a statistical contrast and a design-protected treatment comparison is important when interpreting estimates and p-values.

Small groups and wide confidence intervals are another central feature. A point estimate alone can appear definite even when the data support a broad range of possibilities. Reading the interval alongside the estimate is essential, especially for secondary outcomes. The unadjusted nature of the comparisons and the absence of a reported multiplicity plan further limit confirmatory interpretation.

Finally, planning illustrations based on observed effects are not observed power and are not the trial’s original design justification. They are conditional illustrations that depend on assumptions and on the observed effect used as an input. This distinction helps prevent retrospective calculations from being mistaken for design evidence. The record’s missing details about endpoint definitions, missing data, analysis populations, and multiplicity also show why registry results need to be interpreted with attention to what is documented as well as what remains unreported.

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 primary endpoint is reported only as counts, without a confidence interval, formal comparison, or detailed DLT definition. The groups are small, particularly the Stage 1 45 mg/m2 group, so the observed count does not precisely characterize the underlying frequency. The registry record does not report how DLTs were adjudicated, how long each participant was followed within the stated timeframe, or whether exposure differed across participants.

The trial has different design roles across its stages. Stage 1 was a safety lead-in with dose escalation, while Stage 2 was randomized. The available recalculated secondary comparisons use a Stage 1 group as a reference for Stage 2 arms. Differences between stages and their participant groups may affect these contrasts, and no adjustment for stage differences is reported. The listed crossover group adds another treatment context, but the registry record does not provide enough analysis detail to establish how it relates to the main randomized comparison.

The trial was open-label, and the registry does not report blinding of outcome assessment. The investigator selected among several treatment-of-choice options, so the comparator represents different possible regimens rather than a single named treatment. The record does not report how treatment choice was distributed or whether outcomes were analyzed separately by selected regimen.

For secondary outcomes, the registry record does not report a multiplicity plan, missing-data methods, analysis populations, or detailed analysis methods. Numerical results for several listed endpoints are not reported in the registry record. The available estimates are therefore incomplete and should not be treated as a comprehensive assessment of efficacy or safety. These limitations restrict precision and interpretation; they do not by themselves determine the direction of any treatment effect.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: SGI-110 in Combination With Carboplatin in Ovarian Cancer — NCT01696032.
• Registry record last updated 2024-08-27; status last verified 2024-07.
• Results first posted 2021-05-25.
• 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

The registry identifies NCT01696032 as a completed Phase 2, randomized, controlled, open-label, parallel trial of SGI-110 with carboplatin in platinum-resistant recurrent ovarian cancer. Enrollment was 120. The study had a Stage 1 dose-escalation safety lead-in and a randomized Stage 2 comparing guadecitabine plus carboplatin with investigator-selected treatment of choice.

The primary endpoint was Stage 1 participants with DLTs, followed for up to 12 months. The posted counts were 0 of 14 in the 30 mg/m2 group and 4 of 6 in the 45 mg/m2 group. Available secondary estimates include clinical benefit rate and CA-125 levels, but their recalculated comparisons use a Stage 1 reference group and are unadjusted. The registry record does not report key details including the DLT definition, multiplicity plan, missing-data methods, and analysis populations for secondary outcomes.