A Study Of Inotuzumab Ozogamicin Plus Rituximab For Relapsed/Refractory Aggressive Non-Hodgkin Lymphoma Patients Who Are Not Candidates For Intensive High-Dose Chemotherapy: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study Of Inotuzumab Ozogamicin Plus Rituximab For Relapsed/Refractory Aggressive Non-Hodgkin Lymphoma Patients Who Are Not Candidates For Intensive High-Dose Chemotherapy (NCT01232556), a randomized phase 3 study. The primary outcome is Overall Survival.

2. Clinical Question

The trial asked whether adding inotuzumab ozogamicin to rituximab could improve overall survival compared with investigator’s choice of rituximab plus gemcitabine or rituximab plus bendamustine. It enrolled people with relapsed or refractory CD22-positive aggressive non-Hodgkin lymphoma who were not candidates for intensive high-dose chemotherapy. The primary endpoint was overall survival, defined as time from randomization to death due to any cause.

The registry describes the study’s objective as demonstrating superiority on overall survival. That is a question about comparative outcomes, not simply whether either group had participants who survived for a particular length of time. The posted results report median survival in each group and a hazard ratio with a confidence interval and p-value. Those measures describe the observed survival experience and the uncertainty around the comparison. They do not, on their own, establish that the treatments are equivalent or that a clinically important difference is impossible.

3. Trial Design

This was a randomized, open-label, parallel-group phase 3 study sponsored by Pfizer. The registry lists enrollment of 338. Participants were allocated to inotuzumab ozogamicin plus rituximab or to investigator’s choice of rituximab plus gemcitabine or rituximab plus bendamustine. Masking was listed as none, so the study was not described as blinded.

The study began on 2011-04-04, and the primary completion date was 2014-03-28. Its registry status is terminated; the reason for termination is not reported in the registry record. The listed overall survival timeframe was from randomization up to 5 years after the last dose or up to the final study visit, whichever occurred first. The primary endpoint population was the ITT population.

The study also listed progression-free survival, response outcomes based on NCI International Response Criteria for NHL, duration of response, health status assessed with EQ-5D, and health-related quality of life assessed with FACT-Lym. The registry record lists these secondary outcome areas, but it does not report results for every listed measure. The reported study design and results do not establish why the trial was terminated, what its original sample-size justification was, or what statistical choices were made beyond those described in the record. Those details are not reported in the registry record.

Random allocation is intended to support a fair comparison by assigning participants to treatment groups through a chance-based process. In general, randomization helps reduce systematic differences between groups at the start of a trial, although it does not guarantee that every measured or unmeasured characteristic will be identical. Open-label treatment assignment can affect care, reporting, and assessments, particularly for outcomes that involve judgment or participant report. Overall survival is based on death from any cause, but follow-up and censoring still matter when estimating survival over time.

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: ITT Population.

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: Intent-to-treat (ITT) Population - included all participants who are randomized into the study.

6. Primary Endpoint

The registry lists the primary outcome as Overall Survival, measured over From randomization up to 5 years after last dose or up to final study visit, whichever occurs first..

Overall survival measures the time from randomization until death from any cause. In this trial, participants who were still alive at their last contact or at the end of the study were censored at that date. The registry describes the analysis population as ITT and lists a follow-up window extending from randomization up to 5 years after the last dose or the final study visit, whichever occurred first.

Overall survival is used in trials because it captures a direct, clinically meaningful event and does not depend on assigning a cause to death. The median is the time by which half of the estimated study population has experienced the event, according to the survival estimate. It gives a compact summary, but it does not describe the entire survival pattern or show how individual participants fared.

The registry says Kaplan-Meier methods were used to estimate overall survival and that a stratified Cox proportional hazards regression produced the hazard ratio and its 95% two-sided confidence interval. The hazard ratio summarizes the relative event rate over follow-up under the model; it is not a comparison of the two medians. The registry also lists a Log Rank comparison. These measures complement one another: the medians describe each group’s estimated survival distribution, while the hazard ratio and its interval address the between-group comparison.

7. Results Posted in the Registry

• Inotuzumab Ozogamicin+Rituximab vs Rituximab+Gemcitabine or Rituximab+Bendamustine: hazard ratio 1.08 (95% CI 0.82–1.44), reported p = 0.708 (as reported in the registry).

8. Secondary Efficacy Endpoints

The registry also posts results for secondary outcomes. For each outcome below, the endpoint type, the test applied to the posted summary statistics, the result and its interpretation are given in turn. Secondary endpoints are usually not adjusted for multiple testing, so they support rather than confirm the primary finding.

Progression-Free Survival (PFS)

• Endpoint type: Time to event: compared as a hazard ratio.
• Test: Hazard ratio as reported by the sponsor (Log Rank); it was not recalculated.
• Result:
  • Inotuzumab Ozogamicin+Rituximab vs Rituximab+Gemcitabine or Rituximab+Bendamustine: hazard ratio 0.92 (95% CI 0.72–1.19), reported p = 0.271 (as reported in the registry).
• Interpretation:
  • The 95% confidence interval includes no difference (a ratio of 1), so these data do not show a statistically significant difference between Inotuzumab Ozogamicin+Rituximab vs Rituximab+Gemcitabine or Rituximab+Bendamustine and the control. The interval gives the range of effects that remain compatible with the data.

Percentage of Participants With A Best Overall Response of CR or Partial Response (PR) Per NCI International Response Criteria for NHL

• Endpoint type: Binary: the percentage of participants with the event (counts rebuilt from the posted percentages and group sizes).
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • Rituximab+Gemcitabine or Rituximab+Bendamustine: 29.7% (51/172) versus 29.5% (49/166) on control; risk difference 0.1 percentage points (95% CI -9.6 to 9.9), risk ratio 1.00 (0.72–1.40); unadjusted two-proportion z-test p = 0.979.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Rituximab+Gemcitabine or Rituximab+Bendamustine and the control. The interval gives the range of effects that remain compatible with the data.

Percentage of Participants With A Best Overall Response of CR, Unconfirmed CR (unCR), PR, or Unconfirmed PR (unPR) Per NCI International Response Criteria for NHL

• Endpoint type: Binary: the percentage of participants with the event (counts rebuilt from the posted percentages and group sizes).
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • Rituximab+Gemcitabine or Rituximab+Bendamustine: 43.6% (75/172) versus 41.0% (68/166) on control; risk difference 2.6 percentage points (95% CI -7.9 to 13.2), risk ratio 1.06 (0.83–1.37); unadjusted two-proportion z-test p = 0.623.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Rituximab+Gemcitabine or Rituximab+Bendamustine and the control. The interval gives the range of effects that remain compatible with the data.

The secondary results cover several kinds of outcomes: progression-free survival, response, duration of response, health status, and health-related quality of life. The posted numerical results include progression-free survival and two response definitions. Duration of response and the questionnaire outcomes are listed, but their results are not reported in the registry record.

Across the posted secondary outcomes, the reported comparisons do not show statistically significant differences by their reported p-values. This broad pattern is consistent with the primary result in the limited sense that none of these posted secondary comparisons supplies clear statistical evidence of a between-group difference. It does not turn the primary result into evidence of equivalence, and it does not establish that all endpoints measure the same underlying effect. Progression-free survival and overall survival capture different events; response outcomes capture tumor response under specified criteria rather than survival time.

Secondary outcomes usually provide supporting or exploratory evidence, especially when a trial evaluates multiple endpoints. When many outcomes are examined, the chance of at least one apparently notable result can increase unless the analysis plan addresses multiplicity. The registry record does not report a multiplicity adjustment or a hierarchy governing these outcomes. Consequently, isolated secondary findings would need cautious interpretation, and non-significant secondary tests do not prove the absence of effects.

The sponsor’s response analyses use a stratified method, while the registry’s unadjusted recalculations are simpler comparisons of the posted counts. They should not be conflated. The registry does not report a complete statistical analysis plan for the secondary outcomes, so details such as the handling of missing assessments and the intended inferential role of each endpoint are not established by the record.

Secondary outcomes registered for this trial, for reference (only those above were recalculated):

• Progression-Free Survival (PFS)
• Percentage of Participants With A Best Overall Response of CR or Partial Response (PR) Per NCI International Response Criteria for NHL
• Percentage of Participants With A Best Overall Response of CR, Unconfirmed CR (unCR), PR, or Unconfirmed PR (unPR) Per NCI International Response Criteria for NHL
• Duration of Response
• Health Status as Assessed by the European Quality of Life 5 Dimension (EQ-5D) Questionnaire
• Health Related Quality of Life as Assessed by the Functional Assessment of Cancer Therapy for Lymphoma (FACT-Lym) Questionnaire

9. Statistical Interpretation

10. Statistical Methods Explained

The primary endpoint is time-to-event data: participants can have different follow-up lengths, and some may not experience the event before their last observed contact. The registry states that overall survival was estimated with the Kaplan-Meier method, with censoring at the date of last contact or the end of the study. It also states that a stratified Cox proportional hazards regression was used to calculate the hazard ratio and its 95% two-sided confidence interval. The comparison method listed in the results is Log Rank.

How does Kaplan-Meier estimation work?

General statistical context: Kaplan-Meier estimation uses information from participants who are still under observation at each event time. A participant who has not died by the time follow-up ends contributes information up to the censoring date; censoring is not counted as a death. The method combines the observed event experience over follow-up to estimate the probability of remaining alive. This allows participants with different follow-up durations to contribute without treating every participant as if they had the same observation period.

The estimated median is the point in time at which the estimated survival probability reaches the halfway mark. It is a useful summary, but it leaves out much of the shape of the survival curve. Two groups can have the same median while differing at other follow-up times. Confidence intervals around medians describe uncertainty in their estimates, not the probability that an individual participant will survive for a particular period.

What does the hazard ratio describe?

General statistical context: A hazard is the instantaneous event rate among participants who have not yet experienced the event. A hazard ratio compares those rates between groups over follow-up. It is not a ratio of median survival times and does not directly state the change in an individual participant’s probability of survival. A hazard ratio near 1 indicates that the estimated relative event rates are similar under the fitted model, while its confidence interval conveys uncertainty in that estimate.

The registry specifies a stratified Cox proportional hazards regression for the primary hazard ratio. Stratification allows the analysis to account for the strata used in the study’s analysis, but the stratification factors are not reported in the registry record. The proportional hazards model summarizes the relative hazards over time. Whether the proportional hazards assumption was checked, and how any departures would have been handled, is not reported in the registry record. The reported hazard ratio should therefore be understood as the sponsor’s stated model-based summary, not as proof that relative hazards were constant at every time.

Why report a Log Rank test as well?

General statistical context: The Log Rank test compares event experience between groups across follow-up while accounting for censoring. It produces a test of evidence against a no-difference null hypothesis, whereas the hazard ratio estimates the size and direction of a relative effect. The test result and the estimate answer related but different questions. A p-value alone does not quantify the size of an effect or its precision; the hazard ratio and confidence interval add that information.

For the primary result, the registry reports a hazard ratio of 1.083, a 95% confidence interval of 0.82 to 1.44, and p = 0.708. The interval includes 1, and the p-value is not statistically significant by conventional criteria. These findings do not establish superiority, and they should not be treated as proof of equivalence.

How do the secondary analyses differ?

The registry reports a Log Rank method for progression-free survival, with a hazard ratio and p-value. The response outcomes are percentages, rather than time-to-event outcomes, and the sponsor’s reported method is Cochran-Mantel-Haenszel. In general, a Cochran-Mantel-Haenszel analysis compares groups while accounting for specified strata. The strata used for these response analyses are not reported in the registry record.

For binary outcomes, the registry also contains unadjusted recalculations based on posted counts. Such calculations summarize the observed proportions without necessarily reproducing the sponsor’s stratified analysis. The sponsor’s own analysis is the relevant reported trial analysis; an unadjusted recalculation can be informative as a descriptive check, but it should not be substituted for the planned or reported analysis. The registry does not give enough detail to determine all covariates, strata, missing-outcome rules, or analysis-set definitions for the secondary outcomes.

General statistical context: The choice of method depends on the type of outcome and the way follow-up or grouping is handled. Time-to-event methods incorporate timing and censoring; percentage comparisons focus on whether a response was observed within the defined assessment window. These methods produce different estimands and should not be treated as interchangeable evidence. In particular, a response percentage does not answer the same question as overall survival, and agreement or disagreement among endpoints does not automatically validate one result over another.

11. Confidence Intervals and Precision

A confidence interval describes the range of estimates that are reasonably compatible with the observed data under the statistical method and its assumptions. It is not a guarantee that the true effect lies inside the interval, nor does it describe the range of outcomes for individual participants. A narrower interval indicates greater statistical precision; a wider interval indicates more uncertainty about the effect estimate.

For overall survival, the hazard ratio was 1.083, with a 95% confidence interval from 0.82 to 1.44. The interval includes 1 and covers values on both sides of 1. It therefore does not rule out either a lower or a higher relative hazard for the inotuzumab combination under the comparison convention. The median survival estimates were 9.5 months in both groups, but the confidence intervals around those medians—7.0 to 14.5 months and 7.7 to 14.1 months—show uncertainty in each group’s estimate.

Progression-free survival also has uncertainty around both the group medians and the hazard ratio. The reported hazard ratio was 0.924, with a 95% confidence interval from 0.72 to 1.19. As with the primary endpoint, that interval includes 1. A point estimate alone can appear to suggest a direction, but the interval is needed to understand how precisely that direction is estimated.

Response percentages have confidence intervals as well. These intervals show uncertainty in each group’s estimated response percentage; they are not, by themselves, a confidence interval for the difference between groups. The sponsor’s reported analyses for these outcomes give p-values but not an effect estimate or confidence interval for the between-group comparison in the registry record. Thus, the precision of the comparative response effect cannot be fully assessed from the sponsor-reported analysis alone.

Confidence intervals do not address every source of uncertainty. They reflect sampling uncertainty under the stated methods, while open-label conduct, outcome measurement, missing-data handling, and unreported analysis decisions may also matter. The registry record does not report enough information to quantify all such considerations.

12. Planning a Similar Trial: Sample Size

This is a planning illustration, not a re-evaluation of the original trial. It asks: if a new study wanted to detect a difference like the one observed, or a smaller one, how large would it need to be? It is deliberately not “observed power” of the finished trial, which is not a meaningful quantity.

The planning illustrations are calculations based on the observed effect, not the original trial design. For time-to-event outcomes, the figures are total events required, not participants per arm. For the other outcomes, the figures are participants per arm. The calculations assume a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. They consider detecting the stated fraction of the observed effect.

The 80% power column represents a hypothetical design that would have an 80% chance of detecting the specified effect under the stated assumptions, if that effect were the true one. The 90% power column represents a higher chance of detecting that same specified effect under those assumptions. These are design-planning concepts; they do not describe the chance that the trial’s result is correct.

For overall survival, planning for 100% of the observed effect gives 4,939 total events for 80% power and 6,611 total events for 90% power. Planning for 75% of the observed effect gives 8,780 total events for 80% power and 11,753 total events for 90% power. Planning for 50% of the observed effect gives 19,753 total events for 80% power and 26,444 total events for 90% power. The illustrations for progression-free survival and response outcomes use their respective observed effects and outcome-specific event or participant units.

Smaller effects are harder to distinguish from random variation, so planning for a smaller effect requires more events or participants under the same assumptions. The calculations are not observed power: they are hypothetical planning illustrations, not a measure of whether the completed trial was adequately powered or a retrospective probability that its result is true. The original sample-size justification and design assumptions are not reported in the registry record.

13. How the Sample Size Is Derived

Power is the probability that a study of a given size rejects the null hypothesis when a specific true effect exists. It depends on four things: the effect to detect, the variability of the outcome, the significance level (here a two-sided α of 0.05, so z_1−α/2 = 1.960) and the desired power (z_1−β = 0.842 for 80% and 1.282 for 90%).

Formula

d is the total number of events needed (Schoenfeld's formula) with 1:1 allocation and proportional hazards.

Worked calculation for this trial

1. Hazard ratio HR = 1.083, so ln(HR) = 0.0797.
2. d = 4 × (1.96 + 0.842)² / [ln(1.083)]² = 4,938.2, rounded up to 4,939 events in total for 80% power.

The smaller-effect rows in the table above repeat the same calculation with the effect multiplied by 0.75 and 0.5 (for ratios, applied to the logarithm of the ratio).

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
hr <- 1.083                              # observed hazard ratio
events_needed <- function(power, frac = 1) ceiling(round(4 * (z_a + qnorm(power))^2 / log(hr^frac)^2, 6))   # Schoenfeld, 1:1 allocation
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: events = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(events_needed(0.80, f), big.mark = ","), format(events_needed(0.90, f), big.mark = ",")))

# Simulation: given d total events under proportional hazards, events in the treatment arm are Binomial(d, hr / (1 + hr)).
# The score (log-rank type) statistic is (D1 - d / 2) / sqrt(d / 4).
set.seed(1232556)
d    <- events_needed(0.80)
reps <- 20000
D1 <- rbinom(reps, d, hr / (1 + hr))
z  <- (D1 - d / 2) / sqrt(d / 4)
cat(sprintf("Simulated power with %s events: %.3f (%d replicates)\n", format(d, big.mark = ","), mean(abs(z) > z_a), reps))

Output from running the code

100% of observed effect: events = 4,939 (80% power), 6,611 (90% power)
 75% of observed effect: events = 8,780 (80% power), 11,753 (90% power)
 50% of observed effect: events = 19,753 (80% power), 26,444 (90% power)
Simulated power with 4,939 events: 0.801 (20000 replicates)

The simulation uses the fact that, with d events in total and proportional hazards, the number of events in the treatment arm is binomial. It is an approximation that conditions on the total number of events, and it should land close to the target power.

14. Analyses Reported by the Sponsor

The registry’s primary result is the sponsor-reported analysis: a Log Rank comparison with a hazard ratio of 1.083, a 95% confidence interval from 0.82 to 1.44, and p = 0.708. The survival hazard ratio is not presented as an independent unadjusted recalculation; the registry describes it as the result from stratified Cox proportional hazards regression. The reported analysis is therefore the basis for interpreting the trial’s primary comparison.

For the response outcomes, the sponsor reports Cochran-Mantel-Haenszel p-values. The registry also contains simple unadjusted recalculations based on the posted group counts. Those recalculations answer a more limited descriptive comparison and may differ from a stratified analysis because they do not reproduce that analysis structure. They should be read as supplementary, not as replacements for the sponsor’s own result.

The registry does not report a discrepancy between the posted summaries and the sponsor’s analyses. Even so, the reporting detail differs by endpoint: the primary analysis includes a hazard ratio and interval, while the reported response analyses provide p-values without comparative effect estimates and intervals. The registry record does not explain all analysis details needed to recreate the sponsor’s methods independently.

15. Multiplicity, Missing Data and Analysis Populations

The registry lists overall survival as the primary endpoint and identifies multiple secondary outcomes, including progression-free survival, response measures, duration of response, health status, and health-related quality of life. Testing several outcomes creates a multiplicity issue: the more statistical comparisons made, the greater the opportunity for chance findings if each is interpreted in isolation. General statistical context: prespecified hierarchies, adjusted significance thresholds, or other multiplicity procedures can help control the interpretation of a family of tests.

Whether the trial used a multiplicity procedure, how outcomes were prioritized, and whether any gatekeeping sequence was planned are not reported in the registry record. The posted p-values therefore should not be assumed to have been adjusted, unless the record explicitly says so. The registry reports Cochran-Mantel-Haenszel p-values for the response outcomes and Log Rank p-values for time-to-event comparisons, but it does not give a full account of how these tests fit into a broader inferential plan.

The record identifies the primary analysis population as ITT. It does not provide a detailed description of the ITT definition, exclusions after randomization, or analysis populations for every secondary outcome. The primary endpoint definition explains censoring at the date of last contact or the end of the study. However, general handling of missing outcome assessments, loss to follow-up, deviations from assigned treatment, and sensitivity analyses is not reported in the registry record.

Censoring is part of a time-to-event analysis and is not the same as a complete description of missing-data methods. In general, survival methods rely on assumptions about how censoring relates to future event risk. The registry does not report diagnostic checks or sensitivity analyses for those assumptions. Similarly, the open-label design may affect subjective outcomes and assessment decisions, but the record does not describe steps taken to limit such effects.

These gaps do not invalidate the reported results, but they limit independent assessment of the full analysis strategy. The posted estimates should be read alongside the fact that the registry record does not contain the complete protocol or statistical analysis plan details needed to resolve these questions.

16. 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 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: disease progression (12.0%), pneumonia (4.9%), febrile neutropenia (4.2%), pyrexia (3.0%), asthenia (1.8%).

The registry lists serious adverse events among participants at risk in each group. For inotuzumab ozogamicin plus rituximab, 164 participants were at risk and 61 were affected. For investigator’s choice, 167 participants were at risk and 63 were affected. The other adverse-event category lists 155 affected among 164 at risk in the combination group and 158 among 167 at risk in the comparator group.

The most frequently listed serious terms across the posted results are disease progression at 12.0%, pneumonia at 4.9%, febrile neutropenia at 4.2%, pyrexia at 3.0%, asthenia at 1.8%, and dehydration at 1.8%. The registry does not assign these term percentages to a specific treatment group in the information, so they should not be treated as group-specific comparisons.

These counts and percentages describe reported events; they do not establish that one treatment caused an event or that the groups differ in safety. The record does not provide deaths in the adverse-event summary, nor does it report confidence intervals or a statistical comparison for these figures. The event collection timeframe and additional details about severity grading are not reported in the registry record.

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

The primary result describes the observed overall survival comparison in this randomized trial. Both groups had a reported median overall survival of 9.5 months. The sponsor-reported hazard ratio was 1.083, with a 95% confidence interval from 0.82 to 1.44 and p = 0.708. The estimate does not provide evidence that the combination was superior on the primary endpoint.

This conclusion is narrower than saying that the treatments are equal. A superiority test that does not reach statistical significance cannot establish equivalence, and the confidence interval still allows a range of effects. An equivalence or non-inferiority conclusion would require a suitable design and prespecified margins, which are not reported in the registry record. The trial’s stated aim was superiority, not a claim that any two outcomes within a particular range should be treated as the same.

The medians are summaries of estimated group survival, not predictions for individual participants. The hazard ratio is a model-based relative summary and is not a direct statement of how many months an individual might gain or lose. Neither statistic answers every question about tolerability, quality of life, or suitability for a particular person. The registry results alone are not a basis for treatment decisions.

The secondary results add context but do not replace the primary endpoint. Different outcomes measure different aspects of disease and treatment experience, and the multiplicity plan is not reported. The registry also lists the study as terminated without reporting why. That status cannot be explained by the posted survival estimate alone. Overall, the posted data support a careful statement about the measured comparison and its uncertainty, not a broader conclusion about all patients, all outcomes, or the reason the study ended.

18. Why This Trial Matters Statistically

This trial is statistically notable because it compared an investigational combination with active treatment choices in a randomized phase 3 design, rather than with no treatment. That makes the central question a comparative one: whether overall survival differed between the two assigned strategies in the study population. The distinction matters because an outcome in one group, considered alone, cannot establish the relative effect of the assigned treatments.

The primary result illustrates why a point estimate, a confidence interval, and a hypothesis test should be considered together. The hazard ratio is close to 1, its interval includes 1, and the reported p-value is not statistically significant. The same overall conclusion cannot be drawn from the medians alone; they are useful descriptive summaries but do not capture the full time-to-event comparison. The result does not demonstrate superiority, but neither does it establish equivalence.

The trial also offers a useful example of how secondary endpoints can broaden the clinical picture while complicating statistical interpretation. Survival, progression, response, and patient-reported outcomes are distinct measures. Multiple comparisons and incomplete reporting of the analysis plan make it important to distinguish confirmatory evidence from supportive or exploratory findings.

Finally, the registry provides enough detail to identify the primary endpoint, population, and core analysis methods, but not enough to reconstruct every design decision. Sample-size illustrations based on the observed effect are not the trial’s original justification. The original design assumptions, multiplicity strategy, missing-data approach, and reason for termination are not reported in the registry record. Recognizing those limits is part of interpreting the numerical result responsibly.

19. 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 registry record does not report the reason the study was terminated, so the termination status should not be attributed to the primary result or to any other specific event. The record also does not report the original sample-size justification, planned number of events, or original design assumptions. The planning calculations based on observed effects cannot fill that gap.

The study was open-label with no masking listed. This feature is relevant to interpretation, particularly for response assessments and patient-reported outcomes, although the registry does not report whether or how potential assessment bias was addressed. Overall survival is less dependent on subjective assessment than some secondary endpoints, but follow-up, censoring, and analysis choices still influence the estimate.

The primary analysis is described as ITT, with Kaplan-Meier estimation and a stratified Cox model. The strata, model diagnostics, proportional hazards checks, and detailed censoring procedures beyond the stated definition are not reported in the registry record. The record also does not provide a complete statistical analysis plan, so the planned treatment of multiplicity, missing data, and secondary endpoint testing cannot be independently assessed.

The primary confidence interval includes effects on either side of 1, leaving uncertainty about the direction and size of the comparative effect. The registry’s response analyses provide sponsor p-values without reported comparative effect intervals, limiting assessment of their precision. Some listed secondary outcome results are not reported in the registry record. Finally, the serious adverse-event summary lacks a reported group-specific breakdown for the listed serious terms, a statistical comparison, and details about the collection window. These limitations constrain interpretation but do not themselves establish that the reported analyses are incorrect.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: A Study Of Inotuzumab Ozogamicin Plus Rituximab For Relapsed/Refractory Aggressive Non-Hodgkin Lymphoma Patients Who Are Not Candidates For Intensive High-Dose Chemotherapy — NCT01232556.
• Registry record last updated 2019-01-08; status last verified 2018-12.
• Results first posted 2018-08-21.
• 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.

23. Record Summary

NCT01232556 was a randomized, open-label, parallel-group phase 3 study of inotuzumab ozogamicin plus rituximab versus investigator’s choice of rituximab plus gemcitabine or rituximab plus bendamustine in relapsed or refractory CD22-positive aggressive non-Hodgkin lymphoma among people not candidates for intensive high-dose chemotherapy. The registry lists enrollment of 338 and identifies overall survival as the primary endpoint in the ITT population.

Median overall survival was 9.5 months in each group. The sponsor-reported hazard ratio was 1.083, with a 95% confidence interval from 0.82 to 1.44 and p = 0.708. The posted primary result does not establish superiority, and it does not establish equivalence. The registry status is terminated, but the reason is not reported in the registry record. Details about the original sample-size justification, multiplicity procedures, and comprehensive missing-data methods are also not reported.