← Clinical Trials
Hepatocellular Carcinoma Phase 3 Completed NCT02435433

REACH-2: Complete Statistical Analysis of Ramucirumab in Hepatocellular Carcinoma

An independent statistical review of the randomized phase 3 REACH-2 trial evaluating ramucirumab plus best supportive care versus placebo plus best supportive care in participants with hepatocellular carcinoma and elevated baseline alpha-fetoprotein.

Trial start: 2015-07-20  ·  Primary completion: 2018-03-15  ·  Enrollment: 399
Scope of this record

This page separates reported trial results from statistical interpretation. The numerical results presented here are restricted to the statistical analyses from the ClinicalTrials.gov record. ClinicalTrials.gov provides the official trial registry record.

1. Trial at a Glance

REACH-2 was a randomized, parallel, double-masked phase 3 treatment trial comparing ramucirumab plus best supportive care (BSC) with placebo plus BSC in participants with hepatocellular carcinoma and elevated baseline alpha-fetoprotein. The registry reports overall survival as the primary endpoint and provides statistical analyses for overall survival, progression-free survival, time to radiographic progression, objective response rate, and time to deterioration of FHSI-8.

399
Enrolled
Phase 3
0.710
OS HR
95% CI 0.531–0.949
0.452
PFS HR
95% CI 0.339–0.603
4.6
ORR Odds Ratio
95% CI 0.6–37.3
FeatureREACH-2
Trial nameREACH-2
ClinicalTrials.gov identifierNCT02435433
PhasePhase 3
ConditionHepatocellular Carcinoma
PopulationParticipants with hepatocellular carcinoma and elevated baseline alpha-fetoprotein
AllocationRandomized
Design modelParallel
MaskingDouble
Primary purposeTreatment
Enrollment399.0
Arms listed in registry profile5
InterventionsRamucirumab; Placebo
Primary endpointOverall Survival (OS)
Primary endpoint typeTime-to-event
Hypothesis type reported for analysisSuperiority
Trial statusCompleted
Registry record: This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record.

2. Clinical Question

The central statistical question was whether ramucirumab plus best supportive care differed from placebo plus best supportive care with respect to overall survival in participants with hepatocellular carcinoma and elevated baseline alpha-fetoprotein. The registered primary endpoint was a time-to-event outcome measured from randomization to death from any cause.

Population

Participants with hepatocellular carcinoma and elevated baseline alpha-fetoprotein.

Intervention

Ramucirumab plus best supportive care (BSC).

Comparator

Placebo plus best supportive care (BSC).

Primary question

Does ramucirumab plus BSC produce a different overall-survival experience from placebo plus BSC under the trial's superiority framework?

3. Trial Design

01
Randomize399 enrolled
02
Double maskRandomized parallel design
03
TreatmentRamucirumab or placebo
04
Follow-upTime-to-event and response outcomes
05
AnalysisStratified statistical comparisons
Allocation
Randomized.
Design model
Parallel.
Masking
Double.
Primary purpose
Treatment.

The registry profile lists 5 arms, while the statistical comparisons posted on ClinicalTrials.gov for the primary and secondary analyses compare ramucirumab plus BSC with placebo plus BSC. The ClinicalTrials.gov record also identify three additional registry cohorts through the serious-adverse-event reporting: an open-label ramucirumab plus BSC cohort, a ramucirumab MEE cohort, and a placebo MEE cohort.

RANDOMIZED COMPARISON

Ramucirumab + BSC

  • Ramucirumab
  • Best supportive care
  • Primary and secondary efficacy analyses compare this group with placebo + BSC.
RANDOMIZED COMPARISON

Placebo + BSC

  • Placebo
  • Best supportive care
  • Primary and secondary efficacy analyses compare this group with ramucirumab + BSC.
Important distinction: the five-arm registry profile should not be interpreted as a five-group efficacy comparison. The statistical analyses posted on ClinicalTrials.gov for the efficacy outcomes explicitly define the comparison as ramucirumab + BSC versus placebo + BSC.

4. Endpoints

EndpointRegistry definition / time frameType
Overall Survival (OS) From Date of Randomization to Death from Any Cause (Up to 28 Months) Time-to-event
Progression Free Survival (PFS) From Randomization to Objective Progression or Death from Any Cause (Up to 28 Months) Time-to-event
Time to Radiographic Progression From Randomization to Objective Progression (Up to 28 Months) Time-to-event
Percentage of Participants With a Best Overall Response of Complete Response (CR) or Partial Response (PR): Objective Response Rate (ORR) From Randomization to Objective Progression (Up to 28 Months) Binary
Time to Deterioration of Functional Assessment of Cancer Therapy (FACT) Hepatobiliary Symptom Index-8 (FHSI-8) From Randomization to the First Date of Deterioration Observation (≥ 3-point decrease) (Up to 28 Months) Time-to-event

Overall survival definition

For OS, time was measured from the date of randomization to the date of death from any cause. Participants not known to have died on or before the data cut-off date were censored on the last date on or before the cut-off date on which they were known to be alive.

The registry therefore defines OS as a conventional right-censored time-to-event endpoint. Importantly, censoring does not mean that the participant is treated as having survived forever; it means that the available follow-up establishes survival only up to the participant's censoring date.

5. Statistical Methodology

Primary analysis: stratified log-rank test

The primary OS comparison was performed using a log-rank test, with the analysis text identifying the analysis as stratified. The ClinicalTrials.gov record does not specify the individual stratification factors, so no additional factors are inferred here.

The log-rank framework is designed for comparing survival distributions between groups while accounting for the timing of events and censoring. Rather than reducing each participant to a simple yes/no event indicator, it uses the longitudinal ordering of events and the number of participants at risk over follow-up.

Conceptual survival comparison
Observed events are compared with the events expected under the null hypothesis at successive event times.

For a time-to-event endpoint, the analysis uses information accumulated throughout follow-up rather than relying on a single fixed-time proportion.

Hazard ratio as the effect measure

The registry reports the treatment effect for OS, PFS, and time to radiographic progression as a hazard ratio (HR). A hazard ratio below 1 indicates a lower estimated instantaneous event rate in the ramucirumab group relative to the placebo group, within the framework used to estimate the reported HR.

Interpretation of a hazard ratio
HR < 1  →  lower estimated instantaneous event rate in the ramucirumab group

An HR is a relative time-to-event measure. It is not an absolute risk difference, a probability of benefit for an individual participant, or a statement that the same percentage reduction applies at every follow-up time.

Stratified analysis

The ClinicalTrials.gov record identifies stratified analysis for each of the reported comparisons. Stratification can preserve the structure of a randomized comparison when important baseline factors are incorporated into the analysis. The exact stratification variables are not included in the ClinicalTrials.gov record and therefore are not specified here.

Objective response rate: Cochran-Mantel-Haenszel test

ORR is a binary endpoint: participants are classified according to whether their best overall response was complete response or partial response. The registry reports a Cochran-Mantel-Haenszel test and an odds ratio for this comparison, again with a stratified analysis.

The Cochran-Mantel-Haenszel framework can combine information across strata while accounting for the stratified structure of a comparison. The odds ratio then summarizes the relative odds of the binary outcome between the two groups.

Analysis populations

The OS analysis used all randomized participants. The registry specifically states that censored participants remained included in the analysis. The ORR analysis used all randomized participants who received at least one dose of study drug. The FHSI-8 analysis used all randomized participants who had evaluable FHSI-8 data.

EndpointAnalysis population in the ClinicalTrials.gov recordMethod
Overall Survival All randomized participants; 50 censored in the ramucirumab arm and 21 in the placebo arm Log-rank test; stratified analysis
Progression Free Survival All randomized participants; 25 censored in the ramucirumab arm and 9 in the placebo arm Log-rank test; stratified analysis
Time to Radiographic Progression All randomized participants Log-rank test; stratified analysis
ORR All randomized participants who received at least one dose of study drug Cochran-Mantel-Haenszel test; stratified analysis
Time to FHSI-8 Deterioration All randomized participants who had evaluable FHSI-8 data Log-rank test; stratified analysis

6. Primary Result: Overall Survival

Overall survival was the single registered primary endpoint. The analysis compared ramucirumab + BSC with placebo + BSC from randomization to death from any cause, with follow-up defined in the registry as up to 28 months.

Hazard ratio for overall survival

0.710

95% CI: 0.531–0.949   ·   P = 0.0199

Two-sided 95% confidence interval  ·  Superiority hypothesis

Primary endpointComparisonEffect measureEstimate95% CIP-value
Overall Survival Ramucirumab + BSC vs Placebo + BSC Hazard ratio 0.710 0.531–0.949 0.0199
Clinical Biostats interpretation

The reported OS hazard ratio of 0.710 means that the estimated instantaneous rate of death was lower in the ramucirumab + BSC group than in the placebo + BSC group under the statistical framework represented by the reported hazard ratio. Expressed as a simple relative interpretation, 1 − 0.710 = 0.290, so the estimate corresponds to approximately a 29% lower estimated hazard.

The HR does not mean that 29% of participants benefited, that 29% of deaths were prevented, or that every participant experienced exactly a 29% reduction in mortality risk. It is a population-level relative time-to-event estimate.

The two-sided 95% confidence interval of 0.531–0.949 describes statistical uncertainty around the estimated HR. It does not describe the range of outcomes that an individual participant could experience. Because the interval remains below 1, the reported estimate is compatible with a lower hazard in the ramucirumab group throughout the confidence interval.

The P-value of 0.0199 addresses the statistical evidence against the null hypothesis under the specified analysis. It does not measure the magnitude or clinical importance of the effect. The HR and its confidence interval provide the effect estimate and its precision; the p-value addresses evidence against the null.

The analysis is based on censored time-to-event data, with 50 participants censored in the ramucirumab arm and 21 in the placebo arm. Censoring assumptions therefore form part of the interpretation. The ClinicalTrials.gov record identifies a stratified log-rank analysis but do not provide enough information to assess a proportional-hazards assumption or to determine the exact model used to generate the HR.

7. Secondary Result: Progression Free Survival

Progression-free survival was measured from randomization to objective progression or death from any cause, with the registry specifying a time frame of up to 28 months.

Hazard ratio for progression or death

0.452

95% CI: 0.339–0.603   ·   P < 0.0001

Two-sided 95% confidence interval  ·  Superiority hypothesis

EndpointComparisonEffect measureEstimate95% CIP-value
Progression Free Survival Ramucirumab + BSC vs Placebo + BSC Hazard ratio 0.452 0.339–0.603 <0.0001
Clinical Biostats interpretation

The PFS hazard ratio of 0.452 indicates a substantially lower estimated instantaneous rate of progression or death in the ramucirumab + BSC group relative to placebo + BSC under the reported time-to-event analysis. As a direct transformation of the reported estimate, 1 − 0.452 = 0.548, corresponding to approximately a 54.8% lower estimated hazard.

This does not mean that 54.8% of participants avoided progression, nor does it mean that each participant had exactly a 54.8% lower probability of progression. The endpoint combines objective progression and death, and the HR is a relative time-to-event measure.

The 95% CI of 0.339–0.603 quantifies uncertainty around the estimated HR. Its position below 1 indicates that the entire reported confidence interval corresponds to a lower estimated hazard for the ramucirumab group under the analysis.

The P-value < 0.0001 indicates strong statistical evidence against the null hypothesis under the reported test. It is not a measure of how large the treatment effect is; the HR and confidence interval answer that question more directly.

The analysis included all randomized participants, with 25 censored in the ramucirumab arm and 9 in the placebo arm. As with OS, interpretation depends on the treatment comparison remaining valid under the censoring and time-to-event analysis framework. The ClinicalTrials.gov record does not report a formal proportional-hazards assessment.

8. Secondary Result: Time to Radiographic Progression

Time to radiographic progression was defined from randomization to objective progression, with the registry specifying a time frame of up to 28 months.

Hazard ratio for radiographic progression

0.427

95% CI: 0.313–0.582   ·   P < 0.0001

Two-sided 95% confidence interval  ·  Superiority hypothesis

EndpointComparisonEffect measureEstimate95% CIP-value
Time to Radiographic Progression Ramucirumab + BSC vs Placebo + BSC Hazard ratio 0.427 0.313–0.582 <0.0001
Clinical Biostats interpretation

The reported hazard ratio of 0.427 corresponds to a lower estimated instantaneous rate of objective progression in the ramucirumab + BSC group relative to placebo + BSC. The simple complement, 1 − 0.427 = 0.573, corresponds to approximately a 57.3% lower estimated hazard.

This is not the same as saying that 57.3% of participants were protected from progression. It is a relative time-to-event estimate concerning the rate at which the defined event occurs over follow-up.

The two-sided 95% CI of 0.313–0.582 gives the statistical precision of the reported estimate. Because the full interval is below 1, the reported confidence interval is consistent with a lower estimated hazard of radiographic progression in the ramucirumab group.

The P-value < 0.0001 indicates strong evidence against the null hypothesis under the reported log-rank framework. It should not be interpreted as a percentage measure of effect size, probability that the treatment works, or probability that the result will replicate.

9. Secondary Result: Objective Response Rate

Objective response rate was defined as the percentage of participants whose best overall response was complete response or partial response, measured from randomization to objective progression over the registry-defined period of up to 28 months.

Odds ratio for objective response

4.6

95% CI: 0.6–37.3   ·   P = 0.1697

Two-sided 95% confidence interval  ·  Superiority hypothesis

EndpointComparisonEffect measureEstimate95% CIP-value
Objective Response Rate Ramucirumab + BSC vs Placebo + BSC Odds ratio 4.6 0.6–37.3 0.1697
Clinical Biostats interpretation

The reported OR of 4.6 means that the estimated odds of the defined objective response were 4.6 times as large in the ramucirumab + BSC group as in the placebo + BSC group under the reported stratified categorical analysis.

Odds are not probabilities. An odds ratio of 4.6 therefore cannot be translated into a statement that the response probability was 4.6 times higher. Converting an odds ratio into an absolute probability difference would require the underlying response proportions or baseline odds, which are not reported in the ClinicalTrials.gov record.

The 95% CI of 0.6–37.3 is wide. That width indicates substantial uncertainty around the point estimate. The interval includes 1, so the reported confidence interval also includes no difference in odds.

The P-value of 0.1697 does not provide evidence against the null hypothesis at conventional significance levels under this analysis. It also does not prove that the treatment groups have identical response rates. The appropriate conclusion from these the ClinicalTrials.gov record is that the registry's formal comparison produced an OR of 4.6 with considerable uncertainty and a P-value of 0.1697.

Why this result looks different from the survival analyses: OS, PFS, and time to radiographic progression preserve information about when events occur and are analyzed as time-to-event outcomes. ORR reduces response status to a binary endpoint, so the statistical effect measure is an odds ratio rather than a hazard ratio.

10. Secondary Result: Time to FHSI-8 Deterioration

The registry defines this endpoint as the time from randomization to the first date of deterioration observation, defined as a ≥ 3-point decrease in the Functional Assessment of Cancer Therapy Hepatobiliary Symptom Index-8 (FHSI-8), with a time frame of up to 28 months.

Hazard ratio for FHSI-8 deterioration

0.799

95% CI: 0.545–1.171   ·   P = 0.2382

Two-sided 95% confidence interval  ·  Superiority hypothesis

EndpointComparisonEffect measureEstimate95% CIP-value
Time to FHSI-8 Deterioration Ramucirumab + BSC vs Placebo + BSC Hazard ratio 0.799 0.545–1.171 0.2382
Clinical Biostats interpretation

The HR of 0.799 corresponds to a lower estimated instantaneous rate of the defined FHSI-8 deterioration event in the ramucirumab group relative to the placebo group. The simple complement, 1 − 0.799 = 0.201, corresponds to approximately a 20.1% lower estimated hazard.

The confidence interval of 0.545–1.171 is important because it crosses 1. Thus, the data are compatible with both a lower and a higher event hazard under the reported confidence interval, rather than establishing a single direction with high precision.

The P-value of 0.2382 does not provide statistical evidence against the null hypothesis under the reported log-rank analysis. It does not establish that the groups have identical symptom-deterioration experiences; it indicates that the observed comparison did not meet the conventional threshold for statistical evidence against the null.

The analysis population was restricted to randomized participants who had evaluable FHSI-8 data. That distinction matters because this endpoint does not use exactly the same analysis population definition as the primary OS analysis.

11. Consolidated Statistical Results

EndpointRoleMethodEffect measureEstimate95% CIP-value
Overall Survival Primary Log-rank; stratified Hazard ratio 0.710 0.531–0.949 0.0199
Progression Free Survival Secondary Log-rank; stratified Hazard ratio 0.452 0.339–0.603 <0.0001
Time to Radiographic Progression Secondary Log-rank; stratified Hazard ratio 0.427 0.313–0.582 <0.0001
Objective Response Rate Secondary Cochran-Mantel-Haenszel; stratified Odds ratio 4.6 0.6–37.3 0.1697
Time to FHSI-8 Deterioration Secondary Log-rank; stratified Hazard ratio 0.799 0.545–1.171 0.2382

This table shows why a clinical-trial statistical summary should not reduce the evidence to a single p-value. The five reported comparisons use different endpoints, analysis populations, and effect measures. The OS result is the registered primary endpoint; the remaining four analyses are secondary endpoints.

12. Censoring and Analysis Populations

Time-to-event analyses require a mechanism for handling participants who have not experienced the event by the end of their available follow-up. The registry-reported OS definition explicitly states that participants not known to have died by the data cut-off were censored at the last date on or before the cut-off when they were known to be alive.

AnalysisRamucirumab censoredPlacebo censoredPopulation
Overall Survival 50 21 All randomized participants
Progression Free Survival 25 9 All randomized participants

The fact that censored participants remain in the analysis is fundamental to the interpretation of Kaplan-Meier and log-rank methods. A censored participant contributes information for the period during which their event-free status is known, rather than being treated as if an event had occurred.

Population differences matter: the OS and PFS analyses include all randomized participants, whereas ORR requires at least one dose of study drug and the FHSI-8 analysis requires evaluable FHSI-8 data. Estimates from these endpoints therefore should not be interpreted as though every analysis used an identical population.

13. Statistical Methods Explained

Why was a log-rank test used for overall survival?

Overall survival is a time-to-event endpoint, so both the occurrence of death and the timing of death are relevant. The log-rank test is designed to compare survival experience between randomized groups while incorporating the information contributed by participants who are censored during follow-up.

What does an OS hazard ratio of 0.710 mean?

An HR of 0.710 is a relative measure of the estimated event rate over time. A value below 1 indicates a lower estimated instantaneous rate of death in the ramucirumab group under the reported analysis. It does not mean that exactly 71.0% of participants survive, nor does it represent a 29-percentage-point difference in survival probability.

Why is the confidence interval important?

A point estimate alone hides uncertainty. For OS, the reported HR is 0.710, but the 95% CI extends from 0.531 to 0.949. The interval gives a range of values compatible with the statistical uncertainty represented by the analysis. For the FHSI-8 endpoint, the CI of 0.545–1.171 crosses 1, illustrating why a point estimate below 1 does not automatically establish a statistically supported treatment difference.

Why does the p-value not measure effect size?

A p-value evaluates evidence against a null hypothesis under a specified statistical model and sampling framework. It does not tell us how large the treatment effect is. For example, the PFS analysis has a P-value <0.0001, but the size of the estimated treatment effect is communicated by the HR of 0.452 and its confidence interval of 0.339–0.603.

What does an odds ratio of 4.6 mean for ORR?

An OR of 4.6 means that the estimated odds of achieving the defined objective response were 4.6 times the odds in the placebo group under the reported stratified analysis. Odds are different from probabilities, so the OR cannot be read as a 4.6-fold increase in the percentage of responders without knowing the underlying response rates.

Why is stratification important?

The registry identifies each reported comparative analysis as stratified. Stratified methods account for the structure created by dividing observations into strata before combining the evidence. Because the ClinicalTrials.gov record does not specify the actual stratification variables, the interpretation should stop at the documented fact that the analyses were stratified rather than assuming particular factors.

Why should OS and PFS not be treated as interchangeable?

OS records death from any cause, whereas PFS records objective progression or death from any cause. They are therefore different clinical and statistical endpoints. A treatment effect on PFS does not automatically specify the magnitude of its effect on OS, which is why each endpoint requires its own analysis.

14. Primary Endpoint Versus Secondary Endpoints

The registry identifies one primary endpoint: overall survival. Four additional statistical analyses are posted on ClinicalTrials.gov for secondary endpoints. This hierarchy matters when interpreting the collection of p-values because primary and secondary endpoints can have different roles in a trial's prespecified statistical strategy.

EndpointRoleEndpoint typeEffect measure
Overall SurvivalPrimaryTime-to-eventHazard ratio
Progression Free SurvivalSecondaryTime-to-eventHazard ratio
Time to Radiographic ProgressionSecondaryTime-to-eventHazard ratio
Objective Response RateSecondaryBinaryOdds ratio
Time to FHSI-8 DeteriorationSecondaryTime-to-eventHazard ratio
Multiplicity: the ClinicalTrials.gov record identifies the hypothesis type as superiority but do not provide an alpha-allocation strategy, multiplicity-adjustment procedure, or endpoint-testing hierarchy beyond identifying OS as the single registered primary endpoint. Those features should therefore not be inferred from the reported p-values.

15. What the Hazard Ratios Do — and Do Not — Mean

Relative effect, not absolute effect

The three reported hazard ratios for OS, PFS, and time to radiographic progression are all below 1: 0.710, 0.452, and 0.427, respectively. Each indicates a lower estimated instantaneous event rate in the ramucirumab group under its corresponding time-to-event analysis.

These estimates cannot be interpreted as percentages of participants who benefit, absolute reductions in event probability, or guarantees about individual outcomes.

Confidence intervals provide context

The OS confidence interval is 0.531–0.949; the PFS confidence interval is 0.339–0.603; and the time-to-radiographic-progression confidence interval is 0.313–0.582. These intervals describe uncertainty around the corresponding estimates rather than variation among individual patients.

The FHSI-8 result illustrates uncertainty

The FHSI-8 deterioration HR is 0.799 with a 95% CI of 0.545–1.171. Although the point estimate is below 1, the confidence interval crosses 1. This is a useful reminder that the point estimate and the uncertainty interval must be interpreted together.

16. Understanding the ORR Odds Ratio

The ORR analysis is statistically different from the time-to-event analyses because each participant is classified according to whether the best overall response was complete response or partial response. The reported effect measure is an odds ratio of 4.6.

Odds and probabilities
Odds = p / (1 − p)

The odds ratio compares odds between groups. It should not be treated as a direct ratio of response probabilities, particularly when response is not rare.

The wide 95% confidence interval of 0.6–37.3 is particularly important. A wide interval means that the point estimate of 4.6 is not highly precise. The lower end is below 1 while the upper end is substantially above 1, so the registry-reported analysis leaves considerable uncertainty about the magnitude of the relative odds.

The P-value of 0.1697 should likewise be interpreted in conjunction with the confidence interval. The appropriate statistical statement is not that the OR is "proven" to be 4.6, but that 4.6 is the reported point estimate accompanied by a wide interval and a P-value of 0.1697.

17. Safety Results

The ClinicalTrials.gov record reports serious adverse events by arm as the number affected over the number at risk. These data include the randomized comparison as well as three additional cohorts identified in the registry profile.

Arm / cohortSerious adverse events affectedAt risk
Ramucirumab + BSC72197
Placebo + BSC2795
Open Label Ramucirumab + BSC1847
Ramucirumab MEE Cohort1039
Placebo MEE Cohort321
Serious adverse events: affected participants
Ramucirumab + BSC
72
Placebo + BSC
27
Open Label Ramucirumab
18
Ramucirumab MEE
10
Placebo MEE
3

The denominators matter. The reported serious-adverse-event counts are not directly comparable as simple counts because the numbers at risk differ across arms and cohorts. The ClinicalTrials.gov record does not provide a formal statistical comparison of serious adverse-event rates, so this page does not construct one.

Safety interpretation: the randomized ramucirumab + BSC and placebo + BSC groups can be described separately from the open-label and MEE cohorts. The additional cohorts should not be folded into the randomized efficacy comparison or treated as if they were additional randomized control groups.

18. Trial Timeline

2015-07-20

Trial start

The registry profile gives 2015-07-20 as the study start date.

2018-03-15

Primary completion

The registry profile gives 2018-03-15 as the primary completion date.

Completed

Registry status

The trial is listed as completed, with results posted and 10 outcome measures and 5 statistical analyses reported in the ClinicalTrials.gov record.

19. A Statistical Reading of the Complete Results

Viewed together, the ClinicalTrials.gov record shows a consistent direction for the three reported disease time-to-event endpoints: the estimated hazard ratios for OS, PFS, and time to radiographic progression are all below 1. The FHSI-8 deterioration estimate is also below 1, but its confidence interval crosses 1. ORR uses a different effect measure and has a wide confidence interval.

EndpointDirection of point estimateDoes 95% CI cross 1?Statistical interpretation
Overall Survival HR below 1 No Reported primary analysis provides evidence against the null under the stated framework.
Progression Free Survival HR below 1 No Reported secondary analysis provides strong evidence against the null under the stated framework.
Time to Radiographic Progression HR below 1 No Reported secondary analysis provides strong evidence against the null under the stated framework.
Objective Response Rate OR above 1 Yes Point estimate favors higher odds of response, but the confidence interval is wide and includes 1.
Time to FHSI-8 Deterioration HR below 1 Yes Point estimate favors a lower deterioration hazard, but the confidence interval includes 1.

This pattern illustrates why statistical interpretation should distinguish direction, magnitude, precision, and hypothesis-test evidence. A point estimate below 1 is not equivalent to statistical significance, and a statistically supported result does not by itself communicate the absolute clinical effect.

20. Missing Statistical Information and What Should Not Be Inferred

The ClinicalTrials.gov record is sufficient to describe the principal statistical comparisons, but they do not contain every component of a complete statistical analysis plan. In particular, the ClinicalTrials.gov record does not specify the exact stratification factors, alpha-allocation procedure, multiplicity adjustment, interim-analysis strategy, missing-data or imputation methodology, or a formal proportional-hazards assessment.

What is documented

Randomization, double masking, parallel design, superiority hypotheses, stratified log-rank analyses, a stratified Cochran-Mantel-Haenszel analysis, hazard ratios, odds ratios, confidence intervals, p-values, and defined analysis populations.

What is not specified here

The ClinicalTrials.gov record does not identify the individual stratification factors or provide detailed multiplicity, interim-analysis, imputation, or proportional-hazards procedures.

This distinction is important in an educational statistical record. Absence of a detail in the ClinicalTrials.gov record is not a reason to reconstruct it from assumptions or from general knowledge of the disease area.

21. Important Limitations and Interpretation Issues

22. Why This Trial Matters Statistically

REACH-2 provides a useful teaching example because the ClinicalTrials.gov record connect several fundamental clinical-trial methods within one analysis record: randomized comparison, double masking, time-to-event endpoints, censoring, stratified log-rank testing, hazard ratios, confidence intervals, categorical response analysis, odds ratios, and distinct analysis populations.

ConceptHow it appears in REACH-2
RandomizationThe trial allocation is randomized.
Double maskingThe registry profile identifies the study as double masked.
Time-to-event endpointsOS, PFS, time to radiographic progression, and time to FHSI-8 deterioration.
Log-rank testUsed for the reported time-to-event comparisons.
Stratified analysisIdentified in the analysis records for the time-to-event and ORR comparisons.
Hazard ratioReported for OS, PFS, time to radiographic progression, and FHSI-8 deterioration.
Confidence intervalsTwo-sided 95% intervals are reported for all five registry-reported statistical analyses.
Cochran-Mantel-Haenszel testUsed for the reported ORR comparison.
Odds ratioReported for objective response rate.
Analysis populationsDifferent endpoint-specific populations are documented.
CensoringExplicitly documented for the OS and PFS analyses.

23. Statistical Methods Explained: A Practical Reading of REACH-2

Start with the endpoint

Ask whether the outcome is a time-to-event measure or a binary response measure. That determines why REACH-2 uses hazard ratios for several endpoints but an odds ratio for ORR.

Then identify the population

Check whether all randomized participants were analyzed or whether the endpoint required treatment exposure or evaluable measurements.

Then read the estimate

For an HR, values below 1 indicate a lower estimated event rate in the ramucirumab group. For an OR, values above 1 indicate higher estimated odds of the defined response.

Then read the interval

The confidence interval tells you how precisely the effect was estimated. Wide intervals signal greater uncertainty than narrow intervals.

24. What the Statistical Results Support

The ClinicalTrials.gov record supports several precise statistical statements. The primary OS analysis reported an HR of 0.710 with a two-sided 95% CI of 0.531–0.949 and P = 0.0199. The secondary PFS analysis reported an HR of 0.452 with a 95% CI of 0.339–0.603 and P < 0.0001. Time to radiographic progression reported an HR of 0.427 with a 95% CI of 0.313–0.582 and P < 0.0001.

The ORR analysis reported an odds ratio of 4.6, but its 95% CI was 0.6–37.3 and P = 0.1697. The FHSI-8 deterioration analysis reported an HR of 0.799, with a 95% CI of 0.545–1.171 and P = 0.2382.

These findings should be kept endpoint-specific. The OS result is the registered primary analysis; the other results provide secondary statistical evidence about progression, radiographic progression, response, and symptom deterioration. They should not be collapsed into one composite treatment-effect number.

25. Related Tutorials

Learn more about the methods used in this trial:

26. Related Statistical Calculators

27. Sources

The statistical values and endpoint definitions on this page are taken from the ClinicalTrials.gov record. The PubMed links are provided as the linked publication records associated with the trial data; no additional numerical results from those publications are incorporated into this page.

Continue through Clinical Biostats

Explore statistical tutorials and calculators covering the methods used to interpret randomized clinical trials such as REACH-2.

28. Record Summary

REACH-2 provides a compact example of how several core clinical-trial statistical methods work together. The trial was randomized, parallel, double masked, and phase 3, with 399 participants enrolled. Overall survival was the registered primary time-to-event endpoint. Its reported stratified log-rank analysis produced an HR of 0.710 with a two-sided 95% CI of 0.531–0.949 and P = 0.0199.

The secondary analyses extend the statistical picture. PFS produced an HR of 0.452, time to radiographic progression produced an HR of 0.427, ORR produced an OR of 4.6, and time to FHSI-8 deterioration produced an HR of 0.799. The confidence intervals and p-values show that these estimates differ substantially in precision and statistical evidence.

The most important statistical lesson is that these results should be read as a collection of endpoint-specific analyses rather than as one number. Hazard ratios describe relative time-to-event effects, odds ratios describe relative odds for a binary outcome, confidence intervals communicate precision, and p-values address evidence against a null hypothesis. Analysis populations and censoring rules further determine what each estimate actually represents.

Clinical Biostats methodology: A trial-results page should not merely repeat a registry record. The goal is to reconstruct the statistical story of the trial while clearly separating reported numerical evidence from educational interpretation and avoiding unsupported assumptions about methods that are not documented in the ClinicalTrials.gov record.