Oxaliplatin-Based Chemotherapy and Chemoradiotherapy or Chemoradiotherapy in Esophageal or Gastroesophageal Carcinoma: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Oxaliplatin-Based Chemotherapy and Chemoradiotherapy or Chemoradiotherapy in Esophageal or Gastroesophageal Carcinoma (NCT00525915), a randomized phase 2 study. The primary outcome is Pathologic Complete Response Rate.

2. Clinical Question

The comparison is between treatment strategies, not simply between receiving and not receiving chemotherapy or radiation: both arms included chemoradiotherapy and surgery, while Arm B added induction chemotherapy beforehand. The outcome is based on examination of the surgical specimen. Thus, the reported response rates describe participants who reached surgery and had their specimen assessed. They should not automatically be read as response rates for every randomized participant.

The registry describes additional objectives, including survival, resection, safety, relapse, and exploratory marker questions. It does not post results for those secondary objectives in the record. The primary posted result therefore addresses a specific pathological response question, rather than the full range of outcomes named in the study objectives.

3. Trial Design

The registry record identifies this as a randomized phase II study with no masking and a parallel design. It was conducted for people with resectable esophageal or gastroesophageal carcinoma. The record lists enrollment of 126 and describes M.D. Anderson Cancer Center as the sponsor. The study start is recorded as 2005-04, primary completion as 2013-03, and status as completed.

Arm A received 5-Fluorouracil and oxaliplatin with weekday radiation for 5 weeks, followed by surgery. Arm B received preoperative chemotherapy with 5-Fluorouracil and oxaliplatin, followed by the chemoradiotherapy schedule used in Arm A and then surgery. The record describes the Arm B induction phase as 8 weeks, and notes that its overall schedule was 8 weeks longer than Arm A. These schedules make the randomized comparison a contrast between two sequences of multimodal treatment.

Random allocation is intended to create groups that are comparable on average at the point of assignment, though it cannot guarantee that every measured or unmeasured characteristic will be balanced in a particular trial. The open-label design means masking was not used. Pathological assessment of a surgical specimen is a defined outcome, but the record does not report whether pathologists were blinded to treatment assignment.

The posted pathologic complete response assessment concerns participants who had surgery: 55 in Arm A and 54 in Arm B. Those are the reported outcome group sizes, not a statement about the total number randomized to each arm. Safety fields use a different at-risk count: 63 in each arm. The registry record does not explain the difference between these assessment populations, nor does it describe exclusions or participant flow in sufficient detail to reconcile them.

The primary completion date indicates when primary outcome collection was recorded as complete; it does not by itself explain how long every participant was followed. The registry record names survival, R0 resection, safety, relapse, and exploratory correlative studies as secondary objectives, but secondary results are not posted in the outcome data covered. The trial's original sample-size justification and detailed statistical analysis plan are not reported in the registry record.

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: Participants who had surgery were reported for the pathCR rate assessment in this outcome.

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: Of the 126 randomized, 2 participants in Arm A declined therapy after randomization; and 13 (6 in Arm A and 7 in Arm B) experienced treatment failure before surgery leaving 55 in Arm A participants and 54 in Arm B who underwent surgery.

6. Primary Endpoint

The registry lists the primary outcome as Pathologic Complete Response Rate, measured over Surgery post chemotherapy (approximately 10-11 weeks).

The primary endpoint is the pathologic complete response rate: the percentage of participants whose surgical specimen showed no residual cancer after preoperative treatment. The registry describes this assessment as occurring at surgery after chemotherapy, with a timeframe of approximately 10-11 weeks. It also describes the response categories as pathologic complete response, less than 50% residual cells, and greater than 50% residual cells.

Pathological response is used because it evaluates the tumor in tissue removed at surgery, rather than relying only on symptoms or an imaging-based assessment. It can describe how much cancer remains in the specimen after preoperative treatment. In this trial, it is a direct endpoint for comparing the response associated with the two treatment sequences.

The posted outcome is restricted to participants who had surgery. That restriction matters for interpretation: the reported percentage is not necessarily the percentage among all randomized participants, and the record does not explain why some enrolled participants are absent from the surgical assessment groups. The trial objectives also mention comparing the proportion with less than 50% residual cancer, but that separate category is not posted as an outcome result in the record. The available primary result therefore speaks to complete pathological response, not the full distribution of residual disease or every clinical outcome.

As a general point, a binary endpoint records whether a defined event occurred, rather than the degree or timing of change on a continuous scale. The response rate is easy to interpret as a proportion, but it does not by itself describe durability, survival, relapse, or treatment burden. Those questions require their own outcome data and analyses.

7. Results Posted in the Registry

• Arm B: Pre-Op Chemo + Chemo With Radiation Treatment: 25.9% (14/54) versus 12.7% (7/55) on control; risk difference 13.2 percentage points (95% CI -1.4 to 27.8), risk ratio 2.04 (0.89–4.65); unadjusted two-proportion z-test p = 0.081.

8. Statistical Interpretation

9. Statistical Methods Explained

The posted comparison is a binary-outcome comparison: each assessed participant either had a pathologic complete response or did not. The registry gives the response rates and the assessment group sizes, while the comparison summaries are unadjusted recalculations from those posted counts. The registry record does not report a sponsor-authored inferential analysis for this outcome, an adjusted model, or the original statistical analysis plan.

What does the risk difference describe?

The risk difference is the response proportion in Arm B minus the response proportion in Arm A, expressed on the probability scale. The recalculated estimate is 0.132, with a 95% confidence interval from -0.014 to 0.278. A positive value indicates a higher estimated response proportion in Arm B; a negative value indicates a higher estimated proportion in Arm A. The interval includes values on both sides of no difference, so the data do not pin down the direction with precision. The estimate is unadjusted: it does not account for any participant characteristics or other factors beyond the two posted groups.

How do risk ratios and odds ratios differ?

A risk ratio compares the response proportion in one arm with the proportion in the other. An odds ratio compares the odds of response, rather than the proportions themselves. These measures use different scales and can have different numerical values, especially when an outcome is not rare. The recalculated risk ratio is 2.037, with a 95% confidence interval from 0.892 to 4.653. The recalculated odds ratio is 2.4, with a 95% confidence interval from 0.883 to 6.521. Both intervals include the null value for their respective measures. Neither relative measure removes the uncertainty apparent in the risk-difference interval.

What does a confidence interval convey?

In general, a confidence interval is a range produced by a statistical procedure that reflects sampling uncertainty under its assumptions. It is not a guarantee that the true effect lies in the reported range, and it does not account automatically for bias, incomplete follow-up, outcome misclassification, or choices made after seeing the data. A wide interval signals that the estimate is compatible with a broad range of effects. Here, the interval around the risk difference spans a small value favoring Arm A as well as values favoring Arm B, so the estimate is not highly precise.

How should the p-value be read?

The unadjusted p-value is 0.081. In general, a p-value measures the compatibility of observed data, or more extreme data, with a null hypothesis under the statistical model. It is not the probability that the null hypothesis is true, and it is not a measure of clinical importance. Its interpretation depends on the test and assumptions. The registry record does not identify a sponsor analysis or full inferential procedure that would let readers assess additional modeling choices. The p-value should therefore be read alongside the effect estimate and its confidence interval, rather than as a standalone decision rule.

Why does the analysis population matter?

The outcome was reported among participants who had surgery. Randomization applies at assignment, but restricting an outcome analysis to participants who undergo surgery can make the assessed groups differ from the original randomized groups if the likelihood of reaching surgery is related to prognosis or treatment. The registry record does not report the reasons for nonassessment, a missing-data approach, or an intention-to-treat analysis for this endpoint. Consequently, the result describes the recorded surgical assessment groups; it cannot automatically be generalized to everyone randomized.

General statistical practice would distinguish the randomized population, the treated population, and the population with an evaluable endpoint, and would explain how missing outcomes were handled. These definitions and procedures are not reported in the registry record. The posted comparison also does not establish whether the outcome assessment was blinded. A pathological specimen is a concrete endpoint, but the record does not give enough detail to evaluate possible assessment bias.

The trial had multiple stated objectives, including several secondary outcomes and exploratory studies. The registry record does not report whether or how multiplicity was controlled, whether interim analyses were conducted, or whether the primary comparison was adjusted for any covariates. These omissions are not evidence that such procedures did or did not occur; they mean they cannot be assessed from the registry record.

10. Confidence Intervals and Precision

The main estimate is the recalculated risk difference of 0.132, with a 95% confidence interval from -0.014 to 0.278. The interval extends from a small negative difference to a larger positive difference, so the estimated contrast is imprecise about both direction and magnitude. It includes no difference. The result should not be reduced to a statement that one arm was definitively better or that the strategies were the same.

The relative-effect intervals also span the null value: the risk ratio is 2.037, with a 95% confidence interval from 0.892 to 4.653, and the odds ratio is 2.4, with a 95% confidence interval from 0.883 to 6.521. These ranges are broad. Relative measures can appear large while remaining uncertain, particularly when response counts are limited. Looking across the absolute and relative scales gives a fuller view of the uncertainty than relying on a single point estimate.

Precision is affected by the amount of information in the endpoint data. The response assessment groups include 55 participants in Arm A and 54 in Arm B, and the posted response counts are 7 and 14. The record does not provide a confidence interval for each arm's response rate, but it does provide intervals for the recalculated comparisons. No additional precision should be inferred from the rounded arm rates alone.

A confidence interval represents sampling uncertainty under the assumptions of its calculation. It does not address the potential consequences of the surgical-only assessment population, missing outcomes, unreported analysis choices, or possible measurement bias. The registry record does not describe those issues in enough detail to quantify their impact. Thus, the interval is one component of uncertainty, not a complete accounting of it.

11. 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 figures are participants per arm, not total participants across the study and not total events. They are illustrations based on the observed effect, using two-sided significance level 0.05, equal allocation between the arms, and a normal approximation. The calculation considers detecting the stated fraction of the observed effect. The original trial's own sample-size justification and design assumptions are not reported in the registry record.

For planning based on 100% of the observed effect, the figures are 140 participants per arm for 80% power and 186 participants per arm for 90% power. For planning based on 75% of the observed effect, they are 232 participants per arm for 80% power and 311 participants per arm for 90% power. For planning based on 50% of the observed effect, they are 484 participants per arm for 80% power and 648 participants per arm for 90% power.

In general, power is the probability that a planned study will detect an effect of a specified size under the assumed model and significance threshold, if that effect is present. The 80% and 90% columns represent different target probabilities of detection under those planning assumptions. The higher target requires more participants in each scenario. Smaller assumed effects also require more participants, because they are harder to distinguish from random variation.

These are planning illustrations, not observed power. They do not describe the probability that this completed trial's result is true, and they do not show what power the trial actually had under its original design. The curve should be read as a planning aid linked to the assumed effect size, not as an additional trial outcome.

12. 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

n is the number of participants in each arm, p₀ and p₁ are the control and treatment proportions, and p̄ is their average.

Worked calculation for this trial

1. Control proportion p₀ = 0.127; treatment proportion p₁ = 0.259; average p̄ = 0.193.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.5584 = 1.0945.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.5506 = 0.4634.
4. n = (1.0945 + 0.4634)² / (0.1320)² = 139.3, rounded up to 140 participants per arm 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
p0 <- 0.127272727272727                          # control proportion
p1 <- 0.259259259259259                          # treatment proportion
n_per_arm <- function(power, frac = 1) {
  q1 <- p0 + (p1 - p0) * frac               # smaller effect = fraction of the observed difference
  pbar <- (p0 + q1) / 2
  z_b <- qnorm(power)
  ceiling(round((z_a * sqrt(2 * pbar * (1 - pbar)) + z_b * sqrt(p0 * (1 - p0) + q1 * (1 - q1)))^2 / (q1 - p0)^2, 6))
}
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: n per arm = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(n_per_arm(0.80, f), big.mark = ","), format(n_per_arm(0.90, f), big.mark = ",")))

# Check against R's built-in function (unrounded n per arm)
cat("power.prop.test:", round(power.prop.test(p1 = p0, p2 = p1, power = 0.80)$n, 1), "\n")

# Simulation: how often does the z-test reject at the 80%-power sample size?
set.seed(525915)
n    <- n_per_arm(0.80)
reps <- 20000
x0 <- rbinom(reps, n, p0); x1 <- rbinom(reps, n, p1)
ph <- (x0 + x1) / (2 * n)
z  <- ((x1 - x0) / n) / sqrt(ph * (1 - ph) * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(z) > z_a, na.rm = TRUE), reps))

Output from running the code

100% of observed effect: n per arm = 140 (80% power), 186 (90% power)
 75% of observed effect: n per arm = 232 (80% power), 311 (90% power)
 50% of observed effect: n per arm = 484 (80% power), 648 (90% power)
power.prop.test: 139.3 
Simulated power at n = 140 per arm: 0.801 (20000 replicates)

The simulation draws many pairs of event counts at the planned sample size and counts how often the two-proportion z-test rejects the null. It should land close to the target power; at very small required sample sizes it can fall below it, because the normal approximation is rough there.

13. Multiplicity, Missing Data and Analysis Populations

The registry describes a primary objective that includes pathologic complete response and the proportion of participants with less than 50% residual cancer. The posted primary outcome result covers pathologic complete response. It does not report the second response category comparison as a separate result. The listed secondary objectives include survival at specified follow-up points, median survival time, R0 resection, safety, local and systemic relapse, exploratory correlative studies, and possible interactive effects involving radiation type and treatment arm. Secondary results are not posted in the registry record.

When a study asks multiple questions, each additional test can increase the chance of at least one apparently notable result arising by chance. General approaches to this issue include defining a primary comparison in advance, adjusting significance thresholds, or treating secondary analyses as exploratory. The registry record does not report whether multiplicity was controlled, how outcomes were prioritized in the analysis, or whether the study used a hierarchical testing procedure. The absence of this information means the multiplicity strategy cannot be evaluated; it does not establish that no strategy was used.

The registry record also does not report interim analyses, stopping rules, or the original statistical analysis plan. It does not specify whether the primary comparison was adjusted for baseline characteristics, whether there were prespecified subgroup analyses, or how any model diagnostics were considered. These details affect how confidently an inferential result can be interpreted, particularly in a trial with several stated aims.

The pathologic response assessment population consists of participants who had surgery, while the safety fields list 63 at risk in each arm. The record does not explain the difference between these populations or state how participants without an assessable surgical specimen were handled. It does not report a missing-data method, an intention-to-treat analysis for the primary endpoint, or reasons for absent assessments. A surgical-only result may be vulnerable to selection if reaching surgery is related to treatment or prognosis. The size and direction of any such effect cannot be determined from the registry record.

As a general principle, a transparent report would identify the analysis population, missing-data approach, prespecified outcomes, and multiplicity plan. Those details are not reported in the registry record, limiting assessment of how the posted comparison relates to the full randomized population and the study's several objectives.

14. 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: nausea/vomiting (4.8%), dehydration (3.2%), fever (3.2%), pulmonary embolism (3.2%), atrial fibrillation (3.2%).

The registry safety fields list 63 participants at risk in each arm. Serious adverse events are recorded for 10 in Arm A and 7 in Arm B. These are group-level counts, not event rates recalculated for a common follow-up duration. The record does not report the timing, duration, or person-time underlying these fields, so the counts should not be interpreted as comparative rates adjusted for exposure.

The listed serious adverse event terms include nausea/vomiting at 4.8%, and dehydration, fever, pulmonary embolism, atrial fibrillation, and dyspnea at 3.2% each. These are the registry's displayed percentages for the terms; the record does not provide further detail about event severity grading or whether participants could contribute to more than one term. The terms should not be added together to infer a total.

The other adverse event fields list 63 affected among 63 at risk in each arm. The registry record does not provide a detailed breakdown for those other events in the data. Death counts are not reported in the registry record. Differences in counts alone do not establish a treatment-related safety difference, particularly without information about exposure, event ascertainment, timing, and statistical comparison.

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

The posted data show a higher recorded pathologic complete response rate in Arm B than in Arm A among participants who had surgery: 26.0 in Arm B and 13.0 in Arm A. The recalculated risk-difference estimate is 0.132. Its 95% confidence interval, from -0.014 to 0.278, includes no difference and permits effects in either direction. The unadjusted p-value is 0.081. Taken together, these figures indicate uncertainty around the size and direction of the contrast, rather than a precise estimate.

This result does not show that either treatment strategy improves survival, reduces relapse, increases the chance of resection, or changes quality of life. Those outcomes are not posted in the registry record. It also does not report the proportion with less than 50% residual cancer, despite that category appearing in the stated objective. The primary result is about complete response in a surgical specimen, not every aspect of disease status or treatment experience.

The result does not establish equivalence between the strategies. A confidence interval that includes no difference can reflect uncertainty as well as a genuinely small effect; it is not proof that the treatment strategies have the same effect. Conversely, the point estimate alone does not establish a reliable advantage for Arm B. The sample-size illustrations are not observed power and should not be used to make claims about the certainty of this completed trial.

The outcome groups were participants who had surgery, and the registry does not explain the difference between those group sizes and the safety at-risk counts. If participants without a surgical assessment differed from those assessed, the reported response proportions may not represent the full randomized groups. The registry record does not provide enough detail to evaluate that possibility.

In general, randomized comparisons can support causal interpretation when the analysis preserves the randomized groups and the outcome is collected and analyzed appropriately. Here, the record identifies randomization but does not report enough about the primary analysis population, missing outcomes, blinding of pathological assessment, or analysis plan to establish how those considerations were addressed. The posted result is best understood as an uncertain, unadjusted comparison of the recorded surgical response groups.

16. Why This Trial Matters Statistically

This trial is statistically notable because it compares treatment sequences rather than contrasting a complex treatment with no active treatment. Both arms included chemotherapy with radiation and surgery; the distinguishing feature was induction chemotherapy before that shared sequence in Arm B. The endpoint therefore asks whether the recorded pathological response differs between two multimodal strategies.

The primary outcome is clinically interpretable as a proportion with no residual cancer in a surgical specimen, but the analysis population is narrower than the overall enrollment. The registry reports response assessment among participants who had surgery, while safety fields use a different at-risk count. That distinction illustrates why endpoint definitions and analysis populations are central to interpreting trial percentages. Without a clear participant flow or missing-outcome account, readers cannot determine how closely the surgical groups reflect the randomized groups.

The estimated difference is positive for Arm B, but the confidence interval is broad and includes no difference. The example shows why a point estimate and a p-value should not be interpreted in isolation. Effect estimates describe the observed contrast, while intervals communicate sampling uncertainty; neither resolves potential bias or limitations in reporting.

The study also had several additional stated objectives, but secondary outcome results are not posted. This highlights a practical issue in registry-based evidence review: planned questions and available results may not align fully. Readers can assess the posted primary endpoint, but cannot draw conclusions about unreported survival, relapse, resection, or exploratory analyses.

Finally, the planning illustrations make clear that detecting smaller effects generally requires more participants under the stated assumptions. Such calculations can inform future study planning, but they are not evidence of the original design rationale or of the completed study's observed power. The trial is therefore useful as a case in careful reading of endpoint scope, uncertainty, and analysis-population reporting.

17. 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.
• Participant counts were reconstructed from posted percentages and arm sizes, so small rounding differences are possible.

The primary response result is reported only for participants who had surgery. The registry record does not explain why the outcome assessment groups contain 55 participants in Arm A and 54 in Arm B, while the safety fields list 63 at risk in each arm. It does not report participant flow, reasons for not undergoing surgery, or a method for handling missing response assessments. This limits the ability to relate the result to the randomized population.

The outcome measure reports pathologic complete response, but the stated primary objective also refers to the proportion with less than 50% residual cancer. That additional category is not reported as a result. The record also does not post secondary findings for survival, R0 resection, relapse, or the other listed objectives, so the response estimate cannot be placed alongside those outcomes.

The study was open label, and the registry record does not state whether pathological assessors were blinded to treatment assignment. It does not provide detail on how the specimen categories were applied, whether assessments were centrally reviewed, or how disagreements were handled. These details could affect assessment of measurement consistency, but their impact cannot be inferred from the posted record.

The posted inferential summaries are unadjusted recalculations from the response counts. The registry record does not report an adjusted analysis, the original statistical analysis plan, the primary sample-size justification, interim analyses, or a multiplicity strategy. It also does not report confidence intervals for the individual arm response rates. The reported comparison should therefore be read with the uncertainty and analytical limitations described, without assuming additional procedures.

The serious adverse event figures are counts, and the record does not provide exposure-adjusted rates, detailed event timing, or death counts. Safety comparisons may be difficult to interpret without those details. Finally, the registry's completion status and dates do not supply information about follow-up duration for each participant or the completeness of longer-term outcomes.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: Oxaliplatin-Based Chemotherapy and Chemoradiotherapy or Chemoradiotherapy in Esophageal or Gastroesophageal Carcinoma — NCT00525915.
• Registry record last updated 2015-02-09; status last verified 2015-01.
• Results first posted 2015-02-09.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

21. Record Summary

NCT00525915 is a completed randomized phase II study of oxaliplatin-based chemotherapy and chemoradiotherapy strategies for resectable esophageal or gastroesophageal carcinoma. The parallel-group study was open label and enrolled 126 participants. Arm A received chemotherapy with radiation followed by surgery; Arm B received preoperative chemotherapy before the same chemoradiotherapy approach and surgery.

The primary posted endpoint is pathologic complete response among participants who had surgery. The reported rates are 13.0 in Arm A and 26.0 in Arm B. The unadjusted recalculated risk difference is 0.132, with a 95% confidence interval from -0.014 to 0.278 and p = 0.081. The interval includes no difference, and the outcome population is not the full enrollment group.

Serious adverse events are recorded for 10 participants in Arm A and 7 in Arm B, with 63 at risk in each arm. Secondary outcome results, the trial's original sample-size rationale, and detailed analysis and missing-data procedures are not reported in the registry record.