Validation of a Computed Tomography (CT) Based Fractional Flow Reserve (FFR) Software Using the 320 Detector Aquilion ONE CT Scanner.: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Validation of a Computed Tomography (CT) Based Fractional Flow Reserve (FFR) Software Using the 320 Detector Aquilion ONE CT Scanner. (NCT03149042), a registered phase not stated study. The primary outcome is Comparison of CT Based FFR With Invasive FFR, ROC Analysis.

2. Clinical Question

This is a diagnostic-validation question, not a randomized treatment-effect question. The record describes patients undergoing clinically mandated invasive angiography and compares a CT-based estimate with an invasive measurement. That structure can assess agreement or diagnostic discrimination relative to the chosen reference measurement, but it does not by itself establish effects on clinical outcomes or compare treatment strategies.

3. Trial Design

The registry identifies NCT03149042 as a completed study sponsored by the State University of New York at Buffalo. Its condition is coronary atherosclerosis, and its listed intervention is CCTA. The registry enrollment is 75. The brief study summary separately describes a pilot clinical study including approximately 50 patients over a year and half at GVI. The difference in how enrollment is described is not reconciled in the registry record, so the figures should be read as distinct record statements rather than as a single resolved participant count.

The described patient group consisted of people scheduled for clinically mandated elective invasive coronary angiography at Buffalo General Hospital. CCTA images were processed with a research-based CT-FFR algorithm. The primary endpoint compares CT-based FFR with invasive FFR using receiver operating characteristic analysis. The primary timeframe is 24 hours. The primary-endpoint description reports invasive measurements in 42 LADs, 11 LCXs and 8 RCAs, and says that nine patients had multi-vessel disease with invasive FFR measured in two vessels. These vessel counts are not participant totals.

The summary also describes comparison of patient and phantom CT-FFR, bench-top FFR in patient-specific 3D-printed phantoms, invasive FFR, and 30 days of follow-up. Secondary outcomes include phantom comparisons and several forms of comparison with invasive FFR. These are descriptions of intended or listed assessments; completion details and numerical results are not reported in the registry record.

The record does not describe a randomized allocation. It states that allocation is not stated and that the arms are not a like-for-like control comparison. Accordingly, the record supports describing a diagnostic validation design, not a randomized controlled comparison. Further details of recruitment, participant flow, exclusions, and the analysis population 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: From the total number of patients, nine patients had multi-vessel disease and I-FFR was measured in two vessels. In total I-FFR was measured in: 42 LADs, 11 LCXs and 8 RCAs. Invasive -FFR measurement location was from the ostium to two lesion lengths below the distal throat of the lesion .

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

6. Primary Endpoint

The registry lists the primary outcome as Comparison of CT Based FFR With Invasive FFR, ROC Analysis, measured over 24 hours.

The primary endpoint is the area under the receiver operating characteristic curve for CT-based FFR compared with invasive FFR. The registry describes invasive FFR at or below 0.8 as positive. The CT-based estimate was adjusted to the location of invasive FFR measurement, described as two lesion lengths below the distal end of the lesion.

In general, a receiver operating characteristic curve describes how sensitivity and specificity change across possible thresholds for a diagnostic measure. Its area summarizes the measure’s ability to rank reference-positive observations above reference-negative observations across thresholds. It is a measure of discrimination, not a direct measure of whether an individual diagnosis is correct, and it does not by itself establish calibration, clinical utility, or patient outcomes.

The endpoint description calls the unit the probability of accurate diagnosis and specifies a 95% confidence interval. The primary timeframe is 24 hours. No numerical area estimate, confidence limits, sensitivity, specificity, or threshold-specific result is reported in the registry record. Without those results, the endpoint’s observed discrimination and its precision cannot be assessed from the record. The listed endpoint definition explains what the investigators intended to evaluate; it is not evidence of a particular level of diagnostic performance.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary method named in the registry is receiver operating characteristic analysis. In general, this approach evaluates how a continuous or ordered diagnostic measure separates observations assigned to positive and negative reference categories. At each possible decision threshold, sensitivity describes the share of reference-positive observations classified as positive, while specificity describes the share of reference-negative observations classified as negative. The ROC curve displays that trade-off across thresholds. Its area is a summary of discrimination across thresholds rather than a result at one selected threshold.

The reference classification in this study is invasive FFR at or below 0.8, designated positive in the endpoint description. This defines what counts as reference-positive for the stated ROC analysis. It does not mean that invasive FFR is a perfect biological truth; it is the comparator chosen by the study. The registry does not report a separate adjudication procedure, a prespecified operating threshold for CT-based FFR, or threshold-specific sensitivity and specificity for the primary endpoint.

The CT-FFR description says patient CCTA images were imported into Vitrea segmentation software and used in a research-based algorithm. Four image volumes acquired at 70%, 80%, 90% and 99% of the R-R interval were analyzed. The aortic root and the LAD, LCX, and RCA were automatically segmented and then manually adjusted to obtain centerlines and contours. The algorithm used changes in vessel diameter and computational fluid dynamics to calculate CT-based FFR. The user adjusted the distal pressure measurement location to match the invasive FFR location, described as two lesion lengths below the distal end of the lesion.

These details matter because a diagnostic comparison depends on how both measurements are defined and located. Matching measurement locations can make the comparison more directly aligned at the lesion level. Segmentation and manual adjustment are part of the image-processing procedure described by the registry, but reproducibility, blinding, and operator variability are not reported in the registry record. The record also does not provide enough detail to assess algorithm tuning, calibration, or whether model development and validation data were kept separate.

The primary endpoint includes a 95% confidence interval. In general, a confidence interval describes uncertainty in an estimate under the chosen statistical model and sampling assumptions. It does not guarantee that the interval contains a fixed true value, nor does it compensate for biases in selection, measurement, or analysis. Because the registry does not report the area estimate or its interval, readers cannot evaluate precision or compare the observed estimate with a performance benchmark.

The record lists vessel counts and notes that some patients contributed measurements in more than one vessel. In general, observations clustered within a person may be correlated because they share patient characteristics and measurement conditions. If that dependence is ignored, standard errors and confidence intervals may be too small. Methods such as clustered estimation or patient-level resampling can account for dependence in suitable settings, but the registry does not state which approach, if any, was used. No particular method should be inferred from the vessel counts alone.

Secondary outcomes list comparisons involving CT-based FFR and bench-top FFR using patient-specific 3D-printed phantoms, as well as ROC analysis, Pearson correlation, sensitivity, and specificity for bench-top FFR compared with invasive FFR. These methods answer different questions. In general, correlation measures the strength of association between measurements, but a strong correlation does not establish that two methods agree closely. Sensitivity and specificity describe classification at a selected threshold, while ROC analysis evaluates discrimination across thresholds. The registry does not give the numerical secondary results or the thresholds used for those analyses.

For planning, the record describes illustrations based on the observed effect, a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. It says the effect to detect is a stated fraction of the observed effect. These are planning assumptions, not the trial’s own design justification and not observed power. Time-to-event planning quantities are described as total events, while other planning quantities are described as participants per arm. No planning values or trial-specific sample-size justification are reported in the registry record. The record does not report a formal statistical analysis plan, interim-analysis approach, or multiplicity adjustment.

10. Multiplicity, Missing Data and Analysis Populations

The registry lists a primary ROC analysis and secondary outcomes concerning phantom comparisons, ROC analysis, Pearson correlation, sensitivity, and specificity. It also reports 9 outcome measures and 0 analyses in the registry analysis count. The named outcomes describe a range of questions and statistical summaries, but the record does not explain how confirmatory conclusions were to be selected across them.

In general, when several outcomes or analyses are examined, the chance of at least one apparently notable result can increase if each is interpreted separately. Prespecified outcome hierarchies or multiplicity adjustments can address that concern. The registry does not report a multiplicity strategy, so it is not possible to determine whether any adjustment was planned or applied. No numerical results are reported to assess how such choices might affect interpretation.

Missing-data handling and analysis populations are not reported in the registry record. The record does not describe participant flow, exclusions, unavailable images, failed segmentation, missing invasive measurements, or the number of observations included in each outcome analysis. The enrollment entry, the approximate patient count in the brief summary, and the adverse-event risk count do not establish the denominator for the primary ROC analysis.

The primary description reports vessel-level measurements and identifies patients with measurements in more than one vessel. It does not state whether the analysis was conducted at the vessel or patient level, or how dependence among vessels from the same patient was handled. These omissions limit assessment of the effective information and uncertainty behind any estimate. A complete interpretation would require the analysis population, missingness approach, dependence method, and the reported estimates with their uncertainty.

11. Safety Results

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

Deaths

Other (non-serious) adverse events

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

The adverse-event record lists the CCTA Coronary group with 52 at risk. It reports 0 serious adverse events among those at risk, 0 other adverse events among those at risk, and 0 deaths. No serious adverse-event terms are listed. These are registry counts for the recorded group; they are not a comparative estimate because the record does not describe a like-for-like control arm.

In general, a zero count in a registry entry means that no events are recorded in that entry. It does not prove that the underlying risk is zero, and it does not show that every possible event was detected or reported. The record does not provide a detailed adverse-event collection method, event definitions, severity grading beyond the serious category, follow-up completeness, or a safety-analysis population. The listed risk count should not be assumed to be the denominator for the primary diagnostic analysis.

The study’s main question concerns diagnostic measurement rather than a randomized treatment comparison. The safety counts therefore do not establish comparative safety or support a broad conclusion about harms. They describe the adverse-event figures as recorded, with the limitations of the registry information.

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

The registry describes a study designed to compare CT-based FFR with invasive FFR in patients undergoing elective invasive coronary angiography. Its primary endpoint is ROC area, using invasive FFR at or below 0.8 as the positive reference classification. The record also describes comparisons involving bench-top FFR and patient-specific 3D-printed phantoms. These details establish the research question and measurement framework.

The registry does not report the primary ROC area, its 95% confidence interval, or numerical results for the secondary outcomes. It therefore does not establish how well CT-based FFR discriminated between reference-positive and reference-negative vessels, how precise that estimate was, or how the phantom methods compared with invasive measurements. The brief summary’s expectation that the work might lead to an algorithm with better predictability is a rationale, not a reported finding.

In general, diagnostic discrimination is only one part of evaluating a test. It does not alone show agreement at a chosen operating threshold, calibration of predicted values, clinical usefulness, or effects on patient outcomes. The registry record does not report those additional evaluations in a way that permits conclusions. The mention of 30 days of follow-up in the summary does not supply follow-up outcomes, and it is distinct from the primary endpoint timeframe of 24 hours.

The record also does not describe a randomized control comparison, and it explicitly says the arms are not a like-for-like control comparison. Accordingly, the listed information should not be interpreted as a treatment-effect estimate. It supports a description of a diagnostic-validation study and its intended comparisons, while leaving actual performance and comparative conclusions unreported.

Care is also needed with the different counts in the record. Enrollment is listed as 75, the brief summary describes approximately 50 patients, the adverse-event record gives 52 at risk, and the primary endpoint reports vessel counts. These entries have different contexts and do not identify a common analysis denominator. They cannot be added or used interchangeably to infer the number of observations contributing to a particular result.

13. Why This Trial Matters Statistically

This study is statistically relevant because it puts a computational estimate derived from CT images alongside an invasive physiological measurement. The record describes image segmentation, manual adjustment, computational fluid dynamics, and matching the pressure-measurement location to the invasive measurement. Those features make the comparison dependent not only on the underlying diagnostic concept but also on image processing and measurement alignment.

The record also links patient measurements with an approach using patient-specific 3D-printed coronary phantoms and bench-top flow measurements. Such comparisons raise distinct statistical questions: whether measurements are associated, whether they agree, and whether they classify observations similarly against a reference. The listed use of correlation, sensitivity, specificity, and ROC analysis reflects that these questions are not interchangeable. In general, correlation alone cannot establish close agreement, and an ROC summary does not describe every operational aspect of a diagnostic method.

Repeated vessel measurements within a patient are another important design feature. The registry notes multi-vessel disease and reports vessel-level counts. In general, within-patient measurements may share information, so the unit of analysis and treatment of clustering matter for uncertainty estimates. The registry does not report those analysis details or the primary estimate, leaving readers unable to evaluate how the study quantified precision.

Finally, the study illustrates the difference between a stated validation aim and a documented result. The registry describes why the investigators considered phantom-based optimization useful, but it does not report numerical findings that establish performance or improvement. Statistical interpretation requires the estimate, its uncertainty, the analysis population, and methods suited to the data structure. Their absence limits conclusions while leaving the design question clear.

14. 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 most direct limitation of the registry record is that it does not report the numerical primary endpoint or its confidence interval. Without the ROC area, readers cannot assess discrimination; without the interval, they cannot assess precision. Numerical results for the listed secondary comparisons are also not reported in the registry record.

The record does not reconcile the enrollment entry of 75 with the brief summary’s description of approximately 50 patients. It does not provide a participant flow, explain exclusions, or identify the analysis denominator for the primary endpoint. The adverse-event count of 52 at risk is recorded for a separate purpose and does not resolve the primary-analysis population.

The primary description gives vessel-level counts and notes that some patients had invasive measurements in two vessels. It does not report whether the analysis accounted for within-patient dependence, or whether the analysis unit was the vessel, patient, or another unit. Confidence intervals and tests can be affected by this choice.

Important operational and statistical details are also not reported in the registry record: blinding, reproducibility of segmentation and manual adjustment, missing-data handling, prespecified thresholds beyond the invasive FFR classification, multiplicity procedures, and a formal analysis plan. The record does not state whether algorithm development and validation were separated. These gaps prevent assessment of potential bias, robustness, and generalizability from the registry information alone.

The study is described as a pilot, and the arms are not a like-for-like control comparison. The record therefore does not support treatment-effect conclusions. Its summary mentions 30 days of follow-up, but follow-up results are not reported, and the primary endpoint timeframe is 24 hours.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Validation of a Computed Tomography (CT) Based Fractional Flow Reserve (FFR) Software Using the 320 Detector Aquilion ONE CT Scanner. — NCT03149042.
• Registry record last updated 2020-11-17; status last verified 2020-11.
• Results first posted 2020-06-23.
• 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.

18. Record Summary

NCT03149042 is listed as completed and sponsored by the State University of New York at Buffalo. It concerns coronary atherosclerosis and describes CCTA-based estimation of FFR compared with invasive FFR in patients scheduled for elective invasive coronary angiography. The primary endpoint is ROC area, with invasive FFR at or below 0.8 designated positive and a 95% confidence interval specified. The endpoint timeframe is 24 hours.

The record also describes comparisons involving patient-specific 3D-printed phantoms and bench-top FFR. It lists 9 outcome measures and 0 analyses in its registry analysis count. Enrollment is recorded as 75, while the brief summary describes approximately 50 patients; the record does not reconcile these statements. It reports 0 serious adverse events, 0 other adverse events, and 0 deaths among 52 at risk in the CCTA Coronary group.

Numerical primary and secondary performance results, analysis populations, missing-data methods, and the statistical analysis plan are not reported in the registry record. The documented information therefore characterizes the study question and methods but does not establish the observed diagnostic performance.