Academic-Industrial Partnership for Translation of Acoustic Angiography: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Academic-Industrial Partnership for Translation of Acoustic Angiography (NCT03544268), a non-randomized phase 4 study. The primary outcome is Sensitivity and Specificity.

2. Clinical Question

The study aimed to distinguish malignant from benign lesions using imaging assessments before biopsy. Sensitivity concerns detection among lesions that are malignant; specificity concerns correctly classifying lesions that are benign. These measures address different types of classification error, so a comparison of one alone would not fully answer the stated question.

The brief summary also lists secondary aims: comparing area under the curve and radiologist preference for the techniques across lesion characteristics. Those aims do not replace the primary comparison. The registry record does not report analyzable primary results or numerical secondary results, so it cannot establish how the techniques compared on those measures.

3. Trial Design

This was a completed, phase 4, single-center study sponsored by UNC Lineberger Comprehensive Cancer Center. The registry lists a start date of 2019-03-25, primary completion on 2021-05-25, and an enrollment of 20. The study is described as nonrandomized, unmasked, and sequential. It is not a randomized trial in which one group receives acoustic angiography and another comparable group receives only b-mode ultrasound.

The brief summary describes a one-arm study of patients from the UNC Breast Clinic who had suspicious lesions and were scheduled for biopsy. Participants were to undergo acoustic angiography alongside b-mode ultrasound before biopsy. The arm descriptions also include an Image Optimization group: the record says 30 participants would be recruited to help optimize imaging parameters, in addition to clinical patients. The brief summary describes a recruitment goal of 60 patients, while the registry enrollment is 20. These figures have different descriptions, and the record does not clarify recruitment achieved in each group or reconcile the figures.

The listed arms therefore should not be treated as like-for-like clinical comparison groups. The study’s central diagnostic comparison is between imaging methods assessed in a reader study, rather than between randomized participant groups. The record does not report the reader-study procedures in enough detail to establish how images were assigned, whether readers assessed images independently, or how reader variability was handled.

The primary outcome timeframe is listed as 2 years. The registry describes five outcome measures: the primary sensitivity-and-specificity comparison and secondary measures involving area under the curve, radiologist preference, vessel tortuosity, and a model-based malignancy score. The registry record does not report a statistical analysis plan, an explanation for the enrollment target, or whether any interim analyses were planned.

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: The data were unable to be analyzed because we could not visualize the vessels to calculate this value. The system was unable to achieve the imaging resolution to visualize the vasculature.

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 Sensitivity and Specificity, measured over 2 years.

The primary endpoint was a reader-study comparison of the sensitivity and specificity of acoustic angiography with conventional b-mode ultrasound for evaluating breast lesions. Sensitivity describes how often an imaging assessment identifies a lesion as malignant when it is malignant. Specificity describes how often it identifies a lesion as benign when it is benign. Used together, these measures describe performance across the two diagnostic categories relevant to the study question.

The endpoint was intended to assess whether the imaging techniques differ in their ability to classify lesions against the malignant-versus-benign distinction. Biopsy was part of the clinical context, but the registry’s primary-outcome description is the comparison of imaging sensitivity and specificity, not a treatment effect or a clinical outcome.

The registry record says the data could not be analyzed because the vessels could not be visualized to calculate the value; it attributes this to insufficient imaging resolution to visualize the vasculature. Thus, the planned endpoint does not have a reported estimate in the record. This is an inability to evaluate the endpoint as intended, not a reported finding that the imaging methods have equal performance or that either method has a particular level of sensitivity or specificity.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The central method was a reader-study comparison of sensitivity and specificity for two imaging techniques. In general, a reader study asks readers to interpret images and compares their classifications against a reference classification. In this study, the stated clinical distinction was malignant versus benign breast lesions. The registry does not report enough detail to establish the reference-standard procedures, the readers’ experience, or how images were presented. Biopsy is described as the clinical context, but the precise endpoint adjudication method is not reported in the registry record.

What do sensitivity and specificity describe?

As general statistical concepts, sensitivity is the proportion of malignant cases classified as malignant by a test, while specificity is the proportion of benign cases classified as benign. Each measure has its own denominator: sensitivity is evaluated among cases known to be malignant, and specificity among cases known to be benign. They are not interchangeable. A method can have different performance on these two dimensions, so reporting both is useful for a diagnostic comparison.

These measures depend on having classifiable cases and a suitable reference classification. If the imaging feature required for an assessment cannot be visualized, a sensitivity or specificity estimate based on that feature may not be calculable. The registry’s explanation is that vessels could not be visualized at the resolution achieved. This is a problem with evaluating the planned measure, not a numerical estimate of diagnostic performance. The record does not report how many cases were affected or whether an alternative assessment was attempted.

What would a confidence interval add?

In general, a confidence interval expresses the uncertainty around an estimated quantity under the assumptions of the statistical method. For a comparison of diagnostic measures, it can show which values are reasonably compatible with the observed data. A narrow interval indicates greater precision than a wide interval, though interpretation also depends on the clinical or scientific question and the analysis design. No confidence intervals for sensitivity, specificity, or their comparison are reported in the registry record. Since the primary values could not be calculated, precision cannot be assessed from the posted primary results.

A p-value, in general, assesses how compatible data are with a specified null hypothesis under a statistical model; it does not measure the size or importance of an effect. A p-value alone would not solve the interpretive problem of an endpoint that could not be calculated. The registry record does not report primary p-values or an inferential analysis for the primary comparison.

How should the secondary measures be understood?

Area under the curve is a summary of how well a diagnostic score separates cases across possible classification thresholds. In general, it can describe discrimination without selecting a single threshold, but its interpretation depends on the score, the reference classifications, and the data used. The registry lists area under the curve as a secondary outcome but does not report a numerical estimate or its uncertainty.

Radiologist preference is a different kind of outcome from diagnostic accuracy. It records a preference between imaging techniques for lesion characteristics, as described in the brief summary. Preference does not, on its own, establish whether a method classifies malignancy more accurately. The specific preference scale, reader-level analysis, and results are not reported. Vessel tortuosity and the model-based malignancy score are also listed as secondary outcomes; the registry does not report their numerical results or detailed analysis methods.

Why might reader and image structure matter?

As general statistical guidance, observations from the same reader or the same participant may be related rather than independent. If a study includes repeated interpretations, an analysis may need to account for this clustering to avoid overstating precision. How repeated readings or multiple lesion characteristics were handled is not reported in the registry record. The record also does not describe whether the same readers assessed both techniques or whether images were masked during interpretation. Those details matter for understanding potential reader effects and the dependence between measurements.

Over-dispersion is a general issue that can arise when observed variability exceeds what a chosen statistical model expects. Accounting for it can prevent uncertainty from being understated when the model’s assumptions are too restrictive. The registry does not report the fitted models or enough outcome data to assess whether over-dispersion was relevant to any analysis in this study. It would be speculative to claim that a particular model, correction, or variance estimate was used.

What do the planning illustrations mean?

In general, sample-size calculations are planning tools whose results depend on assumptions about the effect to detect, the significance level, allocation, and statistical method. The registry describes its planning illustration as using a two-sided significance level of 0.05, equal allocation between two arms, a normal approximation, and an effect set as a stated fraction of the observed effect. It also notes that time-to-event illustrations count total events, while other illustrations count participants per arm. These are described as planning assumptions, not as the study’s own design justification.

Those assumptions should not be mistaken for observed power. A planning count is not evidence that the completed study had a particular probability of detecting an effect, and it does not replace a reported analysis. Equal allocation is an assumption of the illustration, not a description of the nonrandomized study design. The record does not report a sample-size justification for the trial itself. No conclusion about adequacy of enrollment can be drawn from the planning assumptions alone, especially when the primary endpoint was not analyzable.

10. Multiplicity, Missing Data and Analysis Populations

The registry lists a primary sensitivity-and-specificity outcome and secondary outcomes concerning area under the curve, radiologist preference, vessel tortuosity, and a model-based malignancy score. Considering several outcomes raises a general multiplicity issue: if many statistical tests are conducted, the chance of at least one apparently notable result can increase. Researchers may address this through prespecified outcome hierarchies, adjusted thresholds, or careful emphasis on estimates and uncertainty. The registry record does not report whether any multiplicity procedure was planned or used.

The record lists no analyses and does not provide numerical estimates for the primary or secondary outcomes. It therefore does not establish which outcomes were analyzed, whether all planned analyses were completed, or whether results were selected for reporting. The analysis plan, including any prespecified contrasts or handling of multiple readers and lesion characteristics, is not reported in the registry record.

Missingness also requires care. The stated primary problem is that the required vessels could not be visualized, leaving the data unable to be analyzed for the intended value. The record does not describe this as ordinary loss to follow-up or report how many observations were unusable. It does not state whether missing or unclassifiable images were excluded, whether any alternative method was used, or whether sensitivity analyses were conducted. General methods for missing data rely on assumptions about why measurements are missing; without details about the affected observations and their causes, those assumptions cannot be assessed.

Analysis populations are not specified. The record does not clarify which participants, lesions, images, or reader assessments would have been included in each endpoint analysis. It also does not explain the relationship between the clinical participants, the image-optimization participants, the enrollment figure, and the recruitment goals. These omissions limit the ability to reconstruct who contributed to any outcome or to assess the population represented by the results.

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.

Most frequent non-serious adverse event terms

The registry reports group-specific adverse-event counts. For Acoustic Angiography, 3 participants were listed as at risk for serious events and 0 were affected; the count for other events was 0 among 3 at risk. For Image Optimization, 15 were listed as at risk for serious events and 0 were affected; the count for other events was 3 among 15 at risk. Deaths were recorded as 0 in each group.

These are counts as recorded, not a randomized comparison of safety between equivalent groups. The groups have different recorded numbers at risk, and the study is nonrandomized. The registry does not provide event descriptions for the three other events in Image Optimization, timing, duration of observation, or a detailed account of how events were collected. It also does not report statistical comparisons or uncertainty intervals for these counts.

A count of 0 recorded serious events does not prove that serious events are impossible or that the intervention is safe in other settings. The event counts should be read in the context of this study’s recorded groups and limited information, not as evidence about comparative diagnostic performance.

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

The registry record documents an attempt to compare acoustic angiography with conventional b-mode ultrasound for evaluation of suspicious breast lesions. Its primary question concerned sensitivity and specificity in distinguishing malignant from benign lesions. The record states that the planned primary data could not be analyzed because the system did not achieve imaging resolution sufficient to visualize the vasculature.

That outcome is informative about a limitation encountered in carrying out the planned measurement. It does not provide an estimate of sensitivity, specificity, or the difference between methods. It also does not show that the methods are equivalent, that acoustic angiography is inferior, or that b-mode ultrasound is superior. Such claims would require interpretable comparative outcome data, which are not reported in the registry record.

The secondary outcomes do not change that interpretation. Area under the curve, radiologist preference, vessel tortuosity, and a model-based malignancy score are listed, but numerical results and uncertainty estimates are not reported. Their presence in the record identifies outcomes of interest, not findings.

The record’s enrollment and recruitment descriptions also need to be kept distinct. It lists enrollment of 20, while the study description refers to recruiting 60 patients and the image-optimization arm description refers to 30 participants. The record does not clarify achieved recruitment for the different components or explain how these figures relate. They do not establish the number of participants who contributed to the primary endpoint, which was not analyzable.

Accordingly, the clearest result is that the intended vessel-based primary analysis could not be completed at the achieved imaging resolution. The available record does not establish diagnostic performance, comparative effectiveness, or a definitive safety profile. The findings should be interpreted as a limitation in evaluating the planned endpoint rather than as a positive or negative comparison of the imaging methods.

13. Why This Trial Matters Statistically

This study is statistically instructive because diagnostic performance depends on measurements being available and interpretable. Sensitivity and specificity are not abstract properties that can be estimated independently of image quality, classification rules, and a usable reference standard. The registry’s explanation that vessels could not be visualized at the achieved resolution shows how a technical constraint can prevent the planned endpoint from being calculated at all.

That is different from an imprecise estimate. With an imprecise estimate, a value can be calculated but uncertainty may be substantial. Here, the registry says the necessary value could not be calculated. A confidence interval or significance test cannot substitute for a missing endpoint estimate. This distinction helps prevent an absence of analyzable results from being misread as evidence of no difference.

The study also illustrates why design labels and comparison structure matter. Although the registry includes an image-optimization group, it describes a nonrandomized study and explicitly does not support treating the arms as a like-for-like control comparison. The clinical question compares imaging methods, while group labels alone do not establish a randomized method comparison. Reader-study details would be important for understanding paired assessments, reader variability, and the dependence among interpretations, but those methods are not described in sufficient detail.

Finally, the difference between the enrollment figure and the recruitment descriptions underscores the need to distinguish goals, optimization participants, and participants actually enrolled. Without a clear account of analysis populations, readers cannot infer who contributed to each outcome. The record’s limitations are therefore not only about the missing primary estimate; they also concern the information needed to interpret how the study was conducted and what its outcome data represent.

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 primary limitation is stated directly in the registry: the data could not be analyzed because the system did not achieve sufficient resolution to visualize the vasculature needed to calculate the endpoint. This leaves the planned sensitivity-and-specificity comparison without a reported estimate. The record does not say whether a modified imaging approach or alternative endpoint was evaluated.

The registry does not report numerical results or uncertainty for the secondary outcomes. It also does not provide the reader-study protocol details needed to assess reader selection, image presentation, masking, repeated interpretations, or the handling of reader-to-reader variation. Although masking is listed as none, the consequences for interpretation cannot be determined without knowing what information readers had and how assessments were conducted.

The record is nonrandomized and warns that its arms are not a like-for-like control comparison. The recruitment descriptions are not reconciled: the record lists enrollment of 20, while the brief summary refers to 60 patients and the image-optimization description refers to 30 participants. The record does not specify the achieved recruitment or analysis population for each component.

Details of the reference classification, missing or unclassifiable observations, endpoint thresholds, and statistical analysis plan are not reported in the registry record. The record also does not describe any multiplicity approach, sample-size justification, or interim analysis plan. These gaps prevent a full assessment of endpoint construction, precision, and the potential for selective analysis. The listed adverse-event counts have limited accompanying information and should not be treated as a definitive comparative safety assessment.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Academic-Industrial Partnership for Translation of Acoustic Angiography — NCT03544268.
• Registry record last updated 2022-04-01; status last verified 2022-03.
• Results first posted 2022-04-01.
• 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

The registry describes a completed phase 4, nonrandomized, unmasked, sequential study sponsored by UNC Lineberger Comprehensive Cancer Center. Its stated primary aim was to compare the sensitivity and specificity of acoustic angiography and conventional b-mode ultrasound for evaluating breast lesions. The study involved suspicious lesions at the UNC Breast Clinic and imaging before scheduled biopsy.

The registry records enrollment of 20 and describes a recruitment goal of 60 patients; the image-optimization description refers to 30 participants. The relationship between those figures is not clarified. The primary outcome timeframe is listed as 2 years.

The primary data could not be analyzed because vessels could not be visualized at the resolution achieved. Numerical results for the listed secondary outcomes are not reported in the registry record. The adverse-event counts are reported by group, but the study is not a randomized safety comparison. The record therefore describes the study’s objective and a technical limitation, without establishing comparative diagnostic performance.