ARFI, VisR and DDAI Ultrasound for Improving Discrimination of Malignant and Unresponsive Breast Cancer: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for ARFI, VisR and DDAI Ultrasound for Improving Discrimination of Malignant and Unresponsive Breast Cancer (NCT03785782), a registered phase not stated study. The primary outcome is Correlation of Peak Displacement (PD) for Detecting Malignancy.

2. Clinical Question

3. Trial Design

This was an unblinded, open-label, exploratory clinical study of ultrasound-derived measurements in women with breast lesions. The stated purpose was to evaluate the diagnostic relevance of metrics related to stiffness, elasticity, viscosity and anisotropy. The registry describes two aims: assessing whether ARFI, VisR and DDAI ultrasound can detect malignant breast masses, and assessing whether these methods can distinguish masses that respond to neoadjuvant chemotherapy from those that do not.

The biopsy aim concerned women with BIRADS-4 or BIRADS-5 masses. Its arm description says that 40 participants were to be recruited, with the investigators aiming for approximately 10 in each of the BIRADS categories 4a, 4b, 4c and 5. The separate chemotherapy aim describes 40 participants with malignant masses undergoing neoadjuvant chemotherapy. These are planned descriptions for separate aims, not evidence that each planned group contributed that number to the primary analysis.

The record lists enrollment as 45. The relationship between that enrollment figure and the recruitment descriptions for the two aims is not reported in the registry record. The primary analysis description reports 34 included subjects, with six exclusions for unsuitable lesions, insufficient image quality or withdrawal. Biopsy findings were malignant for 12 participants and benign for 22.

The study is marked completed. Its recorded start date is 2018-05-10 and its primary completion date is 2021-07-15. The registry lists ultrasound as the intervention and does not describe random allocation. The arms are not a like-for-like control comparison: one concerns women scheduled for biopsy, and the other concerns women scheduled for neoadjuvant chemotherapy. Further details of masking, recruitment flow and the allocation process 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: In total 34 subjects were included in the analysis. Six subjects were excluded because their lessons were unsuitable for the study criteria, had insufficient image quality, or were withdrawn. Biopsy finding of malignant:12 participants, Biopsy finding of benign: 22 participants. Women scheduled for neoadjuvant chemo are not included since they already have a malignant disease diagnosis.

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 Correlation of Peak Displacement (PD) for Detecting Malignancy, measured over Baseline imaging.

The primary outcome is the correlation of peak displacement (PD) for detecting malignancy, measured at baseline imaging. PD is defined in the registry as the maximum displacement of tissue in a breast lesion after an Acoustic Radiation Force Impulse excitation. The ultrasound data were processed to calculate PD, which was then related to biopsy findings classified as malignant or benign. The outcome is described as a mean of a continuous measure, with standard error as the dispersion measure and microns as the unit.

The purpose of measuring PD is to assess whether an ultrasound-derived tissue measurement can help discriminate between lesions with different biopsy classifications. The registry says that separate receiver operating characteristic (ROC) curves were constructed for ultrasound metrics. In general, ROC analysis evaluates how a diagnostic measure separates two outcome categories across possible decision thresholds. It can describe discrimination without requiring the reader to assume a single threshold, although a curve alone does not establish clinical usefulness.

The primary outcome description gives the analysis population and the measurement approach, but the numerical result is not reported in the registry record. The record also does not report a correlation estimate, ROC summary measure, confidence interval, selected threshold or sensitivity and specificity for PD. Therefore, the endpoint's intended role is clear, but the recorded information does not establish how well PD distinguished malignant from benign findings.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does a ROC analysis describe?

In general, a receiver operating characteristic (ROC) curve describes how a diagnostic measure separates two outcome categories as the decision threshold varies. For this study's primary question, the categories were malignant and benign biopsy findings. A threshold is a rule that classifies a measurement on one side as a positive result and on the other side as a negative result. Changing that rule changes the balance between identifying cases in one category and avoiding positive classifications in the other. Examining that trade-off across thresholds is the basis of ROC analysis.

The registry says that separate ROC curves were constructed for ultrasound metrics. It does not report a numerical summary of the curves, the thresholds considered, or results such as sensitivity and specificity. In general, a ROC curve is about discrimination: whether a measure tends to rank observations in one outcome category differently from those in the other. Discrimination is not the same as calibration, which concerns agreement between predicted values and observed outcomes, and neither by itself establishes that using a measure improves decisions.

What does correlation mean in this endpoint?

The primary outcome is described as the correlation of PD for detecting malignancy. PD is a continuous measurement: it represents the maximum tissue displacement after an ARFI excitation, with microns listed as the unit. The study related calculated PD to biopsy findings classified as malignant or benign. In general, correlation describes the degree to which two variables vary together according to a chosen statistical measure. When one variable is a binary classification, the exact interpretation depends on how the analysis represents and estimates the relationship. The specific correlation method is not reported in the registry record.

A correlation estimate would not, by itself, provide every detail needed to understand diagnostic performance. It would need to be read with its uncertainty and alongside the ROC analysis and any threshold-specific results. The registry does not report a correlation estimate or confidence interval for PD. It also does not report how the primary endpoint was calculated beyond the description of processing ultrasound data and relating PD to biopsy findings.

Why does uncertainty matter?

In general, an estimate from a study is subject to sampling variability: repeating a study in a different sample could produce a different estimate. A confidence interval expresses the uncertainty associated with an estimate under the statistical model and its assumptions. A narrower interval indicates greater precision than a wider interval, but precision does not establish that a result is important or unbiased. The registry's primary outcome description names standard error as the dispersion measure but does not report the endpoint estimate, its standard error or a confidence interval. Consequently, the precision of the primary result cannot be assessed from the record.

The analyzed biopsy population is reported as 34 subjects, with six exclusions. Exclusions related to image quality or eligibility may affect which observations contribute to an analysis. In general, missing or excluded observations matter when the probability of being absent is related to the measurement or outcome. The record does not report how exclusions were distributed across biopsy categories, whether any measurements were missing among included subjects, or whether analyses assessed the effect of exclusions.

What can the sample-size illustrations tell readers?

The registry describes sample-size planning illustrations as calculations based on the observed effect, using 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, and distinguishes participants per arm for most outcomes from total events for time-to-event calculations. These are planning illustrations, not the trial's own design or evidence of observed power. Their numerical results and the effect used for those illustrations are not reported in the registry record.

In general, sample-size calculations depend on assumptions about the effect to be detected, variability, error rates, allocation and analysis method. A planning calculation cannot show that a study had a particular probability of detecting an effect after its results are known. The record does not report a protocol-based sample-size justification, so the illustrations should not be read as the study's original recruitment target rationale.

What analysis details remain unclear?

The record mentions ROC curves and a mean continuous outcome, but does not report the statistical model, the precise correlation method, covariate adjustment, handling of repeated measurements, or criteria for selecting any diagnostic threshold. It also does not report whether statistical analyses were prespecified or whether interim analyses occurred. In general, such choices can affect estimates and their interpretation. Without the analysis details and numerical results, an independent assessment of the primary analysis is limited.

10. Confidence Intervals for Each Arm

No two-arm comparison is shown for this trial, but each arm's mean can be given with a t-based 95% confidence interval computed from the posted mean, spread and number of participants.

R code and output

arm <- c("Biopsy Finding of Malignant", "Biopsy Finding of Benign")
n   <- c(12, 22)
m   <- c(4.5803, 8.628)
sp  <- c(2.9947, 7.8776)
sd  <- sp * sqrt(n)      # standard errors converted to SDs
for (i in seq_along(n)) {
  se <- sd[i] / sqrt(n[i]); h <- qt(0.975, n[i] - 1) * se
  cat(sprintf("CARM|%s|%d|%.3f|%.3f|%.3f|%.3f\n", arm[i], n[i], m[i], sd[i], m[i] - h, m[i] + h))
}

CARM  Biopsy Finding of Malignant  12  4.580  10.374  -2.011  11.172
CARM  Biopsy Finding of Benign  22  8.628  36.949  -7.754  25.010

11. Multiplicity, Missing Data and Analysis Populations

The registry lists 16 outcome measures and 16 analyses. It describes several ultrasound-derived metrics and two aims: identifying malignant masses and distinguishing chemotherapy responders from nonresponders. The primary outcome description focuses on PD at baseline imaging and biopsy findings. The relationship between the listed outcomes, the analyses and any hierarchy beyond the named primary outcome is not reported in the registry record.

When multiple outcomes or analyses are examined, there can be more opportunities to observe apparently noteworthy findings by chance. In general, a statistical plan may specify how primary and secondary questions are distinguished and whether any adjustment or other strategy is used to address multiplicity. The record does not report such a strategy, nor does it report whether the outcome list was prespecified or changed during the study.

The record describes six exclusions from the primary analysis because lesions were unsuitable for the study criteria, image quality was insufficient, or participants withdrew. It does not give the number associated with each reason or explain how missing measurements were handled. In general, exclusions and missing data can affect the analyzed population, especially if the reasons for absence are related to the measurement or outcome. A complete assessment would require information about missingness by outcome and the analysis approach; these details are not reported in the registry record.

The record reports 34 subjects in the primary analysis but does not describe an analysis population for every listed outcome. It also does not report whether the analysis followed a prespecified intention-to-diagnose, per-protocol or other population definition. The denominator and handling of missing or unusable measurements should therefore not be assumed to be identical across outcomes.

12. 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 registry reports serious adverse events for the group scheduled for biopsy and the group scheduled for neoadjuvant chemotherapy. For the biopsy group, the number at risk for serious adverse events is 39 and the number affected is 0. For the chemotherapy group, the number at risk is 5 and the number affected is 0. The record also reports 0 other adverse events in each group and 0 deaths in each group.

These figures describe the adverse-event entries recorded for the groups and at-risk counts shown. They do not establish that no event occurred outside the period or population captured by those entries. The safety population counts differ from the recruitment descriptions for the study aims; how those counts relate is not reported in the registry record. The record does not provide event definitions, follow-up duration for safety, severity grading details or information about how events were collected. Accordingly, the zero counts should be read as the registry's reported figures, not as a complete account of all possible harms.

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

The study addresses whether ultrasound-derived measurements, including PD, can help distinguish malignant from benign breast lesions, and whether the methods can distinguish masses that respond to neoadjuvant chemotherapy from those that do not. The registry describes the imaging approach, the biopsy-based classification for the primary question and an analysis using ROC curves. It also gives the primary analysis population and reasons for exclusions.

Those details explain the intended diagnostic comparison, but the primary numerical findings are not reported in the registry record. There is no reported PD estimate, correlation result, ROC summary, confidence interval or threshold-specific diagnostic result. The chemotherapy-response findings are also not reported. The record therefore does not allow readers to judge how much separation the measurements showed, how precise any estimates were, or whether performance varied across the two aims.

In general, a diagnostic measurement's ability to discriminate between outcome categories is only one part of evaluating it. The quality and consistency of measurement, the choice of threshold, uncertainty in estimates, and applicability to the population of interest also matter. These considerations cannot be resolved from the study description alone. The exploratory, unblinded design is stated, but further details relevant to evaluating implementation and analysis are not reported in the registry record.

The enrollment entry is 45, while the descriptions of the two aims each state planned recruitment of 40 participants. The primary outcome description reports 34 subjects included in its analysis. These figures refer to different registry entries and study descriptions; the record does not explain their relationship. Planned recruitment should not be interpreted as the observed analysis population or as evidence that every participant contributed to every outcome.

The appropriate reading is limited to what the record documents: a completed exploratory study with a stated diagnostic question and a described analysis population. It does not report enough primary outcome results to support a conclusion about diagnostic performance or chemotherapy response.

14. Why This Trial Matters Statistically

This study raises a statistical question about translating continuous ultrasound measurements into evidence about diagnostic discrimination. PD is a quantitative measurement, whereas the primary reference classification is biopsy finding as malignant or benign. ROC analysis is a standard general framework for examining separation across thresholds, and the registry identifies it as the approach used for ultrasound metrics. The pairing of a continuous measure with a biopsy-based classification makes the analysis question understandable, even though the numerical findings are not reported.

The record also highlights why analysis populations matter. It reports 34 subjects in the primary analysis and six exclusions for eligibility, image-quality or withdrawal reasons, while describing larger planned recruitment figures for the two aims. Those details make it important to distinguish planned recruitment, overall enrollment and outcome-specific analysis populations. The registry does not explain how the figures relate, so the population represented by the primary analysis should not be conflated with the full study description.

There is a further distinction between demonstrating a statistical association and establishing useful diagnostic performance. In general, association describes how measurements relate to outcomes; discrimination concerns how well a measure separates outcome categories; and practical use additionally depends on thresholds and the context in which decisions are made. The registry describes the analysis framework but does not report results that would permit those questions to be answered.

Finally, the study includes separate diagnostic and chemotherapy-response aims. They concern different questions and populations, and should not be treated as a single randomized comparison. The record's exploratory framing makes transparent reporting of estimates, uncertainty, exclusions and analysis methods especially relevant. Those results and details are not reported in the registry record, limiting statistical interpretation while still allowing readers to identify the intended questions.

15. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The registry record does not report the primary numerical result for PD or the ROC findings. It therefore does not permit assessment of effect size, precision or diagnostic discrimination. Although the outcome description states that standard error was used as a dispersion measure, the estimate and its standard error are not reported. The record also does not report threshold-specific results, a correlation estimate or a confidence interval.

The primary analysis included 34 subjects, and six subjects were excluded for reasons that included unsuitable lesions, insufficient image quality or withdrawal. The record does not give the number excluded for each reason, characterize their biopsy classifications, or describe sensitivity analyses addressing exclusions. The relationship between the enrollment entry of 45, the planned recruitment descriptions of 40 for each aim, and the primary analysis population is not reported.

The design is described as unblinded, open-label and exploratory. Details about measurement reproducibility, assessor procedures, prespecified analysis decisions and any adjustment for multiple analyses are not reported in the registry record. The record lists 16 outcomes and 16 analyses but does not explain the analysis hierarchy or multiplicity approach.

The chemotherapy-response aim is described, but its results and outcome definitions are not reported. The record also does not explain how response was classified or how observations were handled for that aim. Finally, the safety entries report event counts and at-risk counts but do not specify follow-up duration or event-collection methods. These omissions restrict interpretation of both diagnostic and safety information.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: ARFI, VisR and DDAI Ultrasound for Improving Discrimination of Malignant and Unresponsive Breast Cancer — NCT03785782.
• Registry record last updated 2025-02-20; status last verified 2025-02.
• Results first posted 2024-12-09.
• 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.

19. Record Summary

The registry record describes NCT03785782, an exploratory, unblinded, open-label study of ARFI, VisR and DDAI ultrasound measurements in breast lesions. Its aims were to assess detection of malignant masses and discrimination between masses that respond to neoadjuvant chemotherapy and those that do not. The primary outcome is the correlation of peak displacement with malignancy at baseline imaging, using biopsy findings classified as malignant or benign and ROC analysis for ultrasound metrics.

The record lists enrollment as 45, a start date of 2018-05-10, a primary completion date of 2021-07-15 and status as completed. The primary analysis description reports 34 subjects, including 12 with malignant and 22 with benign biopsy findings; six subjects were excluded. Numerical primary results and chemotherapy-response findings are not reported in the registry record. Serious adverse events, other adverse events and deaths are recorded as 0 in each of the two listed groups.