Quantitative ultrasound radiomics guided adaptive neoadjuvant chemotherapy in breast cancer: early results from a randomized feasibility study

Abstract

Background: In patients with locally advanced breast cancer (LABC) receiving neoadjuvant chemotherapy (NAC), quantitative ultrasound (QUS) radiomics can predict final responses early within 4 of 16-18 weeks of treatment. The current study was planned to study the feasibility of a QUS-radiomics model-guided adaptive chemotherapy.

Methods: The phase 2 open-label randomized controlled trial included patients with LABC planned for NAC. Patients were randomly allocated in 1:1 ratio to a standard arm or experimental arm stratified by hormonal receptor status. All patients were planned for standard anthracycline and taxane-based NAC as decided by their medical oncologist. Patients underwent QUS imaging using a clinical ultrasound device before the initiation of NAC and after the 1^st and 4^th weeks of treatment. A support vector machine-based radiomics model developed from an earlier cohort of patients was used to predict treatment response at the 4^th week of NAC. In the standard arm, patients continued to receive planned chemotherapy with the treating oncologists blinded to results. In the experimental arm, the QUS-based prediction was conveyed to the responsible oncologist, and any changes to the planned chemotherapy for predicted non-responders were made by the responsible oncologist. All patients underwent surgery following NAC, and the final response was evaluated based on histopathological examination.

Results: Between June 2018 and July 2021, 60 patients were accrued in the study arm, with 28 patients in each arm available for final analysis. In patients without a change in chemotherapy regimen (53 of 56 patients total), the QUS-radiomics model at week 4 of NAC that was used demonstrated an accuracy of 97%, respectively, in predicting the final treatment response. Seven patients were predicted to be non-responders (observational arm (n=2), experimental arm (n=5)). Three of 5 non-responders in the experimental arm had chemotherapy regimens adapted with an early initiation of taxane therapy or chemotherapy intensification, or early surgery and ended up as responders on final evaluation.

Conclusion: The study demonstrates the feasibility of QUS-radiomics adapted guided NAC for patients with breast cancer. The ability of a QUS-based model in the early prediction of treatment response was prospectively validated in the current study.

Clinical trial registration: clinicaltrials.gov, ID NCT04050228.

Keywords: adaptive chemotherapy; artificial intelligence; breast cancer; neoadjuvant chemotherapy; quantitative ultrasound (QUS); radiomics.

Figure 1

Consort diagram of the study.

Figure 2

B-mode imaging and corresponding QUS-parametric maps at different time points (pretreatment or week 0, and week 1 and 4 of neoadjuvant chemotherapy) for 1 patient from responder and non-responder group. MBF range was from -9.6 dB to 34.0 dB, SS range was from -5.7 dB/MHz to 1.6 dB/MHz, SI range from -7.3 dB to 49.0 dB, AAC range was from 20.2 db/cm³ to 81.6 db/cm³, ASD range was from 40 μm to 171 μm. The scale bar represents a length of 2 cm. MBF, mid-band fit; SS, spectral slope; SI, spectral intercept; AAC, average acoustic concentration; ASD, average spectral diameter.

Figure 3

Sankey diagram for predicted response at week 4 using QUS-radiomics model with the final response on an individual patient basis. Red tiles indicate non-response (predicted or actual) and green tiles indicate response (predicted or actual). In the experimental arm 25/28 patients were not adapted based on information provided to their oncologist whereas 3/28 were adapted. Patient with early surgery is indicated with a yellow tile.

Figure 4

Individual patient predictions based on predictor class scores at week 4 for patient response. R indicates the zone (negative class score) for predicted response and NR indicates the zone (+ve class score) for non-response.

Figure 5

(A) Schematic diagram for the administration of standard AC-T chemotherapy. Weeks are shown from left to right. Typical durations are illustrated. (B-D) indicate the three patients in the Experimental Arm that were adapted (IAA-1, IIA-2, and IIA-3) in Figure 3 . In (B) AC was shortened to move to T. In (C) the AC and T were shortened to move to surgery and in (C) AC was shortened and T was intensified,.