Background: Functional tumor volume (FTV), measured from dynamic contrast-enhanced MRI, is an imaging biomarker that can predict treatment response in breast cancer patients undergoing neoadjuvant chemotherapy (NAC). The FTV-based predictive model, combined with core biopsy, informed treatment decisions of recommending patients with excellent responses to proceed to surgery early in a large NAC clinical trial. Methods: In this retrospective study, we constructed models using FTV measurements. We analyzed performance tradeoffs when a probability threshold was used to identify excellent responders through the prediction of pathology complete response (pCR). Individual models were developed within cohorts defined by the hormone receptor and human epidermal growth factor receptor 2 (HR/HER2) subtype. Results: A total of 814 patients enrolled in the I-SPY 2 trial between 2010 and 2016 were included with a mean age of 49 years (range: 24 to 77). Among these patients, 289 (36%) achieved pCR. The area under the ROC curve (AUC) ranged from 0.68 to 0.74 for individual HR/HER2 subtypes. When probability thresholds were chosen based on minimum positive predictive value (PPV) levels of 50%, 70%, and 90%, the PPV-sensitivity tradeoff varied among subtypes. The highest sensitivities (100%, 87%, 45%) were found in the HR−/HER2+ sub-cohort for probability thresholds of 0, 0.62, and 0.72; followed by the triple-negative sub-cohort (98%, 52%, 4%) at thresholds of 0.13, 0.58, and 0.67; and HR+/HER2+ (78%, 16%, 8%) at thresholds of 0.34, 0.57, and 0.60. The lowest sensitivities (20%, 0%, 0%) occurred in the HR+/HER2− sub-cohort. Conclusions: Predictive models developed using imaging biomarkers, alongside clinically validated probability thresholds, can be incorporated into decision-making for precision oncology.

Background: This multicenter and retrospective study investigated the additive value of tumor morphologic features derived from the functional tumor volume (FTV) tumor mask at pre-treatment (T0) and the early treatment time point (T1) in the prediction of pathologic outcomes for breast cancer patients undergoing neoadjuvant chemotherapy. Methods: A total of 910 patients enrolled in the multicenter I-SPY 2 trial were included. FTV and tumor morphologic features were calculated from the dynamic contrast-enhanced (DCE) MRI. A poor response was defined as a residual cancer burden (RCB) class III (RCB-III) at surgical excision. The area under the receiver operating characteristic curve (AUC) was used to evaluate the predictive performance. The analysis was performed in the full cohort and in individual sub-cohorts stratified by hormone receptor (HR) and human epidermal growth factor receptor 2 (HER2) status. Results: In the full cohort, the AUCs for the use of the FTV ratio and clinicopathologic data were 0.64 ± 0.03 (mean ± SD [standard deviation]). With morphologic features, the AUC increased significantly to 0.76 ± 0.04 (p < 0.001). The ratio of the surface area to volume ratio between T0 and T1 was found to be the most contributing feature. All top contributing features were from T1. An improvement was also observed in the HR+/HER2- and triple-negative sub-cohorts. The AUC increased significantly from 0.56 ± 0.05 to 0.70 ± 0.06 (p < 0.001) and from 0.65 ± 0.06 to 0.73 ± 0.06 (p < 0.001), respectively, when adding morphologic features. Conclusion: Tumor morphologic features can improve the prediction of RCB-III compared to using FTV only at the early treatment time point.

Background Breast cancer is a heterogeneous disease and can be categorized into clinically or biologically meaningful subtypes. Predictive models built by MRI biomarkers performed better when they are optimized by breast cancer subtype than models optimized in the full cohort [1]. Functional tumor volume (FTV) measured from breast MRI has been used to assess tumor response to neoadjuvant therapy longitudinally in the I-SPY 2 TRIAL. Tumors show distinct morphological patterns, or phenotypes, on MRI. Previous studies demonstrated that either qualitative or quantitative measurements characterizing these phenotypes may provide additional information about treatment response [2,3]. In this study, we investigated if MRI morphologic phenotypes defined by unsupervised clustering is associated with breast cancer subtype and pathologic complete response (pCR) after neoadjuvant chemotherapy (NAC). Methods A cohort of 990 patients enrolled in the I-SPY 2 TRIAL were included in this retrospective analysis. Patients were randomized to one of nine experimental drug arms or standard NAC, and pCR was assessed at surgery. DCE-MRI data acquired at pretreatment (T0) and early treatment (T1) were analyzed. Four subtypes of breast cancer were defined by immunohistochemistry (IHC) based on hormone receptor (HR) and HER2 status. Radiomic features were extracted by PyRadiomics [4] using FTV masks from DCE-MRI. MRI morphologic phenotypes were determined based on unsupervised hierarchical clustering approach on extracted radiomic shape features plus FTV using Pearson correlation with agglomerative ward linkage. The associations between the unsupervised clusters of radiomic features and FTV with four IHC subtypes and pCR were evaluated using χ2 test of independence. Cramer’s V [5] were computed to measure the strength of association (higher Cramer’s V means stronger association). P-value < 0.05 was considered statistically significant. Results Three clusters were generated by unsupervised hierarchical clustering in a population of 910 patients included in our analysis (80 patients excluded due to missing pCR or DCE-MRIs). At T0, the unsupervised clusters showed statistically significant but weak association with pCR (Cramer’s V = 0.088, p = 0.029), but the association between the clusters and HR/HER2 subtypes did not reach significance (Cramer’s V = 0.055, p = 0.48). The unsupervised clusters based on T1 shape radiomic features showed statistically significant association with both pCR and HR/HER2 subtypes (p < 0.001 for both) with Cramer’s V of 0.231 and 0.154, respectively. Our results showed stronger association between pCR and cancer subtypes with MRI shape radiomic features at T1 than at T0. Various pCR rates were observed in MRI clusters at T1. They were 56%, 36%, and 23% in Cluster 1, 2, 3, respectively. Table 1 shows pCR rates by HR/HER2 subtype in each cluster. In all sub-cohorts, pCR rate was highest in Cluster 1 and lowest in Cluster 3. In HR+/HER2-, the pCR rate in Cluster 1 was 2-fold of the pCR rates in Clusters 2 and 3-fold of Cluster 3. pCR rate was statistically significantly different depending on the MRI clusters in the sub-cohorts except for the HR/HER2+ sub-cohort: HR+/HER2-, p< 0.001; HR+/HER2+, p=0.021; HR-/HER2+, p=0.083; HR-/HER2-, p< 0.001. Conclusion MRI phenotype generated by unsupervised clustering using radiomic shape features at both pretreatment and early-treatment time points was associated with pCR outcome. Stronger association was observed at early-treatment time point. The association differed by subtype, with the strongest observed in HR+/HER2- and triple negative subtypes. Our results suggest that radiomic shape features derived from DCE-MRI may be helpful for early prediction of tumor response to NAC. Citations 1. npj Breast Cancer 6, (2020). 2. Tomography 6, (2020). 3. Annals of Surgical Oncology 20, 3823–3830 (2013). 4. Cancer Research 77, e104–e107 (2017). 5. Korean Stat Soc 42, 323–328 (2013). Table 1. pCR rate by HR/HER2 subtype in each MRI cluster at T1 Citation Format: Nu N. Le, Natsuko Onishi, David C. Newitt, Jessica E. Gibbs, Lisa J. Wilmes, Efstathios Gennatas, Barbara LeStage, Laura J. Esserman, Nola M. Hylton, Wen Li. Association of MRI morphologic phenotype from unsupervised learning with breast cancer subtypes and treatment response [abstract]. In: Proceedings of the 2022 San Antonio Breast Cancer Symposium; 2022 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2023;83(5 Suppl):Abstract nr PD16-07.

Background: The apparent diffusion coefficient (ADC) presents a biomarker that is sensitive to tumor cellularity. ADC maps can be calculated from non-contrast diffusion-weighted magnetic resonance imaging (DW-MRI) measurements. ACRIN 6698, a sub-study of clinical trial I-SPY 2, investigated mean ADC – averaged over the whole tumor – as a marker to predict pathologic complete response (pCR) [1]. This work compares a group of histogram-based ADC metrics in addition to mean ADC for early prediction of pCR in patients stratified by breast cancer subtype. Methods: We performed a retrospective analysis of DW-MRI, dynamic-contrast enhanced (DCE) MRI, and clinical outcome (i.e., pCR at surgery) in a cohort of 79 female patients who were diagnosed with high-risk, stage II/III breast cancer. Patients underwent neoadjuvant chemotherapy (NAC) with paclitaxel (12 weeks), followed by doxorubicin plus cyclophosphamide (12 weeks). The included population represents a subset of the I-SPY 2 cohort and comprises 48 patients with hormone receptor [HR]+/HER2-, and 31 patients with HR-/HER2-. DW- and DCE-MRI acquisitions were performed according to the I-SPY 2 protocol at pretreatment (T0) and after three weeks (T1) and were analyzed to find early treatment percentage (%) change (T0 to T1) in any metric M; where %-change = 100 × (M(T1) – M(T0))/M(T0). Histogram analysis provided nine region-of-interest (ROI)-based ADC metrics (Table 1). ROIs were manually delineated by expert observers in three-dimensional ADC maps, focusing on diffusion-restricted regions [2]. DCE-MRI was analyzed for the integral I-SPY 2 imaging marker of %-change in functional tumor volume (FTV) between T0 and T1. Statistical analysis compared the predictive power of ADC metrics and FTV, including: the receiver-operating-characteristic (ROC) curve from a logistic regression model to predict pCR as ‘positive’, area-under-the-curve (AUC) assessment, and rank-sum Wilcoxon test (p< 0.05: statistically significant). Results (Table 1): 16 out of 79 (20.3%) patients reached pCR at surgery, with 18.8% pCR among HR+/HER- and 22.6% among HR-/HER2- groups. For all nine computed ADC statistics (listed as median [Q1, Q3], across all patients), %-change was higher in patients who reached pCR than patients with non-pCR (highest value for metric ‘MIN’: 23.9% [-0.9%, 52.5] vs. 16.6% [0.4%, 27.6%], though without statistical significance: p=0.237). Likewise, %-change of FTV was also stronger in pCR patients than non-pCR patients (-58.8% [-80.6%, -22.5%] vs. -28.2% [54.2%, -2.7%], with statistical significance: p=0.036). For all patients combined (n=79), among the various reported ADC metrics, %-change in ‘PCTL_95’ (95th percentile of histogram) yielded the highest AUC (0.7; 95% CI = [0.56, 0.83]; p=0.012). %-change in FTV showed the second highest AUC (0.67; 95% CI = [0.52, 0.82]; p=0.036). By subtype, AUC was highest for %-change of ‘PCTL_95’ (0.69; 95% CI = [0.5, 0.87]; p=0.072) in the HR+/HER2- subgroup; and highest for both %-change of ‘MEAN’ (AUC = 0.73; 95% CI = [0.49, 0.94]; p=0.065) and ‘PCTL_75’ (AUC = 0.73; 95% CI = [0.49, 0.94]; p=0.073) triple negative (HR-/HER2-) subgroup. By comparison, %-change of FTV yielded AUCs of 0.64 (95% CI = [0.41, 0.85]; p=0.191) and 0.71 (95% CI = [0.51, 0.9]; p=0.098) in the HR+/HER2- and triple-negative subgroups, respectively. Conclusion: Various tumor ADC metrics from non-contrast DW-MRI demonstrate potential biomarkers for assessing responsiveness to NAC at an early treatment timepoint. ADC may have predictive performance that is comparable to FTV, depending on the breast cancer subtype. Observations for %-change in ‘MEAN’ ADC at T1 differed from previous reports [1], which may be explained by the small sample size and single (paclitaxel) drug arm. Additional studies are warranted to include patients of experimental arms and of HER2+ subtypes. [1] Partridge et al., Radiology 289(3):618-27 (2018) [2] Nu et al., Tomography 8: 1208-20 (2022) Results of statistical analysis of ADC-based and FTV markers for predicting treatment response at 3 weeks into NAC. Median [Q1, Q3] values represent the median and interquartile range over the respective patient population regarding the %-change of the respective ADC (FTV) metric from T0 to T1. T0: pretreatment, T1: 3-week timepoint, ADC: apparent diffusion coefficient, MRI: magnetic resonance imaging, DW: diffusion-weighted, DCE: dynamic-contrast enhanced, pCR: pathologic complete response, AUC: area under the ROC curve, 95% CI [LL, UL]: confidence interval lower and upper limit, PCTL_x: xth-percentile of tumor ADC histogram, MEAN: mean of ADC within ROI, MIN: minimum of ADC within ROI, MAX: maximum of ADC within ROI, FTV: functional tumor volume, **: statistically significant. Citation Format: Judith Zimmermann, Julia Carmona-Bozo, Nu N. Le, Natsuko Onishi, Lisa J. Wilmes, Jessica E. Gibbs, Jiachao Liang, David C. Newitt, Savannah Partridge, Patrick Bolan, Bonnie N. Joe, Elissa R. Price, Barbara LeStage, Nola M. Hylton, Wen Li. Diffusion-weighted magnetic resonance imaging for subtype-specific prediction of pathologic complete response in neoadjuvant chemotherapy [abstract]. In: Proceedings of the 2022 San Antonio Breast Cancer Symposium; 2022 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2023;83(5 Suppl):Abstract nr P6-01-33.

This study presents a retrospective analysis of inter-reader variability in tumor ADC measurements extracted from three types of ROIs using data from the multi-center I-SPY 2 trial. Overall, good agreement was observed between two readers. Mean ADCs measured in restricted ROIs were highly reproducible by manual delineation. Mean ADCs of multi-slice restricted ROIs showed higher ICCs than single-slice restricted ROIs.

This study tested the hypothesis that a change in the apparent diffusion coefficient (ADC) measured in diffusion-weighted MRI (DWI) is an independent imaging marker, and ADC performs better than functional tumor volume (FTV) for assessing treatment response in patients with locally advanced breast cancer receiving neoadjuvant immunotherapy. A total of 249 patients were randomized to standard neoadjuvant chemotherapy with pembrolizumab (pembro) or without pembrolizumab (control). DCE-MRI and DWI, performed prior to and 3 weeks after the start of treatment, were analyzed. Percent changes of tumor ADC metrics (mean, 5th to 95th percentiles of ADC histogram) and FTV were evaluated for the prediction of pathologic complete response (pCR) using a logistic regression model. The area under the ROC curve (AUC) estimated for the percent change in mean ADC was higher in the pembro cohort (0.73, 95% confidence interval [CI]: 0.52 to 0.93) than in the control cohort (0.63, 95% CI: 0.43 to 0.83). In the control cohort, the percent change of the 95th percentile ADC achieved the highest AUC, 0.69 (95% CI: 0.52 to 0.85). In the pembro cohort, the percent change of the 25th percentile ADC achieved the highest AUC, 0.75 (95% CI: 0.55 to 0.95). AUCs estimated for percent change of FTV were 0.61 (95% CI: 0.39 to 0.83) and 0.66 (95% CI: 0.47 to 0.85) for the pembro and control cohorts, respectively. Tumor ADC may perform better than FTV to predict pCR at an early treatment time-point during neoadjuvant immunotherapy.

Background: Checkpoint blockade pembrolizumab has demonstrated great potential to improve pathologic outcome for HER2- breast cancer. The apparent diffusion coefficient (ADC) is a non-contrast MRI-derived biomarker that is sensitive to changes in tumor cellularity. Clinical trial ACRIN 6698, a sub-study of I-SPY 2, demonstrated that ADC can predict pathologic complete response (pCR). This study compares the utility of ADC for early prediction of pCR in patients with HER2- breast cancer randomized to pembrolizumab versus standard neoadjuvant chemotherapy (NACT) in I-SPY 2. Methods: A retrospective analysis of imaging and clinical data was performed on a cohort of 249 women diagnosed with high-risk, stage II/III breast cancer. All patients were randomized to the standard NACT (paclitaxel) or pembrolizumab plus paclitaxel for 12 weeks, followed by doxorubicin plus cyclophosphamide. MRI exams performed at pretreatment (T0) and 3 weeks after the treatment started (T1) were analyzed. Tumor ADC was calculated within manually delineated region-of-interests on diffusion-weighted MRI. The percent change of ADC from T0 to T1 was evaluated in the prediction of pCR after NACT. Statistical analysis included Wilcoxon rank sum test and the area under the ROC curve (AUC). A p-value <0.05 was considered statistically significant. Results: A subcohort of 103 patients with analyzable diffusion-weighted MRI exams and known pCR (n=30)/non-pCR (n=73) outcome were included in this analysis. Among 103 patients, 62 had HR+/HER2- and 41 had triple negative breast cancer. Twenty-eight patients (17 HR+/HER2- and 11 triple negative) were randomized to receive pembrolizumab and 75 (45 HR+/HER2- and 30 triple negative) to standard NACT. Tumor ADC increased at 3 weeks in both standard and pembrolizumab cohorts with median ADC change of 11.5% (interquartile range [IQR]: 4.6, 16.2)% and 14.4% (IQR: 0.2, 19.9)%, respectively. In the pembrolizumab cohort, the difference in ADC change between non-pCR and pCR groups was estimated as -9.7% (95% confidence interval [CI]: -22.4, -0.9), with ADC increasing more in the pCR group. The AUC of predicting pCR in the pembrolizumab cohort was estimated as 0.73 (95%CI: 0.52, 0.93), while it was estimated as 0.63 (95% CI: 0.43, 0.83) in the standard NACT cohort. In comparison, the AUCs using functional tumor volume (FTV) to predict pCR were 0.61 (95%CI: 0.39, 0.83) and 0.66 (95% CI: 0.47, 0.85) in the corresponding cohorts (Table 1). The results suggest that ADC had higher association with pCR than FTV in the pembrolizumab cohort and FTV had higher association than ADC in the standard cohort. Conclusions: Tumor ADC, measured using diffusion-weighted MRI, demonstrates potential as a biomarker for assessing early response to immunotherapy in the neoadjuvant setting for high risk HER2- breast cancer. This study is limited by sample size. Future analysis with larger cohorts is warranted. Table 1.Association between MRI variables and pCRN (pCR rate)Difference between non-pCR and pCR groups (95% CI)AUC (95% CI)pDiffusion weighted MRI (percent change of ADC)Pembrolizumab28 (54%)-9.7 (-22.4, -0.9)0.73 (0.52, 0.93)0.041Standard75 (20%)-5.6 (-13.9, 2.1)0.63 (0.43, 0.83)0.13DCE-MRI (percent change of FTV)Pembrolizumab28 (54%)10.6 (-18.3, 43.9)0.61 (0.39, 0.83)0.34Standard74 (20%)25.3 (-0.6, 49)0.66 (0.47, 0.85)0.056 Citation Format: Wen Li, Nu N Le, Natsuko Onishi, David C Newitt, Jessica E Gibbs, Lisa J Wilmes, John Kornak, Savannah C Partridge, Barbara LeStage, Elissa R Price, Bonnie N Joe, I-SPY 2 TRIAL Imaging Working Group, I-SPY 2 TRIAL Consortium, Laura J Esserman, Nola M Hylton. Diffusion-weighted MRI for prediction of pathologic complete response in HER2- breast cancer treated with pembrolizumab plus neoadjuvant chemotherapy [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr P3-03-02.

This study evaluated the inter-reader agreement of tumor apparent diffusion coefficient (ADC) measurements performed on breast diffusion-weighted imaging (DWI) for assessing treatment response in a multi-center clinical trial of neoadjuvant chemotherapy (NAC) for breast cancer. DWIs from 103 breast cancer patients (mean age: 46 ± 11 years) acquired at baseline and after 3 weeks of treatment were evaluated independently by two readers. Three types of tumor regions of interests (ROIs) were delineated: multiple-slice restricted, single-slice restricted and single-slice tumor ROIs. Compared to tumor ROIs, restricted ROIs were limited to low ADC areas of enhancing tumor only. We found excellent agreement (intraclass correlation coefficient [ICC] ranged from 0.94 to 0.98) for mean ADC. Higher ICCs were observed in multiple-slice restricted ROIs (range: 0.97 to 0.98) than in other two ROI types (both in the range of 0.94 to 0.98). Among the three ROI types, the highest area under the receiver operating characteristic curves (AUCs) were observed for mean ADC of multiple-slice restricted ROIs (0.65, 95% confidence interval [CI]: 0.52–0.79 and 0.67, 95% CI: 0.53–0.81 for Reader 1 and Reader 2, respectively). In conclusion, mean ADC values of multiple-slice restricted ROI showed excellent agreement and similar predictive performance for pathologic complete response between the two readers.