Mehran Habibi

NSABP B-59/GeparDouze offers an important opportunity to refine how treatment effects and exploratory biomarkers are interpreted in early triple-negative breast cancer (TNBC). Adding atezolizumab to perioperative chemotherapy increased pathological complete response (pCR) but did not meet the trial’s primary statistical criterion for improving event-free survival (EFS). Its exploratory analyses further suggest that immune features such as tumor-infiltrating lymphocytes (TILs) and basal-like immune-activated (BLIA) versus immune-suppressed (BLIS) transcriptional states may help identify patients most likely to benefit. We argue that these findings are scientifically valuable but belong at different evidentiary levels: improvement in pCR does not establish long-term treatment benefit, and prognostic or within-subgroup associations do not establish a predictive biomarker without evidence of treatment-by-biomarker interaction. In 1,550 patients with stage II–III TNBC, pCR increased from 57.0% to 63.3% with atezolizumab, whereas the primary EFS analysis did not cross the prespecified boundary for superiority (hazard ratio [HR], 0.800; 95% confidence interval [CI], 0.621–1.030; P=0.083).¹ This does not establish absence of benefit—the confidence interval remains compatible with a clinically meaningful effect—but it defines the evidentiary boundary of the randomized comparison.

Background: Gene expression assays help personalize adjuvant chemotherapy decisions for hormone receptor-positive, HER2-negative (HR+HER2-) early breast cancer (EBC). The 70-gene risk of distant-recurrence signature, MammaPrint, demonstrated clinical utility in guiding chemotherapy de-escalation in genomically low risk patients in the MINDACT trial. This study evaluates MammaPrint as a continuous predictor of chemotherapy benefit in HR+HER2- EBC using real-world data (RWD) from the FLEX Registry. Methods: The study evaluated 1002 patients treated with endocrine therapy (ET) only or ET with chemotherapy (ET+CT) enrolled in FLEX (NCT03053193) with 5-year median follow-up. Propensity-score matching balanced treatment groups by menopausal status, T-stage, and nodal status. The primary endpoint was distant recurrence-free interval (DRFI). Regression and Cox proportional hazards models assessed chemotherapy benefit across MammaPrint Index (MPI) risk. Results: Most patients were postmenopausal (70.1%), node-negative (70.0%), and had grade 2 tumors (51.2%). The regression models showed that MPI strongly predicted 5-year DRFI in ET only (R2 = 0.99, P < .001) and ET + CT (R2 = 0.90, P < .001) groups, corresponding to an average absolute chemotherapy benefit of 5.6% in High 1 and 10.9% in High 2. Minimal improvement in DRFI with chemotherapy was observed for Low (1.7%) and UltraLow (<1.0%) risk groups. A multivariate Cox model with an MPI-by-treatment interaction term demonstrated that increasing MPI risk was associated with greater chemotherapy benefit on DRFI (HR = 0.15, P = .047). Chemotherapy benefit was significantly associated with premenopausal status, but not age, T-stage, nodal status, or grade. Conclusions: These RWD from the FLEX Registry demonstrate that MPI is predictive of both DRFI prognosis and chemotherapy benefit in HR+HER2- EBC. (NCT03053193)

Plastic EVs have different biodistribution compared to stiff and soft EVs.

Stiff EVs show increased biodistribution and retention in vivo.A, Schematic of EV isolation and staining prior to tail-vein injection of 10 µg of EVs in nude mice. NIR imaging (B) and mean SNR (C) of MDA-MB-231 vesicle biodistribution in dorsal, left, ventral, and right sides (mean ± SEM). Signal intensity is in arbitrary units (a.u.). Three mice in 0.5 kPa condition, and four in 25 kPa condition; one-way ANOVA. NIR imaging (D) and mean SNR biodistribution (E) in the lungs, liver, spleen, kidneys, and bone/bone marrow (mean ± SEM). Signal intensity is in arbitrary units (a.u.). Three mice in MDA-MB-231 0.5 kPa condition and four in 25 kPa condition; one-way ANOVA. F, MDA-MB-231 EV binding assay to ECM proteins collagen type I, collagen type IV, and laminin (mean ± SEM). Three biological repeats for collagen type I and collagen type IV; five for laminin. Paired t test.

Stiff EVs promote cancer cell migration and dissemination. A, Schematic of 2-day GFP+ zebrafish embryo model system used to test the ability of MDA-MB-231 EVs (red, filled triangles) to disperse MDA-MB-231 cancer cells (blue, open triangles). Scale bar, 100 µm. B, Full zebrafish body images of the embryos at time 0 and 24 hours postinjection of MDA-MB-231 cancer cells. Scale bar, 100 µm. Representative images. C, Enlarged images of disseminated MDA-MB-231 cells (white) and EVs (red). Scale bar, 100 µm. Representative images. D, Percentage of injected embryos with cancer cell dissemination to the head or the tail. Total number of fish per condition is 47 for PBS control, 63 for 25 kPa, and 78 for 0.5 kPa condition. Five biological repeats of EVs. One-way ANOVA.

Stiffness measurements of primary patient tissues

Table S1 - Primers for qRT-PCR analysis

Matrix stiffness impacts the quantity of EVs produced by patient tissue. A, Representative primary breast tumor tissue. Sample has been mechanically mapped using microindentation. Dark circles indicate measurements of the elastic modulus expressed in units of kPa. Crosses are nonmeasurable indentations. Dotted lines indicate where the tissue was sectioned into stiff (middle) and soft (right) regions for subsequent vesicle collection. B, Mean compression measurements (kPa, mean ± SEM) of patient tumor and adjacent normal tissues. Ten adjacent normal tissue samples and nine tumor tissue samples. Nonparametric t test. C, Vesicles released per gram of tissue in pathologist determined tumor adjacent or tumor tissue patient samples. Ten normal tissue samples and eight tumor tissue samples. D, Combined tumor-adjacent and tumor tissues samples separated by mean compression measurements (kPa, mean ± SEM). Fourteen samples < 10 kPa and four samples > 10 kPa. E, Tumor samples separated by mean microindentation measurements (kPa, mean ± SEM). Thirteen samples < 10 kPa and six samples > 10 kPa. Nonparametric t test.