Laura J. van ’t Veer PhD, Professor of Laboratory Medicine, Co-leader of the Breast Oncology Program, Director of Applied Genomics, Helen Diller Family Comprehensive Cancer Center, University of California, San Francisco.
MILAN, Italy—Around 25 per cent of patients with newly-diagnosed triple negative breast cancer will not benefit from neoadjuvant checkpoint inhibitor immunotherapy with pembrolizumab—even though it improves outcomes among the remaining majority. This finding comes from the I-SPY2 TRIAL and was reported at the 2024 European Breast Cancer Conference, held in Milan, Italy, by Laura van ’t Veer, Leader of the Breast Oncology Program at the University of California in San Francisco.
A subset of patients with triple-negative early-stage breast cancers, identified in the study through a gene test as having “response predictive sub-types” had a very low likelihood of response to neoadjuvant pembrolizumab, suggesting that such patients could be spared the potential toxicities of immunotherapy.
After her talk in Italy Dr van ’t Veer met up with Audio Journal of Oncology reporter Peter Goodwin to discuss the I-SPY findings.
INTERVIEW: Laura J. van ’t Veer PhD
“Hello, Peter Goodwin her with the Audio Journal of Oncology………..New drugs for breast cancer are traditionally tested in triple negative ……….. from me, Peter Goodwin, Good-bye” 15:46 secs
“Immune subtyping in the Response Predictive Subtypes (RPS) identifies a subset of triple negative (TN) early- stage breast cancer patients with a very low likelihood of response to neoadjuvant immunotherapy (IO): results from 5 IO arms of the I-SPY2 TRIAL”
https://cm.eortc.org/cmPortal/Searchable/ebcc14/config/Normal/#!sessiondetails/0000107210_0
Immune subtyping in the Response Predictive Subtypes (RPS) identifies a subset of triple negative (TN) early-stage breast cancer patients with a very low likelihood of response to neoadjuvant immunotherapy (IO): results from 5 IO arms of the I-SPY2 TRIAL
D.M. Wolf1,, C. Yau2,, J. Haan3,, D. Wehkamp3,, A. Witteveen3,, A. Glas3,, M. Campbell2,, R. Nanda4,, J. Chien5,, R. Shatsky6,, C. Isaacs7,, A. Barcura3,, L. Mittempergher3,, M. Kuilman3,, D. Yee8,, A. DeMichele9,, J. Perlmutter10,, L. Pusztai11,, L. Esserman2,, L.J. van ‘t Veer PhD12,.
1University California San Francisco, Laboratory Medicine, San Francisco, USA.
2University California San Francisco, Surgery, San Francisco, USA.
3Agendia, Research and Development, Amsterdam, The Netherlands.
4University Chicago, Medicine, Chicago, USA.
5University California San Francisco, Medicine, San Francisco, USA.
6University California San Diego, Medicine, San Diego, USA.
7Georgetown University, Medicine, Washington DC, USA.
8University Minnesota, Masonic Cancer Center, Minneapolis, USA.
9University Pennsylvania, Medicine, Philadelphia, USA.
10Gemini Group, Advocacy, Ann Arbor, USA.
11Yale University, Medicine, New Haven, USA.
12University California San Francisco, Helen Diller Family Comprehensive Cancer Center, San Francisco, USA.
Neoadjuvant immunotherapy (IO) has become standard of care for early stage TN breast cancer. However, not all patients respond and IO poses significant risk of permanent, life altering immune related adverse events (iRAEs) including adrenal insufficiency and thyroid dysfunction. Previously we showed immune gene expression signatures dominated by STAT1/chemokine/cytokine/dendritic markers associate with pathologic complete response (pCR) in TN treated with IO and developed a clinically applicable Immune classifier (ImPrint) predicting response to IO for both TN and HR+ that is now being used in I-SPY2.2 as part of the Response Predictive Subtypes. This initial ImPrint classifier performed with high accuracy for TN and HR+ patients combined, though we noticed that this classifier could be further improved by reducing the false-negative rate for TN (ie high negative predictive value). Here we report the performance of a refined version of ImPrint for TN patients (ImPrintTN) from 5 IO arms of the I-SPY2 trial.
150 TN patients from 5 pooled IO arms (anti-PD1, anti-PDL1/PARPi, anti-PD1/TLR9 dual-IO, and anti-PD1 +/- LAG3 dual-IO, all plus taxane/anthracycline) and 128 patients from the taxane/anthracycline concurrent control arm were included in this analysis. Patients in IO arms with FFPE pre-treatment biopsies were divided into treatment- and response-balanced training and test sets; and an IO-response classifier was developed including additional immune signaling and checkpoint markers using pre-treatment mRNA from the training set (n=55). Patient biopsies were classified ImPrintTN+ (likely sensitive) vs. ImPrintTN- (likely resistant), by Agendia Inc using pre-treatment expression data. Performance of ImPrintTN for predicting pCR to IO in the test set, and overall was characterized using standard methods.
Overall, the pCR rate for TN over the 5 pooled IO arms was 54%. 66% of TN patients were ImPrintTN+. pCR rates with IO in the independent test set were 71% in ImPrintTN+ vs. 22% in ImPrintTN- (delta-pCR rate 49%; sensitivity=87%; negative predictive value (NPV)=78%). Similar results were observed in all 5IO arms taken together (test plus training), where pCR rates with IO were 74% in ImPrintTN+ vs. 16% in ImPrintTN- (delta-pCR rate 58%; sensitivity=90%; NPV=84%). In the control arm, pCR rates were 30% in ImPrint+ and 15% in ImPrint-, delta-pCR 15%).
https://www.audiomedica.com/wp-content/2025/08/240320-Laura-van-t-Veer-EBCC-2024-Predictive-Subtypes-AJO-PRODUCTION-MASTER-.mp3
Laura J. van ‘t Veer, PhD
for TN predicts response and non-response to a variety of IO regimens tested in I-SPY2. Within the ImPrintTN- subset, pCR rates to IO regimens are very low, and similar to that of non-IO containing regimens. Prospective validation is ongoing in I-SPY2.2. Our current data suggest that ImPrintTN may help inform prioritization of IO vs other treatments for TN patients to best balance likely benefit vs risk of serious irAEs.