# Project 2: Non-invasive imaging metrics to optimize early treatment switching decisions and prognostic modeling of long-term outcomes

> **NIH NIH P01** · UNIVERSITY OF CALIFORNIA, SAN FRANCISCO · 2023 · $375,951

## Abstract

The overall objective of the Program Project is to optimize every patient’s likelihood of reaching a pathologic complete
response (pCR) by using imaging, histopathology and molecular biomarkers to guide their treatment. Project 2 focuses on
advancing the imaging methods in the evolved design of I-SPY2.2, to identify patients that might benefit from a change in
course of treatment. In the I-SPY2 trial design, MRI measurements of functional tumor volume (FTV) are the biomarker
used to inform the longitudinal model for evaluation of drug arms. In I-SPY2.2, FTV is used at the individual patient level
to tailor treatments, raising the need for greater control over variability in MRI performance. We have been addressing
many of the elements involved in standardization of MRIs performed in the clinical setting through NCI-funded efforts in
the area of quantitative imaging. We also performed retrospective studies using data from 990 patients randomized to one
of 9 experimental drug arms completed by 2016 to better understand the impact of variability on FTV’s performance as a
biomarker and those findings have been used to introduce refinements to the I-SPY2 MRI exam protocol. Specific Aims 1
and 2 focus on iterative improvements to the de-escalation strategy and pre-RCB, as well as the escalation strategy,
respectively. While FTV-based response provides the initial signal for considering a change in treatment, different
strategies are required to improve the level of certainty for recommending escalation or de-escalation, given MRI’s
relative strength in demonstrating extensive disease, and limitation in detecting minimal disease. In the pre-RCB de-
escalation strategy, a negative finding on core biopsy of the tumor bed at 12-weeks is required before the option to omit
AC is offered. In the scenario of escalation, we use MRI response of less than 30% at 3-weeks to flag potential poor
response and recommend repeat imaging at 6-weeks, where the threshold for escalation to Block B is <65% FTV
response. We will build on these initial strategies in several ways. Current MRI prediction models are based on data from
the initial 990 patients enrolled under I-SPY2 and have been optimized within subtypes defined by HR and HER2. We
will refine these models using the more biologically-relevant Response-Predictive Subtype schema and using expanded I-
SPY patient cohorts. More comprehensive MRI prediction models twill be developed, integrating classifiers of shape,
heterogeneity and normal tissue features that can be derived from the same MRI data used to measure FTV. Working with
Project 3 investigators, we will pose the question of added value of ctDNA in both the de-escalation and escalation
strategies, investigating the use of ctDNA at multiple timepoints. Specific Aim 3 addresses the potential additive benefit
of serial FTV measurement to the histopathologic endpoint residual cancer burden (RCB) which has been well-established
as prognostic in the neoadjuvant set...

## Key facts

- **NIH application ID:** 10628610
- **Project number:** 2P01CA210961-06A1
- **Recipient organization:** UNIVERSITY OF CALIFORNIA, SAN FRANCISCO
- **Principal Investigator:** Nola M. Hylton-Watson
- **Activity code:** P01 (R01, R21, SBIR, etc.)
- **Funding institute:** NIH
- **Fiscal year:** 2023
- **Award amount:** $375,951
- **Award type:** 2
- **Project period:** 2017-09-08 → 2028-06-30

## Primary source

NIH RePORTER: https://reporter.nih.gov/project-details/10628610

## Citation

> US National Institutes of Health, RePORTER application 10628610, Project 2: Non-invasive imaging metrics to optimize early treatment switching decisions and prognostic modeling of long-term outcomes (2P01CA210961-06A1). Retrieved via AI Analytics 2026-05-23 from https://api.ai-analytics.org/grant/nih/10628610. Licensed CC0.

---

*[NIH grants dataset](/datasets/nih-grants) · CC0 1.0*