Title: The ImPrint immune signature identifies high risk early breast cancer patients who may benefit from PD1 checkpoint inhibition in I-SPY2

Publication: ASCO 2022, Poster 514

Authors: Kuilman et al.

Background

  • Remarkable increase of novel Immuno-Oncology drugs in many malignancies led to the need for biomarkers to identify who would benefit.
  • Various predictive biomarkers have been developed but none have consistently predicted efficacy.
  • I-SPY2 qualified several expression-based immune biology related signatures that predict response to PD1/PDL1 immune checkpoint inhibition (ICI)
  • We assessed whole transcriptome data of patients with high-risk early-breast cancer (EBC) who received ICI within the neoadjuvant biomarker-rich I-SPY2 trial (NCT01042379), aiming to migrate the I-SPY2 research findings into robust clinical grade signature to predict sensitivity to PD1/PDL1 ICI.

Methods

  • Whole transcriptome microarray data were available from pre-treatment EBC biopsies of 69 HER2- patients of the I-SPY2 Pembrolizumab (4 cycles) (Discovery set- Table1A) and 70 HER2- patients of the I-SPY 2 Durvalumab/Olaparib (Validation set- Table 1B) arms. All patients had a High-Risk 70-gene MammaPrint (MP)3,4 profile, were Luminal or Basal-type based on 80-gene BluePrint molecular subtyping profile
  • Pathologic complete response (pCR) was defined as no residual invasive cancer in breast or nodes at the time of surgery.
  • The most significant predictive genes for pCR (effect size >0.45) were identified by comparing pCR and RD (Residual Disease) groups in the Pembrolizumab arm by iteratively splitting the Discovery set in training and test, balancing for Hormonal Receptor (HR) status and using Leave one out cross validation for performance assessment.
  • Pathway analysis was performed with gene set enrichment analysis (GSEA) using Molecular Signatures Database/Hallmark gene sets (adjusted p-value ≤0.05).
  • Prevalence analysis was performed on a set of 1463 patients enrolled in the I-SPY2 trial (849 HR+HER2-; 614 Triple Negative)

Results

The 53-gene signature ImPrint (Figure 1), predicts pCR to PD1 inhibition with 94% sensitivity and 84% specificity in all patients (Discovery set, Figure 2A). In Triple Negative, sensitivity and specificity are 100% and 70%, and in HR+HER2- 83% and 89%, respectively. The Positive Predictive Value (PPV) is 77% in the HR+HER2- group. In the Validation set (Figure 2B) , ImPrint predicts pCR to PDL1 inhibition with 76% sensitivity and 85% specificity in all patients, 89% sensitivity and 58% specificity in Triple Negative and 70% sensitivity and 97% specificity in HR+HER2-. Notably, the PPV is 93% for the HR+HER2- group.

ImPrint prevalence analysis on 1463 EBC, shows that majority of Triple Negative EBC is Immune+ (n=438/614, 71%), however a clinically relevant subset of HR+HER2- is also Immune+ (n=213/849, 25%) (Figure 3).

Over 90% of the ImPrint genes have known immune response related functions (including PD-L1 and PD-1, as well as immune predictive genes from I-SPY22). GSEA indicates that ImPrint identifies a subset of immune active tumors (Immune+) with enrichment of different immune pathways, such as Interferon-α- and -γ response, known to play key roles in activation of cellular immunity stimulation of antitumor immune-response (Figure 4).

Conclusions

▪ ImPrint predicts pCR to PD1-PDL1 ICI with high sensitivity and specificity in both discovery and validation sets.
▪ Over 90% of the ImPrint genes have known immune response related functions (including genes that codify PD-L1 and PD-1)
▪ ImPrint identifies tumors with an immune active phenotype denoted by the enrichment of several immune-related pathways
▪ ImPrint appears very effective in identifying a subset of HR+HER2- patients who could benefit fromimmunotherapy.
▪ Prospective validation of ImPrint will be performed within the I-SPY 2.2. trial.