Cancer is a disease of the genome, driven by somatic mutation and evolutionary selection, and continually shaped by immune recognition and elimination. My research has repeatedly shown that tumor-intrinsic features predict progression and outcome across diverse cancers, enabling disease forecasting while defining new targets 1-7. For example, we described a ‘big bang’ model of tumor growth wherein malignant and even metastatic potential is specified early such that some tumors are ‘born to be bad’ 1. Similarly, we demonstrated that breast cancer recurrence risk could be predicted decades prior based on somatic genomic copy number aberrations 5-7.
These findings sparked other questions. Do hereditary or host differences influence somatic evolution and the molecular subtype of disease an individual develops? When and how do molecular differences in tumors of the same histology arise? For example, germline BRCA1 mutations are associated with a 55-72% lifetime risk of breast cancer, enriched for estrogen-receptor negative (ER-) disease, while BRCA2 mutations are associated with ER+ disease. However, the basis for these molecular differences is poorly understood. More generally, sporadic cancers are assumed to result from random mutations acquired during cell division and hence attributable to ‘bad luck’. Our findings suggest that this process is not random but influenced by one’s germline genome and immune system – and hence can be predicted.
Analyzing 6,000 patients with pre-invasive, invasive or metastatic breast cancer using a powerful computational framework that accounts for ancestry, germline genotype, and HLA (human leukocyte antigen), we discovered a pivotol role for hereditary and immune features in dictating the subtype of BC and its aggressiveness 8. Specifically, hereditary variants in recurrently amplified oncogenes, such as HER2, combined with a patient’s HLA, dictates the levels of oncogene-derived epitopes that can be presented for recognition by the host’s immune system. Thus, when specific combinations of HLA haplotype and seemingly innocous genetic variants in these oncogenes co-occur in the same individual, the immune system’s ability to detect and surveill nascent pre-invasive tumor cells that display these epitopes, is corrospondingly impacted. As such, individuals with high germline epitope burden (GEB) in the HER2 oncogene are less likely to develop HER2+ BC compared to other subtypes. Our work indicates that high GEB leads to increased MHC-class I presentation of HER2 epitopes, and that immunoediting selects against tumor clones with many copies of HER2.
The same pattern holds true for other subgroups of breast cancer, including the aggressive ER+ luminal tumors, we previously identified via an Integrative Classification scheme (ICs), which have a persistent recurrence risk for two decades 5,6. These four ‘high-risk’ ER+ IC subgroups are defined by copy number amplifications, similar to HER2 (IC1: RPS6KB1/PRR11, IC2: CCND1, IC6: FGFR1, IC9: MYC). Together, these subgroups account for 25% of ER+ disease and most recurrences. These genome unstable subgroups are immune-depleted and refractory to standard-of-care therapies, underscoring the need for need for new approaches. Moreover, the high-risk ICs are enriched amongst young women. Further, pre-invasive ductal carcinoma in situ (DCIS) are addicted to these same oncogenes 7.
While high GEB is protective in DCIS owing to immune-surveillance and elimination, it is a poor prognostic factor in invasive breast cancer and accompanied by a “switch” to immune tolerance. Indeed, low GEB is associated with progression to invasive breast cancer. Invasive tumors that overcome this immune-mediated negative selection are immune cold and myeloid enriched with greater propensity to recur, emphasizing the importance of earlier interventions. I will discuss ongoing efforts to elucidate the molecular basis for immune suppression in these tumors through multi-modal spatial profiling of a large collection of breast tumors sampled throughout disease progression.
Further, I will review new data demonstrating that the patterns of germline-mediated immunoediting we first described in breast cancer, extends to diverse solid tumors, providing a redout of pre-existing immunity to tumor-associated antigens. Our findings also unearth a previously unappreciated source of clonal antigens that may be exploited to treat and intercept aggressive tumors. Finally, I will discuss how this provocative discovery helps to explain the missing heritability of malignancy – with implications for risk stratification and predicting cancer risk in healthy individuals.
References