Radiation therapy as a biological modifier of the breast cancer immune microenvironment

Radiotherapy has attracted considerable attention as a means to alter the immune cell compartment of breast malignancies in support of increased sensitivity to immunotherapy. Recent data from Schalck and colleagues corroborate the notion that dose is a critical determinant of the immunological effects of radiotherapy on the tumor microenvironment.

Immunotherapy with immune checkpoint inhibitors (ICIs) has not been as effective in patients with breast cancer as in patients with other malignancies, such as melanoma and non-small cell lung carcinoma (NSCLC)¹. Indeed, the majority of women with breast neoplasms bear a specific disease subtype commonly known as HR⁺HER2^- that is immunologically silent at baseline, and hence poorly responsive to ICIs². Thus, while encouraging results are being obtained by delivering ICIs in the neoadjuvant setting, a number of strategies have been evaluated to increase the immunogenicity of HR⁺HER2⁻ breast cancer in view of adjuvant or second-line systemic (immuno)therapy³. A large amount of preclinical evidence points to focal radiation therapy (RT) as an effective and safe means to convert immunologically silent HR⁺ breast malignancies into immunologically active tumors that may respond to ICIs⁴. However, multiple factors appear to play a major role in determining whether focal RT ultimately mediates immunostimulatory or immunosuppressive effects on the tumor microenvironment, including (but not limited to) dose and fractionation schedule⁵. Recent findings on longitudinal samples from patients with HR⁺HER2⁻ subjected to neoadjuvant focal RT lend extra support to the notion that dose is a critical determinant of the immunological effects of RT on the tumor microenvironment (TME)⁶.

Schalck and colleagues set out to investigate the effects of neoadjuvant focal RT delivered in either of two biologically equivalent boost doses (7.5 Gy × 1 fraction or 2 Gy × 5 fractions) in a cohort of 19 women with HR⁺HER2⁻ cancer enrolled in a window-of-opportunity Phase II clinical trial (PRECISE, NCT03359954). While the primary objective of the study was the longitudinal quantification of tumor-infiltrating lymphocytes (TILs), genomic and immunological changes imposed by focal RT were also assessed (in a subset of patients) by single-cell (sc) DNA, RNA, and TCR sequencing (seq). TILs were reported to increase in 11/19 patients, although often marginally and with no statistical difference between RT schedules⁷, prompting an analysis of all available longitudinal samples for scRNAseq (from 11 patients), scTCRseq (from 11 patients), and scDNAseq (from 8 patients)⁶.

Longitudinal scRNAseq analysis revealed that most of the malignant and non-malignant components of the HR⁺HER2⁻ breast cancer microenvironment shifted in high-dimensional space upon focal RT. More specifically, surgical specimens were enriched in naïve or central-memory CD4⁺ T cell populations expressing C-C motif chemokine receptor 7 (CCR7), a finding that was validated by immunohistochemistry (IHC) and was not paralleled by an increased relative abundance of CD4⁺FOXP3⁺ regulatory T (T_REG) cells or CD8⁺CCR7⁺ T cells. Conversely, focal RT elicited a decrease in the relative abundance of lymphocytes expressing immune effectors of the granzyme family, with the notable exception of a CD4⁺ T cell population expressing cytotoxic molecules alongside other key anticancer cytokines such as tumor necrosis factor (TNF) and interferon gamma (IFNG). Intriguingly, longitudinal TCR enrichment analysis (which was possible only on 5 patients) revealed that most TCR clonotypes were present exclusively pre- or post-treatment, indicating that RT considerably reconfigured the T cell infiltrate of HR⁺HER2⁻ breast tumors, most likely by eliminating existing TILs while promoting the recruitment of novel ones⁶.

Tumor-infiltrating myeloid cells also exhibited considerable alterations after focal RT. Overall, the relative abundance of tumor-associated macrophages (TAMs) increased upon irradiation, a finding that was validated by the IHC-assisted quantification of CD68⁺ cells. Of note, while the changes in abundance of specific TAM populations exhibited considerable variability across patients, in 9 out of 11 sample pairs, focal RT elicited a relative increase in the abundance of TAMs expressing Fc gamma binding protein (FCGBP) and C-X3-C motif chemokine receptor 1 (CX3CR1), which have previously been associated with clinically relevant immunosuppression in a variety of settings including a cohort of patients with HR⁺HER2⁻ breast cancer subjected to neoadjuvant immunotherapy with a programmed cell death 1 (PDCD1, best known as PD-1) blocker followed by PD-1 blockage alongside stereotactic body radiation therapy (SBRT)^8,9. That said, in 8 out of 11 patients, myeloid cells as a whole expressed increased levels of antigen presentation-related genes, at least hypothetically pointing to a more favorable microenvironment for T cell activation. Nonetheless, myeloid cell populations critically involved in anticancer immunity, such as dendritic cells (DCs), exhibited marginal alterations in relative abundance upon focal RT⁶.

Malignant cells identified by the expression of epithelial cell adhesion molecule (EPCAM) and inferred copy number profiles did not exhibit general transcriptional changes upon focal RT that would outweigh patient-driven transcriptional signatures. That said, a few specific genes were commonly upregulated in surgical samples compared to their paired pre-RT biopsy, including a number of genes coding for NF-κB signal transducers downstream of TNF engagement. Conversely, various genes involved in type I interferon (IFN) signaling were surprisingly downregulated in post- vs pre-RT HR⁺ breast cancer specimens. Along similar lines, no differences were documented with respect to genetic signatures of cell cycle distribution, which was corroborated by the IHC-assisted quantification of the marker of proliferation Ki-67 (MKI67, best known as Ki-67) positivity. Moreover, focal boost RT failed to alter the transcriptional levels of breast cancer subtype-defining markers, including estrogen receptor 1 (ESR1) and erb-b2 receptor tyrosine kinase 2 (ERBB2, best known as HER2)⁶.

Longitudinal scDNAseq analysis revealed that focal RT often changed the proportion of malignant cell subclones, enabling the identification of 4 patients with high genomic selection (HGS) and 4 patients with low genomic selection (LGS). Of note, samples from the HGS cohort were enriched in the transcriptional signature of estrogen signaling, both pre- and post-RT. Conversely, specimens from the LGS cohort exhibited an enrichment for type I IFN-related and IFNG-related transcripts. Accordingly, samples from the LGS cohorts were enriched for a set of IFN-related transcripts previously associated with radioresistance, the so-called “IFN-related DNA damage resistance score” (IRDS)¹⁰. Apparently at odds with this observation, post-RT LGS samples had an increased relative abundance of immune over malignant cells compared to post-RT HGS samples, at least potentially suggesting increased (rather than decreased) radiosensitivity. However, LGS specimens were also enriched in various transcripts associated with active immunosuppression, potentially pointing to an ongoing, but unproductive, immune response⁶.

In summary, Schalck and colleagues demonstrated that a boost dose (biologically equivalent to 7.5 Gy in a single fraction) of focal RT delivered to HR⁺ breast neoplasms prior to surgery elicits immunological changes generally associated with immunosuppression, including (but not limited to) an increase in myeloid over lymphoid cells with a predominance of CXC3CR1⁺ TAMs. This is fundamentally different from preclinical evidence from multiple studies indicating that hypofractionated RT (total dose 24–50 Gy) efficiently converts immunologically silent HR⁺ breast tumors into immunologically active neoplasms that often respond to ICIs⁵.

Of note, we have recently demonstrated that hypofractionated RT (10 Gy × 3 fractions) efficiently prevents subsequently delivered CDK4/6 inhibitors from promoting the repolarization of TAMs towards a CXC3CR1⁺ phenotype associated with resistance to therapy in preclinical models of HR⁺ breast cancer⁸. Such an immunosuppressive TAM phenotype was also associated with the lack of pathological complete responses in patients with HR⁺ breast cancer receiving neoadjuvant PD-1 blockade followed by SBRT (8 Gy × 3 fractions) and concurrent PD-1 blockade^8,9. Conversely, in the dataset from Schalck and collaborators (who examined 7.5 Gy × 1 fraction or 2 Gy × 5 fractions), the relative abundance of CXC3CR1⁺ TAMs increased after irradiation⁶. Taken together, these observations corroborate the critical impact of dose and fractionation on the ability of RT to alter the immunological profile of the TME, which has major implications not only for the activity of RT alone but also for the efficacy of concurrently or subsequently administered therapeutics (Fig. 1).

Fig. 1: Impact of dose and fractionation on the immunological effects of RT.

Accumulating preclinical and clinical evidence indicates that dose and fractionation have a critical impact on the ability of radiation therapy (RT) to reconfigure the immunological microenvironment of solid tumors. This is particularly relevant not only for the clinical activity of RT as a standalone intervention, but also for the efficacy of concomitantly or subsequently delivered immunotherapeutics and other drugs that actively engage tumor-specific immune responses, such as some targeted anticancer agents. That said, it is unlikely that a specific RT dose and fractionation schedule will invariably generate beneficial (A) or detrimental (B) immunological alterations of the tumor microenvironment (TME), with a number of additional parameters (e.g., tumor type, mutational profile, immunological configuration at baseline, etc.) factoring in. CTL cytotoxic T lymphocyte, DC dendritic cell, TAM tumor-associated macrophage, T_REG regulatory T.