Meta-analysis and metagenes

Integrative gene-expression analysis applied to the study of human samples has defined shared themes invariably associated with immune-mediated tissue destruction. Such themes define an in situ T helper 1 (Th-1)-like immune response characterized by the coordinate expression of interferon-ү (IFN-ү)/interferon regulatory factor-1 (IRF-1)-induced transcripts, mRNA’s encoding CXCR3/CCR5 chemokine ligands (i.e., CXCL9–11, CCL3–5), and those encoding immune-effector functional molecules (e.g., granzymes, granulysin [GNLY], and perforin [PRF1]). We refer to these collectively as the immunologic constant of rejection (ICR) pathways.^1-4 Their upregulation has been observed in a plethora of different immune-related conditions, spanning from allograft rejection to flares of autoimmunity.^1-3 In the context of cancer immunotherapy, an efficient induction of these molecular pathways early after treatment correlates with achievement of favorable clinical response later.^5,6 Moreover, patients bearing metastatic tumors that display this polarized immunophenotype respond better to various forms of immunity-related manipulations, including interleukin 2 (IL-2) treatment,⁵ adoptive-therapy,⁷ vaccines,⁴ and anti-cytotoxic T lymphocyte-associated protein 4 (CTLA-4) monoclonal antibody (mAb).⁶ In addition, the presence of largely overlapping gene-signatures has been convincingly associated with favorable prognosis of melanoma, colon, breast, and ovarian cancers, as recently reviewed elsewhere.³ Observations from immune-checkpoint inhibitor trials (e.g., anti-CTLA-4 and anti-programmed cell death 1 [PD-1] mAbs⁸) have also revealed that this inflammatory status is accompanied by the concomitant counteractivation of immunosuppressive mechanisms (i.e., indoleamine-pyrrole 2,3-dioxygenase [IDO] and PD ligand 1 [PD-L1]),^6,9 which likely reflect ongoing immune-escape processes. This suppressed immune response could eventually be reverted by the administration of immune checkpoint inhibitors. However, tumors lacking these 2 features are relatively resistant to immunotherapeutic manipulations.^5-7,9 Because of the pivotal role of the Th-1 like response in mediating tumor rejection, targeted therapy aimed at reprogramming the tumor microenvironment by inducing ICR pathways are the object of effervescent investigations.^3,10,11

In the meta-analysis recently published in OncoImmunology by Stoll et al.,¹² investigators determined whether robust gene-signatures (metagenes) are predictive of beneficial response to neoadjuvant chemotherapy among breast cancer patients. Authors employed a hypothesis-driven approach based on a strong rationale. Rather than testing the expression of the whole transcriptome, they focused on the identification of metagenes (i.e., modules of genes strongly correlated) underlying phenomena conducive to tumor rejection. Gene expression levels evaluated to build metagenes consisted of a wide range of transcripts reflecting intratumoral immune responses, as well as genes induced in response to local stress (i.e., endoplasmic reticulum [ER] stress-, autophagy-, and lysosome-associated transcripts). In fact, as demonstrated by previous works,¹³ the induction of ER stress and autophagy by certain chemotherapeutic agents can drive tumor cells toward an immunogenic form of apoptosis (known as immunogenic cell death) conducive to the elicitation local immune response.¹⁴ For each category, metagenes were delineated by exploring The Cancer Genome Atlas breast cancer data set. Their reproducibility was assessed using 6 additional cancer data sets including colorectal, head and neck, and other breast cancer samples. Among these were 3 breast cancer data sets for which information in regards to clinical neoadjuvant chemotherapy response was available that were used to test metagenes predictive capabilities. The reproducibility of stress-related metagenes was found to be generally poor—perhaps suggesting a lack of coordinated and persistent stress-related events in pre-treatment tumor deposits—such that no association between these metagenes and clinical outcome was detected. However, the immune-related metagene driven by the C-X-C motif chemokine ligand, CXCL13 transcript bore the highest reproducibility across data sets and was strongly associated with the achievement of complete pathological response. Genes embraced by the CXCL13 metagenes¹² largely overlap with those associated with favorable cancer prognosis and response to immunotherapy.^3,4 They include classical ICR genes such as ligands for the chemokine receptors CXCR3 and CCR5 (CXCL9–10, and CCL5 transcripts, respectively), immune-effector genes (e.g., PRF1, granzymes), and Th-1 related genes (IFNG and CD8B).¹² To explain the favorable predictive role of Th-1 like gene signatures in the setting of cancer immunotherapy, it has been proposed that immune-manipulation could restore a naturally occurring, though insufficient, host’s immune response by enhancing its effector functions and, thus, its predictive significance.⁴ Similarly, it is tempting to hypothesize that the immunogenic cell death, eventually induced by antineoplastic drugs, requires the presence of an ongoing intratumoral response to exert its immune-adjuvant effect.

A number of investigations have reported a correlation between T-cell infiltrates, Th-1 related genes, and achievement of complete response following neoadjuvant chemotherapy in breast cancer patients.¹⁵ By defining the CXCL13-meta gene, Stoll et al. added molecular precision to previous observations.^15,16 In fact, only recently, has CXCL13 emerged as critical modulator of intratumoral response.^17-19 This chemokine, which binds CXCR5, is physiologically highly expressed in the follicles of secondary lymphoid organs, where it can be secreted by follicular dendritic cells and T follicular helper (Tfh) cells. In this context, CXCL13 mediates migration of high-affinity CXCR5+ Tfh cells and B cells into B-cell concentrated areas. While a number of studies have reported the presence of tertiary lymphoid structure in a considerable proportion of cancers,²⁰ it was only last year that the presence of CXCL13+ Tfh cells were demonstrated in solid tumors.¹⁸ By analyzing breast cancer samples, Gu-Trantien et al.¹⁷ showed that the presence of tumor-infiltrating CXCL13+ Tfh cells, localizing primarily in peritumoral tertiary lymphoid structures, was associated with improved disease outcome. In parallel, the presence of Tfh cells were shown to correlate with abundance of Th1 cells and B cells within the neoplastic bed.¹⁷ Similar conclusions were recently independently reached by Bindea et al.¹⁹ via analysis of colorectal tumor specimens. Interestingly, authors showed that tumor cells also expressed CXCL13 and that genetic deletion of CXCL13 markedly lowers the density of B cells and Tfh cells in invasive margins.¹⁹ It remains, however, to be fully elucidated whether (and how) the genetic makeup of the host, somatic cancer cell genetic or epigenetic aberrations, or environmental factors, such as lifetime exposure to commensale microbiota,²¹ may interact to influence the development of a favorable cancer immune phenotype. We believe that assessing these critical questions using integrated high-throughput approaches will allow the development of innovative targeted therapy that may dramatically impact therapeutic outcome in the near future.²²

Glossary

Abbreviations:

CTLA-4

cytotoxic T lymphocyte-associated protein 4

ER

endoplasmic reticulum

GNLY

granulysin

IDO

Indoleamine-pyrrole 2,3-dioxygenase

IFN-ү

interferon-ү

IL-2

interleukin 2

IRF-1

interferon regulatory factor-1

mAb

monoclonal antibody

PD-1

programmed cell death 1

PD-L1

PD ligand 1

PRF1

perforin

Tfh

T follicular helper

Th-1

T helper 1

Citation: Bedognetti D, Wang E, Marincola FM. Meta-analysis and metagenes: CXCL-13-driven signature as a robust marker of intratumoral immune response and predictor of breast cancer chemotherapeutic outcome. OncoImmunology 2014; 3:e28727; 10.4161/onci.28727