Good TIMEs: Using the Tumor Microenvironment to Predict Response to Treatment

Tumors are not merely composed of cancer cells but, in fact, consist of an intricate and dynamic ecosystem made up of a heterogeneous bunch of cells and structures, which are collectively referred to as the tumor microenvironment (TME). The TME includes the cancer cells themselves, infiltrating immune cells, stromal cells, and the extracellular matrix (Anderson and Simon 2021). The interactions between tumor cells and the surrounding immune cells are referred to more specifically as the tumor immune microenvironment or TIME.

These components actively communicate with one another and shape not only the evolution of the cancer but also influence disease progression and the response to treatments such as immunotherapy. Most patients with head and neck squamous cancer (HNSC), for example, do not respond to immunotherapy, so information on factors that predict response would be beneficial when it comes to prioritization of expensive treatments and avoiding needless toxic side effects (Napolitano et al. 2019).

In this blog, we discuss a recent study by Misetic et al. (2023) in which they investigated the mechanisms behind the immune-regulatory role of the TIME and how components of the TIME could be used as biomarkers to predict response to immunotherapy.

Determining the Driving Force

The researchers used computational techniques to analyze a dataset of 7,730 samples from 32 different solid cancer types. Interactions between tumor cells and the surrounding immune cells tend to involve genes that drive cancer evolution, aptly known as cancer drivers. The researchers began by compiling a list of cancer drivers based on both experimentally validated and computationally predicted methods and further streamlining this list to include only drivers with damaging alterations in the 7,730 samples.

Next, they developed a method to characterize the features of the TIME of these tumors. The scientists used five gene signatures that would indicate the levels of overall tumor immune infiltration (IS), cytotoxic antitumor infiltration (CYS), antitumor T helper activity (ICR), antitumor inflammation state (TIS), and cancer-promoting inflammation (CPI) of each TIME.

Combining these two steps, Misetic et al. were able to calculate each cancer driver’s probability to predict each feature of the TIME across the different cancer types. They identified an impressive total of 477 drivers whose damage predicted higher or lower levels of the TIME features. Moreover, the identified drivers of the TIME were often recurrently damaged across different cancer types and frequently interacted with more than one TIME feature. Further analysis revealed that these TIME drivers were involved in significantly more biological pathways, particularly immune-related pathways, than non-TIME drivers.

Hot and Cold

Given that the traits of a TIME have an impact on the tumor’s response to treatment, the researchers hypothesized that the analysis of the TIME drivers could also predict responsiveness to immune checkpoint blockade (ICB) therapy, a class of drugs that act to reinvigorate the immune system to attack cancer cells.

Tumors can be classified into two broad subtypes: hot tumors and cold tumors. Hot tumors refer to those that have a high level of T cell infiltration and inflammatory activity, whereas cold tumors are characterized by an absence of cytotoxic T cells and the presence of immunosuppressive cells, such as myeloid-derived suppressor cells (MDSCs), regulatory T cells (Tregs), and tumor-associated macrophages (TAMs). Hot tumors tend to be more responsive to ICB whilst cold tumors rarely respond to ICB (Liu and Sun, 2021).

By analyzing two of the main gene groups where mutations induce cancer, tumor suppressors (encoding proteins with antitumor properties) and oncogenes (encoding proteins with protumor properties), the scientists found that mutations in suppressors were preferentially enriched in TIME drivers that predicted a hot tumor environment, likely enabling the tumors escape from the immune system. In contrast, mutations in oncogenes were enriched in TIME drivers predictive of a cold tumor environment, suggesting these alterations favor tumor growth in a protumor TIME.

Both the number of antitumor TIME drivers (known as the antitumor TIME driver burden or anti-TBD) and the tumor mutational burden (TMB) of a cancer type were able to be used independently as significant predictors of response to ICB, with the combination of the two increasing the predictive power even more so. Additionally, antitumor TIME tumor suppressor burden (anti-TTB) was better able to predict treatment response compared to the TIME oncogene burden.

Untangling the Web

Head and neck squamous cell carcinomas (HNSC) form a group of genetically mixed cancers, subdivided into those caused by human papillomavirus (HPV+ HNSC) and those caused by smoking (HPV- HNSC). Since HNSC shows variable responses to ICB treatment, the researchers set out to determine whether aspects of the TIME would be predictive of treatment efficacy in this cancer type.

The authors of the study were able to rebuild the regulatory networks of 1,443 transcription factors to determine the relationship between alterations in different drivers and the resulting TIME features in the HNSC subtypes. It was observed that alteration in the oncogene DNMT3B was predictive of a hot tumor environment and was recurrently damaged in HPV+ HSNCs.

HPV- HNSCs show more genomic instability than the HPV+ subtype, with a higher frequency of genomic mutations, and so were further subdivided into those with high and low copy number alterations (CNA^high and CNA^low, respectively). The computational analysis revealed that damage to the tumor suppressor CASP8 was enriched in the HPV- CNA^low class of HSNC and was predictive of antitumor activity, whereas the gain of function (GOF) alteration of the oncogene TERT was enriched in the HPV- CNA^high class and corresponded with a cold TIME. Overall, these data suggest that patients with HPV+ and HPV- CNA^low HSNC should be prioritized for ICB treatment, due to the hot TIME conferring an increased likelihood of effectiveness.

The TIME contains a multitude of complex interactions that can sustain or dampen the development of the tumor; hence, the ability to untangle this web is an exciting step in cancer research. Not only do these computational techniques present the potential for patient prioritization for ICB treatment, but they could also facilitate in elucidating the mechanisms behind tumor growth and survival, therefore providing novel signaling pathways that could be targeted for cancer therapies.