Near future of tumor immunology: Anticipating resistance mechanisms to immunotherapies, a big challenge for clinical trials

ABSTRACT

The success of immunotherapies brings hope for the future of cancer treatment. Even so, we are faced with a new challenge, that of understanding which patients will respond initially and, possibly, develop resistance. The examination of the immune profile, especially approaches related to the immunoscore, may foretell which tumors will have a positive initial response. Ideally, the mutation load would also be analyzed, helping to reveal tumor associated antigens that are predictive of an effective cytolytic attack. However, the response may be hindered by changes induced in the tumor and its microenvironment during treatment, perhaps stemming from the therapy itself. To monitor such alterations, we suggest that minimally invasive approaches should be explored, such as the analysis of circulating tumor DNA. When testing new drugs, the data collected from each patient would initially represent an N of 1 clinical trial that could then be deposited in large databases and mined retrospectively for trends and correlations between genetic alterations and response to therapy. We expect that the investment in personalized approaches that couple molecular analysis during clinical trials will yield critical data that, in the future, may be used to predict the outcome of novel immunotherapies.

KEYWORDS:

• cancer
• checkpoint blockade
• circulating tumor DNA
• immunoscore
• personalized medicine

After decades of scepticism, immunotherapies reawakened to occupy a central role in the frontier of innovative cancer therapies. Several experimental demonstrations were critical in unveiling the role of the immune system in cancer control, among which, the identification of tumor associated antigens, the immunosurveillance concept as well as the remarkable success of immune checkpoint blockers (ICB).^1,2 Now immunotherapies are being applied to an expanding variety of solid tumors and we are facing the challenge of understanding why some patients do not respond and how initially sensitive tumors eventually progress. Hopefully, the answers to these questions will lead to new or multimodal therapeutic approaches which may finally extend the magnitude of survival from months to years.

One key step to predict and consequently avoid resistance requires the identification of the immune profile of the tumor. For example, a current consensus classifies colorectal cancer (CRC) in at least 4 distinct molecular subtypes (CMS1 to 4) that are associated with variable prognosis. The subtype CMS1 (microsatellite instability and immune activation features), is characterized by the expression of genes associated with immune infiltrate and have better prognosis.³ Another effort in predicting (CRC) aggressiveness comes from Jerome Galon and colleagues. They have shown that the presence of intratumoral and marginal CD3⁺, CD8⁺, Granzyme B⁺ and/or CD45RO⁺ cells were associated with increased survival in colon cancer, which led to the proposal of an immunoscore of tumors, that has a superior prognostic value than the largely accepted TNM classification.⁴ The immunoscore was validated by an international consortium in more than 3500 patient samples and the promising results were presented in the ASCO meeting this year, showing increased survival of patients with high immunoscored tumors when compared with low immunoscored ones. Beyond the prognosis, the composition of the natural immune reaction influences the clinical response to immunotherapies and such a determination may indicate the most suitable therapeutic intervention for each tumor type.⁵ CMS1, which also harbours a high mutational load, is highly responsive to pembrolizumab, a check point inhibitor, while there is no evidence regarding the efficacy of immunotherapies in the other subtypes.⁶ More recently, clinical trials have yielded encouraging results tyrosine kinase inhibitors of the MEK/ERK signaling pathway were combined with checkpoint inhibitors in patients with microsatellite stable metastatic colorectal cancer with the aim to boost the immune response in a subgroup of patients whose tumors are not likely to respond to immunotherapy.⁷ It is not yet known how immunoscore can be predictive in evaluating the response to immunotherapy and also the level of correspondence between the CMS1 and high immunoscored tumors. These classifications making use of immunologic features represent the first steps toward a personalised immunotherapy.

While the ideal way to understand the molecular mechanisms of resistance is to obtain and compare biopsy samples from baseline with those of progressing lesions, this cannot be easily done because of the risks and pain associated with re-biopsy of visceral metastases and also because a single biopsy may not reflect tumor heterogeneity.⁸ For these reasons, few molecular clues are available that may predict resistance to immunotherapy.

In melanomas, Snyder and colleagues have shown a positive correlation between the mutation load and CTLA-4 blockade, underscoring the importance of neoantigens in tumor rejection.⁹ Even though the predictive response based on mutation load represents an important tool for physicians, the determination of the tumor neoantigen load is still too expensive and complex to be incorporated in routine clinical practice. Illustrating the complexity of the tumor microenvironment, the cross talk between tumor and antitumor immune cells can also lead to resistance. Interestingly, using a mouse model of melanoma, Thomas Tuting and collaborators showed that the inflammation induced by tumor specific cytotoxic T cells is able to induce dedifferentiation of tumor cells and abrogate tumor cell recognition.¹⁰ Also, Antoni Ribas and collaborators recently showed that the late relapse of melanoma in patients treated by PD1 checkpoint blockers are related to alterations in interferon signaling and antigen exposure.¹¹ Our results, in a murine model of lung cancer, also show that even though IFNβ/p19ARF gene transfer reduces tumor growth in a T cell dependent mechanism, PDL1 is upregulated in the tumor, indicating the emergence of compensatory mechanisms.¹²

Thus, ideally, the complexity of tumor dynamics should be assessed continuously during treatment, allowing the early identification of resistance and implementation of countermeasures. Of note, not only do the tumor and its microenvironment seem to influence the response to immunotherapy, but so do host-related factors like age, HLA, diet, metabolism, infections and microbiome (recently revised by Pitt et al).¹³

The increased complexity generated by the genetic, immune and host related heterogeneity represents a challenge for the identification of the optimal treatment combination. Ideally, various steps of the immune response would be therapeutically modulated in association with current therapies, creating a large panel of possibilities. We can expect to face the same difficulties encountered when evaluating targeted therapies, as reviewed by Catenassi.¹⁴ Next generation immune-therapeutic clinical trials should consider strategies to dynamically evaluate the immune response during the treatment and will possibly adopt variations of N of 1 trials in which each patient is a sole unit of observation for determination of efficacy and side effects.

The sequential, multi-step modulation of the immune response will require the development of a reliable panel of biomarkers that indicate an early response to each intervention. To do so in an accelerated fashion and to avoid invasive procedures, the evaluation of circulating nucleotides (ctDNA, circulating tumor DNA) together with the development of in vivo imaging strategies or point of care devices could be a future solution, enabling real time evaluation of tumor immunodynamics and assessment of the tumor heterogeneity.^15,16 The examination of ctDNA is a promising tool to dynamically quantify specific molecular alterations, such as known driver mutations, and study them as markers of residual disease, response and progression.¹⁷ CtDNA could eventually be used to reflect not only the progression, but also the specific characterization of the antitumor immune response.¹⁸ Limitations of the ctDNA method include the determination of the best sensitivity cut-offs for the detection of tumor mutations that have clinical significance, validation of a standard technique and (still) high cost.

In this complex scenario, the planning of new clinical trials must take into consideration the mechanisms of resistance that may be encountered. However, given the large and increasing number of new cancer-directed compounds, it is not timely or financially feasible to conduct randomized trials for all of them in specific, and often rare, tumors with known molecular mechanisms of resistance. Therefore, the establishment of databases for results from single patients treated with molecularly-driven drugs and immunotherapies which could then be pooled to assess preliminary efficacy in a large patient population would be helpful for the early identification of drug efficacy. Moreover, N of 1 trials with collection of ctDNA, and eventually biopsies of the more accessible lesions, throughout the trial would be an intelligent and useful way to prospectively evaluate the biologic progress of treatment-induced resistance mechanisms.

At our own institute, initiatives are in place to increase activities related to immunotherapy, ctDNA analysis and clinical trials. Though such efforts are at early stages, we expect close cooperation between the clinical, clinical research and laboratory research teams that are further supported by an in-house biobank repository. As a public institution, funding is particularly challenging, but measures have been taken to garner support from philanthropic, public and industry sources. The promise of testing and implementing new immunotherapy approaches is certainly challenging, thus critical cooperation between multidisciplinary professionals as well as public and private consortia will be required to achieve this goal.

Because of the complexity stemming for the increasing knowledge of cancer biology, massive funding, and the logistics of running clinical trials, it is clear that the integration of different approaches will demand the cooperation between academia/governments and companies and among industries themselves to expedite the basic research and the development of effective cancer/tumor microenvironment-directed therapies.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Funding

São Paulo Research Foundation grants (Strauss, B.E) 13/25167–5, fellowships (Catani, J.P.P) 14/11524–3; (Adjemian S.) 12/25380–8.