Although immunotherapy has revolutionized lung cancer management, we still need to identify markers to help us better select patients who will benefit from it, thus avoiding toxic treatments and lack of therapeutic benefit in non-responders.
With this objective, we designed a study that included a cohort of patients with advanced and locally advanced lung cancer in which the characteristics of the TCRβ repertoire in mononuclear cells and certain soluble factors were analyzed in peripheral blood samples before the initiation of ICIs in monotherapy or in combination with chemotherapy. Although our study's sample size is limited, the cohort included is representative of the lung cancer population, which may help us draw certain conclusions in this regard.
There is evidence that the analysis of the TCRβ profile provides predictive information on response to ICIs (8). However, the evidence in this regard is quite variable, possibly due to small sample sizes, studies performed in different tumor pathologies, type of immunotherapeutic agent administered and sequencing methods performed (19). One of the advantages of using TCRβ as a biomarker is that it can be determined in an easily accessible sample, such as peripheral blood, without resorting to tumor tissue, which is often scarce in lung cancer (20). Even so, TCRβ has certain limitations regarding its determination, such as the lack of standardization in the cut-off points, the platforms and the cost of the sequencing techniques employed.
In our study, it was observed that there is a trend, although not significant, that the greater the richness or baseline number of different TCRβ clonotypes, the greater the clinical benefit, achieving an increase in OS. Previous studies have shown that treatment with ICIs can have a pharmacodynamic effect by increasing the number of unique TCR clonotypes in peripheral blood and, consequently, a greater possibility of recognizing tumor neoantigens, thus improving therapeutic response (14, 21). This has been proven by analyzing tumor tissue in responder patients with NSCLC receiving treatment with neoadjuvant ICIs, observing tumor tissue enriched with expanded clonotypes (22). Thus, it has been observed that the pre-treatment presence of clones in the tumor shows clonal expansion in blood after treatment, correlating with better response and clearance of circulating tumor DNA (19).
TCRβ diversity refers to the number of clonotypes present. It is to be expected that the greater the diversity of the TCRβ repertoire, the greater the probability that T lymphocytes will recognize tumor antigens and, therefore a best antitumor response. Han, J. et al. 2020 and Huang, A.C. et al. 2017 showed that the diversity of the PD-1 + CD8 + TCR population in blood could indicate that there is a greater proportion of exhausted T cells that can be reactivated with ICI, leading to a more effective immune response in patients with non-small cell lung cancer (23, 24). However, the conclusions reached by previous studies regarding TCRβ diversity are pretty mixed. In fact, in our study, no significant differences in OS or PFS were observed according to the TCRβ Shannon diversity index. However, there was a trend towards better OS after treatment in those with greater TCRβ Shannon diversity. This can be explained by the fact that there are antigen-specific TCRs in the peripheral blood mononuclear cells that are non-tumorigenic and can dilute tumor-specific TCRs (24).
Therefore, our study confirms that a wide repertoire of TCR in blood with greater richness and diversity makes the recognition of tumor antigens more likely and that they are reactivated later with the action of ICIs, thus decreasing the immune escape of tumor cells.
In terms of the convergence analysis, it has been seen that patients with a greater response to ICI have a greater pre-treatment TCR convergence, thus reinforcing the idea that T cells with convergent TCRs target tumor antigens (16). An advantage of TCR convergence as a biomarker is that it is able to detect the T-cell response to tumor neoantigens beyond those originating from nonsynonymous mutations (point mutations that alter the resulting protein sequences). The evidence in this regard is contradictory to what was observed in our study, in which no significant differences in OS or PFS were observed according to TCRβ convergence, and even a tendency was observed that the lower the TCRβ convergence, the higher the OS and PFS. TCR convergence is a process by which a tumor antigen determined by antigenic specificity leads to the expansion of T cells that share TCRs with antigen specificity (same functional TCR) but different amino acid or nucleotide sequences (25). Less convergence means a greater number of different clones with specificity for antigen presented in the HLA molecule. Furthermore, some studies suggest that successful immunotherapy was not reliant on the select expansion of specific T cell clones, but instead induced relatively uniform expansion of most tumor-infiltrating T cells, enhancing effector capabilities (26). However, different therapeutic approaches and distinct cancers could likely yield different results.
It has also been shown that the sequencing method or platform chosen for analysis can vary in determining TCRβ convergence as substitution sequencing errors introduced during sequencing of the TCRβ repertoire can create artifacts that resemble TCR convergence (16). Thus, a study comparing TCRβ sequencing using the Illumina platform with the Oncomine assay was performed and differences were seen between them in the frequency of TCRβ convergence, rearrangements, and detection sensitivity. Both were consistent in detecting TCR clonality and diversity, but Illumina resulted in higher detection of convergent TCRs(16). With prior knowledge of the different substitution error rates in the different sequencing platforms, the most appropriate platform can be chosen accordingly.
In our study, the Oncomine platform was used in which TCR convergence is precalculated from the set of clones reported by the Ion Reporter software that is interpreted as standard. Perhaps the sequencing platform employed with lower detection rate than others, such as Illumina along with the paucity of peripheral blood samples that could be used for TCRβ determination, contributed to these results. Still, further understanding of the mechanisms involved and studies involving larger cohorts are required to consider TCR as a predictive biomarker of response to ICIs.
It has been shown that several factors in peripheral blood prior to initiating treatment with ICI, such as the number of activated CD4 memory T cells and a more clonally diverse TCR repertoire, are associated with the development of severe immune-mediated adverse effects and with a greater response to ICIs (27). In addition, it has been observed that patients receiving treatment with ICIs experience change in TCR clonality that may be related to the severity of the immune-mediated event and the timing of the event (27). In fact, according to other studies, patients with NSCLC with greater increases in PD-1 + CD8 + TCR repertoire clonality intra and post-ICI presentated greater PFS and OS reflecting an expansion of a successful anti-tumoral clonotype. On the contrary, pre-treatment TCR repertoire diversity could be a treatment-agnostic prognostic factor(28)
These findings could be of great utility because the modification in the characteristics of the TCRβ repertoire during treatment could serve as a tool to predict and identify which patients are at higher risk of developing them. In our cohort, it was seen that patients with higher toxicity had higher values of TCRβ convergence (data not showed), and as mentioned, this has been found to be associated with higher response to ICIs.
A proinflammatory gene expression profile in pre-treatment samples is associated with a superior pathologic response after treatment with chemotherapy and immunotherapy (29). Thus, pre-treatment peripheral blood analysis of some cytokines was performed to study whether certain pre-treatment blood soluble factors could predict response or benefit to ICIs.
In our cohort, it was observed that higher pre-treatment levels of IL-2 and IL-15 were associated with more aggressive tumor behavior and worse outcomes: patients survived less and had lower PFS. This is consistent with evidence from other studies (30). In patients with lung cancer, high concentrations of intratumoral IL-15 are associated with a worse prognosis (31). It appears that intratumoral production and/or circulating sIL-15/IL-15Rα complexes contribute to developing a tumor microenvironment favourable for tumor progression and immune escape (32). However, there are exceptions to the behavior described above since, in some solid tumors, the IL-15/IL-15R complex may also play an antineoplastic role (31, 32). Therefore, the role of intratumoral and circulating IL-15 is complex and depends on several factors, such as the type of IL-15 produced, the IL-15-Rα chain isoforms involved in sIL-15/IL-15Rα, the presence of functional IL-15 receptors on tumor cells, as well as their response to stromal and endogenous IL-15. Therefore, it is difficult to say whether it predicts aggressive behavior (32). Although there is no data regarding levels of circulating IL-15 in relation with antiPD1/PDL1 efficacy, a previous study showed that low serum IL-15 levels correlates with better responses to antiCTLA4 treatment in melanoma (31).
Regarding IL-2, this cytokine seems to have a dual effect on the tumor immune microenvironment. On the one hand, ICIs could, by interacting with T cells, increase IL-2 secretion, enhancing the immune response, but recent studies show that IL-2 also induces an immunosuppressive activity by promoting Treg proliferation and activation, which inhibits the antitumor response (34). Therefore, further studies are needed to investigate and clarify the relationship between IL-2 and ICIs.
No statistically significant association was observed in the univariate analysis between IL-10 levels and OS. IL-10 is an immunosuppressive and anti-inflammatory cytokine that regulates the growth and differentiation of different cell types. It is well known that in cancer patients, higher levels of IL-10 in serum correlate inversely with oncologic prognosis (34, 35). Despite this, it has recently been observed that IL-10 may play a role in CD8 + T cell activation and proliferation in cancer and chronic inflammation (34). In addition, IL-10 and PD-1 play immunosuppressive roles through very different pathways (35), and dual blockade has synergistic antitumor action(36, 37). Their efficacy and safety as antitumor therapy has been proven in several studies(39). Considering the heterogeneity of the findings regarding this cytokine, the small size of our sample, and the arbitrary value taken to consider high IL-10 expression (> 2.8), more studies and data are needed to clarify the prognostic significance of IL-10 in the treatment of ICIs.
IFN-γ exerts a dual role. On the one hand, it is a potent inducer of the antitumor immune response, but it can also serve as a tumor escape mechanism. Its direction towards one or the other action will depend on tumor specificity, signal intensity and tumor microenvironment (40). In our work, we found no association between an IFN-γ expression signature and response to ICIs; in fact, responders were observed to have lower levels of MICB, CXCL10 and IFN-γ. Several studies have shown that ICIs increase IFN-γ production contributing to tumor clearance, and it has been demonstrated that resistance to IT could be due to defects in the IFN-γ signalling pathway (40, 41). The IFN-γ and PD-L1 gene signature combination has been associated with greater therapeutic benefit to IT and could constitute a predictive biomarker of response to ICIs (42, 43).
It has been shown in different studies that tumors with higher CXCL10 expression correlate with a better prognosis(45). On the contrary, in our cohort, it was seen that responder patients had lower pre-treatment levels of CXCL10. This soluble factor is involved in T and NK cell mobilization. Thus, patients with low levels of this factor may have a more CXCL10 increase after ICIs treatment and, therefore, a better response. Deep research is needed to confirm this hypothesis.
The important role of NK cells and NKG2D ligands in cancer immunosurveillance suggests that their presence in serum could serve as a prognostic marker (46). Their relationship with survival in cancer patients has been studied (46, 47, 48, 49). The ligands of NKG2D are MICA, MICB and six members of the ULBP family (51). The release of soluble NKG2D ligands represents a form of tumor cell immune evasion strategy since these ligands deregulate NKG2D expression by decreasing NK cell function and T cell activation, and furthermore these soluble ligands compete in receptor binding with ligands expressed on the surface of tumor cells (46). Therefore, it is to be expected that higher levels are associated with worse prognosis and disease progression albeit its relation with ICI response has not been previously determined.
These findings are consistent with our work, in which patients with lower levels of ULBP1, ULBP2, MICB and MICA had better survival outcomes. Higher levels of ULBP1, were significantly associated with lower PFS. Patients with OS ≥ 12 months, had statistically lower levels of MICB, ULBP1 and ULBP4. Although these ligands may play an essential role as predictors of the evolutionary course of cancer, the complex regulation of NKG2D ligands, their variation, and specificity depending on tumor type will have to be taken into account, and a better future understanding of the effects of these soluble factors on immune cells will be necessary (52).
The difference in cytokines serum level values between responders and non-responders could be appreciated as small. However, serum determination is an indicator of the different levels in the tumor microemviroment (TME), meaning the differences could be higher in the TME. Other publications report similar or minor differences between healthy and patient serum sample and between treated and non-tretated patients with statistical differences and biological significance (53). In our research, serum determinations were performed in all patients before starting treatment, so it is normal not to see large differences. However, these slight variations seem to help predict the response to ICIs. Of course, these findings need to be corroborated by subsequent studies using a bigger cohort of patients, but if so, they would be a valuable tool in clinical practice. Other publications in pediatric autoimmune hepatitis also detected minor differences between groups and showed how IL-2 levels predicted treatment response (54). The relevance of our research just lies in the relatively few studies that have been published describing the relationship between cytokines and ICIs responses.
Concerning markers predictive of immune toxicity during treatment, in our study, an association was observed between IL-15 and MICB expression and the development or not of immune-mediated toxicity (data not showed); thus, those who did not present toxicity had higher levels of IL-15 and MICB and lower survival. Therefore, although more evidence is needed, it would be interesting to validate these findings in future studies and to observe the role of these soluble factors as predictors of immune toxicity and thus, of response to ICIs.