Titolo della tesi: The predictive role of immune profile in patients with solid tumours in treatment with immunotherapy: a network analysis of efficacy and toxicity
Background:
Despite immunotherapy has deeply changed the treatment landscape and prognosis of several cancers, only a small percentage of patients achieve long-term benefit in terms of overall survival (OS). In addition, ICIs have particular immune-related adverse events (irAEs). Soluble immune profiles, resulting from the combined evaluation of circulating checkpoints, adhesion and inflammatory molecules, rather than the evaluation of individual marker, could be considered as a portrait of the immune system fitness of each patient at the baseline, which affects the response to treatment and the development of toxicities. The aim of this study was to define an immune profile predicting outocomes to ICIs and irAE development.
Methods:
A prospective, multicenter study evaluating the immune profile of patients with advanced cancer, treated with ICIs was performed. The immune profile was studied evaluating circulating concentration of 12 cytokines, 5 chemokines, 13 soluble immune checkpoints (sIC), 3 adhesion molecules and indoleamine 2,3-dioxygenase (IDO) at the baseline (T0) through a multiplex assay. Four connectivity heat maps and networks were obtained by calculating the Spearman correlation coefficients, according to the response to immunotherapy and onset of cumulative toxicity: responder patients without toxicity, non-responder with toxicity, responder with toxicity, non-responder without toxicity. Then, connectivity heat maps were defined for OS and progression-free survival (PFS) using the median values as cut-off.
Results:
Immune profile of 53 patients with advanced solid tumours treated with ICIs was evaluated at T0. Cumulative toxicity occurred in 18 patients (34%). A subgroup of non-small cell lung cancer (NSCLC) patients with high cytokine/chemokine concentrations was identified in non-responder patients without toxicity. A statistically significant up-regulation of IL17A and all the adhesion molecules in non-responder patients with toxicity with respect to the other ones was detected. CTLA4 was significantly higher in non-responders with toxicity compared to both responders and non-responders without toxicity, as well as sCD80 compared to both non-responders without toxicity and responders with toxicity. Four connectivity maps in responder and non-responder patients with and without toxicity were defined. In patients with toxicity, we observed a clearly different connectivity patterns with a loss of connectivity of mostly of sICs and cytokines correlations. In non-responder patients with toxicity, an inversion of the correlation for some adhesion molecules was observed (from positive or null correlation became negative). Four corresponding connectivity networks were built. Only 14 connections among the four networks were common, while connections specifically observed for each group of patients were: 26 in responder with toxicity, 38 in responder without toxicity, 80 in non-responder with toxicity, 31 in non-responder without toxicity. IL10, IL8, IL4, IL6, INFgamma, INFalpha, TNFalpha, GM-CSF, MIP-1alpha, IL13, sLAG3, sTIM3, sCD27, sCD28, sCD 137 and sPDL-2 showed a statistically significant down-regulation in patients with a longer OS. IL10, IL12p70, GM-CFS and sCD27 were statistically significant down-regulated in patients with longer PFS. No connectivity differences were instead observed when we compared the correlations maps between patients with OS and PFS above and below the respective median values.
Conclusions:
The combined evaluation of soluble molecules, rather than a single circulating factor, may be more suitable to represent the fitness of the immune system status in each patient and could allow to identify different prognostic and predictive outcome profiles. A specific connectivity model for each of the 4 clinical situation patterns, based on response to immunotherapy and irAE onset, was defined by a network analysis. Moreover, an organ-dependent immunity has also been highlighted. In patients who develop irAEs and in patients with the worst prognosis (non-responders who will develop toxicity), peculiar connectivity network of immune dysregulation was defined, which could facilitate their early and timely identification. The detection of these specific immune profiles before treatment, if confirmed in a larger patient population, could lead to the design of a personalized treatment approach fit to the peculiar characteristic of patient immune status, to improve outcomes and preventing avoidable irAEs.