Although immune based therapy has proven efficacious for many patients with microsatellite instability (MSI) high colon cancers, a majority of treated patients see limited benefit. Conversely, select patients with microsatellite stable (MSS) tumors have disease stability after treatment necessitating new and novel strategies to study endogenous immune reactivity in the tumor microenvironment (TME). Using phenotypic and spatial data from infiltrating immune and tumor cells we sought to model cellular mixing in the TME to predict disease specific outcomes in patients with colorectal liver metastases (CRLM).
Formalin-fixed tissue from 195 CRLM were subjected to multiple fluorescent immunohistochemistry (mf-IHC) using tyramide signal amplification allowing phenotyping of tumor cells (TCs), CTLs, regulatory T cells (Tregs), antigen presenting cells (APCs) and measurement of programmed death ligand 1 (PD-L1). After multispectral imaging, cell to cell interactions were analyzed by measurement of engagement, intercellular distance and activation using granzyme B and Ki-67.
Measurement of intercellular distances revealed that tumors with high engagement of CTLs to TCs had improved 5-year disease specific survival (52.3mos vs 42.5mos; p=0.0185). Enhanced CTL activity upon engagement was demonstrated by increased Ki-67 and granzyme positivity compared to unengaged cells (92% and 34%, respectively; p<0.001). Mathematical modeling of individual immune environments was performed by calculation of g-function, a parameter which assesses population mixing as a function of probability of incidence of intercellular distances. Patients with high CTL-TC mixing were more likely male (72% vs 50% for low mixing; p=0.0154) and younger (mean age 57yrs vs 62yrs for low mixing; p=0.0157). They had improved disease specific survival with 42% alive at 5 years compared to only 16% in the low mixing group (p=0.0275). Modeling further revealed that when high levels of PD-L1+ APCs were present in the TME, CTL-TC intercellular distances were increased (mean = 55µm vs 22µm; p=0.0135) and mixing was decreased (g-function 25.1 vs 20.3; p=0.0421).
Evaluation of immune interactions within the TME of metastatic colon cancer using mfIHC in combination with mathematical modeling can characterize cellular mixing of TCs and CTLs providing a novel strategy to better predict outcomes while identifying potential candidates for immune based therapies.