Lung cancer is one of the most common cancers with high
mortality
worldwide despite the development of molecularly targeted therapies
and immunotherapies. A significant challenge in managing lung cancer
is the accurate diagnosis of cancerous lesions owing to the lack of
sensitive and specific biomarkers. The current procedure necessitates
an invasive tissue biopsy for diagnosis and molecular subtyping, which
presents patients with risk, morbidity, anxiety, and high false-positive
rates. The high-risk diagnostic approach has highlighted the need
to search for a reliable, low-risk noninvasive diagnostic approach
to capture lung cancer heterogeneity precisely. The immune interaction
profile of lung cancer is driven by immune cells’ distinctive,
precise interactions with the tumor microenvironment. Here, we hypothesize
that immune cells, particularly T cells, can be used for accurate
lung cancer diagnosis by exploiting the distinctive immune–tumor
interaction by detecting the immune-diagnostic signature. We have
developed an ultrasensitive T-sense nanosensor to probe these specific
diagnostic signatures using the physical synthesis process of multiphoton
ionization. Our research employed predictive in vitro models of lung
cancers, cancer-associated T cells (PCAT, MCAT) and CSC-associated
T cells (PCSCAT, MCSCAT), from primary and metastatic lung cancer
patients to reveal the immune-diagnostic signature and uncover the
molecular, functional, and phenotypic separation between patient-derived
T cells (PDT) and healthy samples. We demonstrated this by adopting
a machine learning model trained with SERS data obtained using cocultured
T cells with preclinical models (CAT, CSCAT) of primary (H69AR) and
metastatic lung cancer (H1915). Interrogating these distinct signatures
with PDT captured the complexity and diversity of the tumor-associated
T cell signature across the patient population, exposing the clinical
feasibility of immune diagnosis in an independent cohort of patient
samples. Thus, our predictive approach using T cells from the patient
peripheral blood showed a highly accurate diagnosis with a specificity
and sensitivity of 94.1% and 100%, respectively, for primary lung
cancer and 97.9% and 94.4% for metastatic lung cancer. Our results
prove that the immune-diagnostic signature developed in this study
could be used as a clinical technology for cancer diagnosis and determine
the course of clinical management with T cells.