Abstract Early detection of cancer can increase survival rates by 5-10 times. In the US, a significant portion of cancers are still detected at a late stage, for example, 32.1% of breast cancer, 45.5% of lung cancer, 22.8% of colon cancer patients were diagnosed at advanced stage. Early detection is challenging because many cancers are asymptomatic in early stages, although cancer screening can detect cancer before symptoms emerge. Current standard-of-care screening programs are limited and directed to single types of cancer. Non-invasive multi-cancer early detection (MCED) is urgently needed because it can screen for several cancer types with a single blood test, increasing cancer screening coverage while avoiding cumulative false positive rates. Developmental MCED approaches usually rely on detection of circulating tumor DNA (ctDNA) released from tumors. However, ctDNA signals scale with tumor burden limiting sensitivity and specificity, particularly for early-stage cancers, while also limiting applicability to precision oncology. Here we report a MCED approach with superior sensitivity and specificity, which can also be used to predict treatment response. We introduce a novel strategy for detecting cancers from blood plasma, profiling the immune response to disease development rather than relying on detection of emerging diseased tissue. The platform is designed to detect alterations in the ratio of immunoglobulins and albumin, as well as class-switching among immunoglobulins, by measuring the overall concentration of amino acid residues incorporated into proteins in patient plasma. By using our biorthogonal chemistry labeling tool, we were able to measure and characterize amino acid concentration signatures (AACS) within the neat blood samples, including cysteine, free cysteine, lysine, tryptophan and tyrosine, which allowed us to observe distinct signatures for varying cancers including breast, prostate, colorectal and pancreatic cancers. We could identify 84% of cancers with <0.5% false positive rate (N=93). In this study, cancer AACS signatures were distinct from signatures measured with non-cancerous immune activation, including autoimmune diseases and infection. Moreover, AACS was also correlated to clinical features, such as cancer metastatic statuses and therapy response. With Cyclin-dependent Kinase 4/6 Inhibitors-treated advanced breast cancer patients as an example, we achieved 100% correct prediction of responding patients and 91.7% accurate predictions of non-responding patients by using AACS in treatment-naïve samples. Altogether, the novel AACS approach is a powerful combination of bioorthogonal chemistry and machine learning analysis in clinical practice, and it can be potentially used for cancer screening and support clinical decisions for treatment selection and patient stratification in new indications. Citation Format: Cong Tang, Patrícia Corredeira, Sandra Casimiro, Wesley Sukdao, Luís Costa, Emma Yates, Gonçalo Bernardes. Immune activation characterization via amino acid concentration signatures for multi-cancer early detection and CDKi treatment response prediction [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 1268.