Recent advancements in immune checkpoint inhibitors (ICIs) targeting PD-1 or CTLA4 have shown significant success in treating various advanced cancers and tumor types. However, a substantial number of patients fail to respond to treatment, and currently there are no consensus biomarkers predictive of response¹. In addition, the wide variety of inhibitors have different responses and there is little information for clinicians to select the most appropriate therapy. Even widely used biomarkers such as PD-1 have predictive limitations. While some advances have been made in tissue based biomarkers such as tumor mutational burden (TMB), there is still a large unmet need to identify biomarkers using liquid biopsies. This article explores the potential use of serum/plasma glycoproteins as prognostic and predictive biomarkers.

Unlocking the Potential of Glycoproteomics

Academic literature has made clear the utility of glycoproteins in understanding cancer etiology, disease progression, and modulation of the immune system². The study of glycoproteins has long relied on mass spectrometry, but the complexity of the data generated made it virtually impossible to study the glycoproteome quantitatively at scale. Leveraging artificial intelligence to accelerate data processing³, InterVenn has made it possible to interrogate this layer of biology and identify crucial information that is missing from the existing -omics landscape.

The quantitative glycopeptide profiles generated by this platform were assessed in a study involving 202 advanced melanoma patients seen at Massachusetts General Hospital⁴. Plasma samples were collected prior to initiation of either anti-PD-1 monotherapy or anti-PD-1/anti-CTLA-4 combination therapy. 64 peptides and glycopeptides displayed significant associations with overall survival (FDR < 0.05) and thirteen of these biomarkers were applied to generate a multivariable classifier that successfully predicted treatment response. In held-out validation and test set data, responders had an OS hazard ratio of 3.2 (p = 8.2E-04) compared to non-responders (Figure 1). This was validated in an independent cohort with HR = 5.6 (p = 0.027).

Similar research conducted in non-small cell lung cancer (NSCLC)⁵ involved 125 patients with a diagnosis of unresectable stage 3 or 4 NSCLC. Baseline samples were run on the platform to assess glycopeptide associations with response to pembrolizumab monotherapy or combination pembrolizumab-chemotherapy. 70% of the cohort was utilized in training a multivariable classifier, utilizing distinct markers from the melanoma context; in the hold-out test set, strong predictive performance was observed (HR = 3.86, p < 0.01). This classifier separated patients likely benefiting from ICI therapy from those likely not benefiting from ICI therapy, with median OS of 23.2 vs. 5.9 months, respectively.

Beyond Prognostic Biomarkers

InterVenn’s published work in immuno-oncology (IO) has predominantly addressed patient response in specific indications across a given class of therapeutics. Work to be published in Fall 2024 will establish that glycoproteins could be used as predictive biomarkers to select between IO agents and combination therapy. These results underscore the potential impact and clinical utilization of circulating glycoprotein expression profiles.

The evidence for glycoproteins’ predictive power in IO response is robust, across multiple indications. As the field of ICI therapy continues to evolve, the ability to personalize IO treatment decisions based on predictive biomarkers will be crucial for improving patient outcomes. These studies suggest glycoproteomics will be an important part of realizing the promise of precision medicine.

References

1. Yang F, Wang JF, Wang Y, Liu B, Molina JR. Comparative Analysis of Predictive Biomarkers for PD-1/PD-L1 Inhibitors in Cancers: Developments and Challenges. Cancers (Basel). 2021 Dec 27;14(1):109. doi: 10.3390/cancers14010109. PMID: 35008273; PMCID: PMC8750062.
2. He, K., Baniasad, M., Kwon, H. et al. Decoding the glycoproteome: a new frontier for biomarker discovery in cancer. J Hematol Oncol 17, 12 (2024). https://doi.org/10.1186/s13045-024-01532-x
3. Wu Z, Serie D, Xu G, Zou J. PB-net: automatic peak integration by sequential deep learning for multiple reaction monitoring. J Proteomics (2020) 223:103820. doi: 10.1016/j.jprot.2020.103820
4. Pickering C, Aiyetan P, Xu G, Mitchell A, Rice R, Najjar YG, et al. Plasma glycoproteomic biomarkers identify metastatic melanoma patients with reduced clinical benefit from immune checkpoint inhibitor therapy. Front Immunol. 2023;14:1187332.
5. Lindpaintner K, Srinivasan A, Mitchell A, et al. A novel, highly accurate liquid biopsy-based glycoproteomic predictor of checkpoint inhibitor treatment benefit in advanced non-small cell lung cancer. Journal for ImmunoTherapy of Cancer 2022;10: doi: 10.1136/jitc-2022-SITC2022.0158