Article Text

Download PDFPDF

217 Exploring glycosylation-dependent circuits for prediction of antitumor immune response and immunotherapy outcomes in melanoma and colorectal cancer
  1. Joaquin P Merlo1,2,
  2. Yamil D Mahmoud1,
  3. Florencia Veigas1,
  4. Marco A Scheidegger1,
  5. Alejandro J Cagnoni1,
  6. Elmer A Fernandez3,
  7. Gabriel A Rabinovich1 and
  8. Karina V Marino1
  1. 1IBYME-CONICET, CABA, Buenos Aires, Argentina
  2. 2Universidad Argentina De la Empresa, Buenos Aires, CABA, Argentina
  3. 3CIDIE-CONICET, Cordoba, Cordoba, Argentina
  • Journal for ImmunoTherapy of Cancer (JITC) preprint. The copyright holder for this preprint are the authors/funders, who have granted JITC permission to display the preprint. All rights reserved. No reuse allowed without permission.


Background Although immunotherapies have emerged as efficient strategies for treating cancer, biomarkers for outcome prediction are still under debate. Relevant biomarkers such as PD-L1, tumor mutational burden, and microsatellite instability (MSI) have shown limited efficacy as stand-alone markers.1 2 Moreover, aberrant glycosylation has been associated to tumor development, but its potential to predict immunotherapy outcomes has been barely explored.3 We studied the expression of glyco-immune genes looking for patterns that could provide information on glycobiological determinants associated with response to immunotherapy.

Methods TCGA data was obtained from GDC portal4 and immunotherapy-treated patients’ bulk transcriptomic data from the ENA (PRJEB23709).5 Deconvolution of the tumor microenvironment was performed using MIXTURE with LM22 signature.6 Gene signatures were obtained from MSigDB7 or literature.8 CMS and MSI classification were obtained from literature.9 Pair-wise comparisons were performed using Wilcoxon test. All data analyses were performed with R software v4.0.3.

Results TCGA-SKCM samples were clustered using glyco-immune genes, resulting in two glycoclusters (GCs). GC2 showed higher overall survival and a highly immunoactive profile (figure 1). From these profiles, we developed a Glyco-Immune Signature (GIS, 18 genes) and validated it using baseline biopsies from immunotherapy-treated patients. These patients recapitulated previously found profiles and responders were enriched in GC2 patients (figure 2). Our signature positively correlated with published response signatures and negatively correlated with resistance signatures. Finally, by building proportional-hazard models, we found that adding the GIS score to resistance scores, increased the capacity to discriminate between high and low-risk patients. To explore the potential of this GIS scores in other malignancies, we expanded our analysis to colorectal cancer. The GIS was associated with MSI status and predicted response to immunotherapy, with MSI-H patients and responders showing higher GIS score in TCGA-COAD data (figure 3). Moreover, high GIS-scoring patients presented a ‘hot’ tumor microenvironment and higher scores of immune-related signatures. Also, when analyzing the correspondence with CMS classification, CMS1 patients were high GIS-scoring patients and CMS2 patients were low GIS-scoring patients (figure 4). Interestingly, ~40% of MSI-L/MSS patients showed high GIS scores.

Conclusions Results show that this Glycoimmune signature could serve as surrogate marker for current clinical biomarkers associated with immunotherapy response.


  1. Parvini S, Majidpoor J, Mortezaee K. The impact of PD-L1 as a biomarker of cancer responses to combo anti-PD-1/CTLA-4. Pathology, research and practice, 2023;247:154583.

  2. Switzer B, Puzanov I, Skitzki JJ, Hamad L, Ernstoff MS. Managing Metastatic Melanoma in 2022: A Clinical Review. JCO oncology practice, 2022;18(5):335–351.

  3. Girotti MR, Salatino M, Dalotto-Moreno T, Rabinovich GA. Sweetening the hallmarks of cancer: Galectins as multifunctional mediators of tumor progression. The Journal of experimental medicine, 2020;217(2):e20182041.

  4. Grossman RL, Heath AP, Ferretti V, Varmus HE, Lowy DR, Kibbe WA, Staudt LM. Toward a Shared Vision for Cancer Genomic Data. The New England journal of medicine, 2016;375(12):1109–1112.

  5. Gide TN, Quek C, Menzies AM, Tasker AT, Shang P, Holst J, Madore J, Lim SY, Velickovic R, Wongchenko M, Yan Y, Lo S, Carlino MS, Guminski A, Saw RPM, Pang A, McGuire HM, Palendira U, Thompson JF, Rizos H, … Wilmott JS. Distinct Immune Cell Populations Define Response to Anti-PD-1 Monotherapy and Anti-PD-1/Anti-CTLA-4 Combined Therapy. Cancer cell, 2019;35(2):238–255.e6.

  6. Fernández EA, Mahmoud YD, Veigas F, Rocha D, Miranda M, Merlo J, Balzarini M, Lujan HD, Rabinovich GA, Girotti MR. Unveiling the immune infiltrate modulation in cancer and response to immunotherapy by MIXTURE-an enhanced deconvolution method. Briefings in bioinformatics, 2021;22(4):bbaa317.

  7. iberzon A, Birger C, Thorvaldsdóttir H, Ghandi M, Mesirov JP, Tamayo P. The Molecular Signatures Database (MSigDB) hallmark gene set collection. Cell Syst. 2015 Dec 23;1(6):417–425.

  8. Jerby-Arnon L, Shah P, Cuoco MS, Rodman C, Su MJ, Melms JC, Leeson R, Kanodia A, Mei S, Lin JR, Wang S, Rabasha B, Liu D, Zhang G, Margolais C, Ashenberg O, Ott PA, Buchbinder EI, Haq R, Hodi FS, Boland GM, Sullivan RJ, Frederick DT, Miao B, Moll T, Flaherty KT, Herlyn M, Jenkins RW, Thummalapalli R, Kowalczyk MS, Cañadas I, Schilling B, Cartwright ANR, Luoma AM, Malu S, Hwu P, Bernatchez C, Forget MA, Barbie DA, Shalek AK, Tirosh I, Sorger PK, Wucherpfennig K, Van Allen EM, Schadendorf D, Johnson BE, Rotem A, Rozenblatt-Rosen O, Garraway LA, Yoon CH, Izar B, Regev A. A Cancer Cell Program Promotes T Cell Exclusion and Resistance to Checkpoint Blockade. Cell. 2018 Nov 1;175(4):984–997.e24. doi: 10.1016/j.cell.2018.09.006. PMID: 30388455; PMCID: PMC6410377.

  9. Komor MA, Bosch LJ, Bounova G, Bolijn AS, Delis-van Diemen PM, Rausch C, Hoogstrate Y, Stubbs AP, de Jong M, Jenster G, van Grieken NC, Carvalho B, Wessels LF, Jimenez CR, Fijneman RJ, Meijer GA; NGS-ProToCol Consortium:. Consensus molecular subtype classification of colorectal adenomas. J Pathol. 2018 Nov;246(3):266–276.

Abstract 217 Figure 1

Immune infiltrate composition analysed with MIXTURE, using LM22 signature (top panel) and gene signature scores calculated as the geometric mean of the genes composing each signature (bottom panel) for TCGA-SKCM cohort samples grouped by glycocluster classification. Each value was scaled and is shown as Z-score. Top annotation shows several molecular and clinical traits for these samples: 1) barplot showing overall survival; 2) color- coded glycocluster classification; 3) color-scaled intratumoral heterogeneity (ITH), tumor mutational burden (TMB), and copy number variation load (CNV); 4) color-coded mutational status for 5 key driver genes, BRAF, KRAS, NRAS, HRAS, and NFI; 5) scatter plot showing absolute infiltrate score. Statistical significance was set to p<0.05 (*: p<0.05, **: p<0.01, ***p<0.001, ****: p<0.0001, ns: p>0.05). GC2 patients showed higher overall survival rates and a significantly higher proportion of naive and memory B cells, plasma cells, CD8 T cells, CD4 activated T cells, and M1 macrophages, with a lower proportion TMB, cytolytic score, interferon-y, and angiogenesis scores, with lower CNV and ITH.

Abstract 217 Figure 2

Stacked barplots show the proportion of GC1 and GC2 patients in non-responders and responders, and boxplot show the distribution of GIS score for patients treated with immunotherapy in the validation cohort (PRJEB23709). Statistical significance was set to p<0.05 (*: p<0.05, **: p<0.01, ***: p<0.001, ****p<0.0001, ns: p>0.05).

Abstract 217 Figure 3

Stacked barplots show the proportion of (A) high and low GIS-scoring patients in MSI-H, MSI-L and MSS patients, or (B) in patients predicted to respond to immunotherapy or not by the TIDE algorithm. Boxplot show the distribution of (C) GIS score by MSI status or by (D) response as predicted by TIDE algorithm. Statistical significance was set to p<0.05 (*: p<0.05, **: p<0.01, ***: p<0.001, ****: p<0.0001, ns: p>0.05).

Abstract 217 Figure 4

Stacked barplots show the proportion of high and low GIS-scoring patients in the four CMS classes.

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.