Glycoprofiling of proteins as prostate cancer biomarkers: a multinational population study

The glycoprofiling of two proteins, the free form of the prostate-specific antigen (fPSA) and zinc-α-2-glycoprotein (ZA2G), was assessed to determine their suitability as prostate cancer (PCa) biomarkers. The glycoprofiling of proteins was performed by analysing changes in the glycan composition on fPSA and ZA2G using lectins (proteins recognising glycans, i.e. complex carbohydrates). The specific glycoprofiling of the proteins was performed using magnetic beads (MBs) modified with horseradish peroxidase (HRP) and antibodies that selectively enriched fPSA or ZA2G from human serum samples. Subsequently, the antibody-captured glycoproteins were incubated on lectin-coated ELISA plates. In addition, a novel glycoprotein standard (GPS) was used to calibrate the assay. The glycoprofiling of fPSA and ZA2G was performed in human serum samples obtained from men undergoing prostate biopsy after an elevated serum PSA, and prostate cancer patients with or without prior therapy. The results are presented in the form of a ROC (Receiver Operating Curve). A DCA (Decision Curve Analysis) to evaluate the clinical performance and net benefit of fPSA glycan-based biomarkers was also performed. While the glycoprofiling of ZA2G showed little promise as a potential PCa biomarker, the glycoprofiling of fPSA would appear to have significant clinical potential. Hence, the GIA (Glycobiopsy ImmunoAssay) test integrates the glycoprofiling of fPSA (i.e. two glycan forms of fPSA). The GIA test could be used for early diagnoses of PCa (AUC=0.84; n=501 samples) with a potential for use in therapy-monitoring (AUC=0.85; n=168 samples). Moreover, the analysis of a subset of serum samples (n=215) revealed that the GIA test (AUC=0.81) outperformed the PHI (Prostate Health Index) test (AUC=0.69) in discriminating between men with prostate cancer and those with benign serum PSA elevation.


56
In 2020, the worldwide incidence of and mortality from prostate cancer (PCa) were estimated as 1.41 57 million and 375,000, respectively; these figures are predicted to increase to 2.24 million and 721,000 by 58 2040 [1]. PCa has a large impact on a patient's quality of life; it significantly influences sexual, bowel and 59 urinary functions [2]. Early detection of cancer is crucial for a chance of curative treatment; however, PCa 60 screening also identifies PCa cases that are not fatal, thereby causing significant social distress, or 61 leading to unnecessary subsequent overtreatment [2]. Detection of indolent, low-risk PCa (i.e. Gleason 62 score 3 + 3 or ISUP grade group 1) may lead to anxiety and depression, especially for patients

76
The glycosylation process takes place in the Golgi apparatus, an organelle continuously receiving and 77 processing a flow of protein cargoes. Its well-organised cisternal structure has been shown to be crucial 78 for its proper functioning. Oncogenesis disrupts the structural integrity of the Golgi apparatus, resulting in 79 the abnormal expression of enzymes, the dysregulation of anti-apoptotic kinases and the hyperactivity of

86
In previous studies, we demonstrated the diagnostic potential of glycosylation changes in free 87 prostate-specific antigen (fPSA). Aberrant sialylation and fucosylation, for example, can be used to 88 diagnose both early-stage PCa and high-grade prostatic intraepithelial neoplasia; they can even be used 89 in the recognition of a castration-resistant form of PCa [18,19]. In our Glycobiopsy Immuno Assay (GIA) 90 test, we use a unique magnetic-beads-based protocol that overcomes the challenges inherent in lectin-91 assisted glycoprofiling of proteins [20][21][22]. The magnetic beads are modified by anti-fPSA antibodies for 92 the selective enrichment of fPSA from human blood serum samples. Subsequently, the magnetic beads 93 with attached fPSA are added to lectin-coated ELISA plates in order to perform glycoprofiling. Finally, the 94 sandwich ELISA protocol is completed by a horseradish peroxidase (HRP) reaction, (see detailed 95 protocol on: www.glycanostics.com). This protocol has proved to be robust and reproducible.

96
In the aforementioned studies [18,19], one serum sample of one particular PCa patient was applied to 97 calibrate the analysis and correct for plate-to-plate variability. Such an approach was feasible for clinical 98 validation using only a limited number of samples and/or for the analysis of samples in a single run/day.

99
The analysis of a large set of samples, or the analysis over a longer period of time, requires a proper 100 calibration. Attempts to resolve this issue by producing fPSA with attached cancer-specific glycans in

128
of the participants whose samples were used in the study are summarised in

139
Two independent clinical validation studies were performed; namely, (i) early diagnostics (early DX; 140 benign vs. PCa, no prior therapy) using 501 samples (unless indicated otherwise, see Table 1) and (ii)

185
In a subgroup analysis, the application of the GIA test to the identification of low-risk and high-risk PCa  fPSA, fPSA% and GIA test at 80% sensitivity revealed the following percentages: tPSA 30%, fPSA 53%, 197 fPSA% 53% and GIA test 70% (Fig. 2). Hence, the GIA test has the potential to significantly reduce the The GIA test can be confirmed by an AUC value of 0.85, which was significantly higher than the AUC 232 value for tPSA (0.61) (Fig. 3)

243
AUC value was 0.69 for the PHI test and 0.81 for the GIA test, respectively (Fig. 4)

246
A detailed analysis run at 80% sensitivity revealed that the following percentages of biopsies could 247 have been identified as negative (avoidable) for the following tests: tPSA 21% of biopsies, fPSA 52% of 248 biopsies, PHI 54% of biopsies and GIA 73% of biopsies (Fig. 5)

274
The main reason for using a DCA in this case is to show the potential benefit of using the GIA as 275 a second-opinion test to determine the need for a prostate biopsy at a given threshold probability. When 276 the probability of having a high-risk PCa is low, a urologist may decide to actively monitor the patient 277 rather than to perform (an avoidable) biopsy. This can eliminate future barriers between patients and 278 clinicians, as patients (not having undergone an unnecessary biopsy) will be less hesitant to return to the 279 care-provider. On the other hand, when there is a higher probability of PCa being present, the fear is that 280 a high-risk tumour might go undiagnosed and would subsequently be harder to cure. The DCA curve 281 analysis indicates a net benefit of using the GIA test compared to tPSA test, since it is more efficient 282 across all threshold probabilities, starting at ~5% (Fig. 6).

287
the principal components were calculated (Fig. 7). The first three principal components, capturing more and PC3, suggesting a strong added value of using these two lectins in the analysis. 95% confident 292 ellipses are shown for benign (green) and PCa (red) cohorts (Fig. 7).  In the present study, we confirmed that two glycoforms of fPSA recognised by WFL, (fPSA WFL ) and

323
The DCA curve analysis, used as another statistical method independent of the ROC curve analysis,

324
indicates a net benefit of using the GIA test compared to tPSA, as it appears to be more efficient across 325 all threshold probabilities, starting at ~5%. PCA also suggests a strong added value of using lectins for 326 PCa diagnostics. It is worth mentioning that both models showed a potential gain in comparison with the 327 "biopsy-all" and "biopsy-none" models across the strategy thresholds.

340
The GIA test has the potential to be used in therapy-monitoring since it distinguished between treated 341 and untreated PCa patients significantly better (AUC value of 0.85) than tPSA (AUC value of 0.61 (Fig.   342 3)).

368
Furthermore, out of 392 negative biopsies considered to be avoidable, 70-73% could have been 369 prevented had the GIA test been used; 21-30% had the tPSA been used; 52-53% with use of the fPSA 370 and 54% with use of the PHI test. Accordingly, the GIA test is able to outperform all the PSA-based 371 serological tests and has the ability to significantly reduce the number of biopsies.