Plasma ghrelin O‐acyltransferase (GOAT) enzyme levels: A novel non‐invasive diagnosis tool for patients with significant prostate cancer

Abstract Early detection of PCa faces severe limitations as PSA displays poor‐specificity/sensitivity. As we recently demonstrated that plasma ghrelin O‐acyltransferase (GOAT)‐enzyme is significantly elevated in PCa‐patients compared with healthy‐controls, using a limited patients‐cohort, we aimed to further explore the potential of GOAT to improve PCa diagnosis using an ample patients‐cohort (n = 312) and defining subgroups (i.e. significant PCa/metastatic patients, etc.) that could benefit from this biomarker. Plasma GOAT‐levels were evaluated by ELISA in patients with (n = 183) and without (n = 129) PCa. Gleason Score ≥ 7 was considered clinically significant PCa. GOAT‐levels were higher in PCa patients vs control patients, and in those with significant PCa vs non‐significant PCa. GOAT‐levels association with the diagnoses of significant PCa was independent from traditional clinical variables (i.e. PSA/age/DRE). Remarkably, GOAT outperformed PSA in patients with PSA‐levels ranging 3‐20 ng/mL for the significant PCa diagnosis [GOAT‐AUC = 0.612 (0.531‐0.693) vs PSA‐AUC = 0.494 (0.407‐0.580)]. A panel of key variables including GOAT/age/DRE/testosterone also outperformed the same panel but with PSA [AUC = 0.720 (0.710‐0.730) vs AUC = 0.705 (0.695‐0.716), respectively]. Notably, GOAT‐levels could also represent a novel predictive biomarker of aggressiveness, as its levels are positively associated with Gleason Score and the presence of metastasis at the time of diagnoses. Altogether, our data reveal that GOAT‐levels can be used as a non‐invasive biomarker for significant PCa diagnosis in patients at risk of PCa (with PSA: 3‐20 ng/mL).

detection" (CCB.030PM). Biorepository node (Córdoba, Spain) is also gratefully acknowledged for the samples. Ciber is an initiative of Instituto de Salud Carlos III, Ministerio de Sanidad, Servicios Sociales e Igualdad, Spain). AUC = 0.705 (0.695-0.716), respectively]. Notably, GOAT-levels could also represent a novel predictive biomarker of aggressiveness, as its levels are positively associated with Gleason Score and the presence of metastasis at the time of diagnoses. Altogether, our data reveal that GOAT-levels can be used as a non-invasive biomarker for significant PCa diagnosis in patients at risk of PCa (with PSA: 3-20 ng/mL).

K E Y W O R D S
GOAT enzyme, non-invasive biomarker, significant prostate cancer

| INTRODUCTION
Prostate cancer (PCa) has emerged as the most frequent cancer type among men, with an estimation of 164 690 new cases in the United States for 2018 (10% of all new cancer cases). 1 The rate of diagnosis has increased since the 1990s with the introduction of the PSA test for early detection of PCa, and metastatic disease and specific mortality have been reduced in most western countries. 2 However, a key limitation in PCa management is that early PCa diagnosis is mainly based on the plasma levels of PSA, a biomarker that exhibits profound drawbacks. For instance, PSA test displays low specificity because of the fact that multiple factors can increase PSA levels, such as benign prostatic hyperplasia or inflammation conditions, and this test is not able to accurately distinguish clinically relevant tumours from indolent cases. 3 This leads to the overdiagnosis of PCa with many unnecessary biopsies and reduced patient quality of life (QoL), as well as to the overtreatment in a considerable number of patients. 4 Likewise, clinical management of aggressive PCa, that is metastatic and castration-resistant PCa (CRPC), also faces major limitations, including unresponsive patients and development of resistance to hormonal and chemical therapies. 5,6 Therefore, there is an important unmet clinical need for the identification and validation of new, reliable and specific biomarkers for early diagnosis of PCa, as well as for prediction of disease prognosis and treatment response, etc., which would improve patient survival and QoL and would reduce substantially the number of unnecessary biopsies in patients with suspect of PCa based on PSA test.
In line with this, and using a limited cohort of patients, we have recently demonstrated that ghrelin-O-acyltransferase (GOAT), a key enzyme regulating ghrelin system activity, [7][8][9] is overexpressed in PCa tissues (at the mRNA and protein level) and its plasma levels are elevated in PCa patients compared to healthy prostate tissues and to plasma from healthy controls, respectively. 10 Moreover, we observed that plasma GOAT levels could discriminate PCa, suggesting that GOAT might serve as a potential novel non-invasive biomarker of PCa. 10 However, in this previous pilot study, we could not establish whether plasma GOAT levels could be a significantly better diagnostic marker than PSA for the diagnosis of PCa, specially on those individuals with PSA levels ranging 3-20 ng/mL (wherein precision of PSA is remarkably poor), and for the diagnosis of significant PCa (Sig PCa). Accordingly, the aims of this study were (a) to valorize the utility of plasma GOAT enzyme levels alone, or in combination with other traditional clinical variables, as a tool for the detection of PCa, using a more representative, ample cohort of patients (n = 312) and by defining specific subgroups (e.g. Sig PCa vs non-Sig PCa) that could benefit from this biomarker; (b) to compare the utility of plasma GOAT vs PSA levels as diagnostic tools in this cohort of patients; and, (c) to determine the utility of plasma GOAT enzyme levels as a novel predictive biomarker of aggressiveness, by analysing its association with Gleason Score (GS), metastatic PCa and earlier CRPC status in the same cohort of patients.

| MATERIAL AN D METHODS
This is a case-control study implemented with patients who donated blood under fasting conditions in the Reina Sofia University Hospital. The study was approved by the Hospital Ethic-Committee, and written informed consent from all patients was obtained. All samples were obtained through the Andalusian-Biobank (Servicio Andaluz de Salud, Spain).

| Patients and samples
Three cohorts of patients were included in the study:   In order to determine and compare the levels of GOAT and PSA in plasma samples from all the patients included in this study (cohorts 1-3 mentioned above; a total of 312 samples), blood was collected early in the morning, after an overnight fast. Each blood sample was placed into a vacutainer tube containing sodium citrate, centrifuged 10 minutes at 1100 g (20°C) and subsequently plasma was aliquoted in tubes and kept at −80°C. Additionally, clinical, anthropometric and pathological features of all the patients were obtained and registered. In addition, testosterone levels were evaluated in patients at risk of PCa (cohorts 2 and 3).

| Determination of plasma GOAT, PSA and testosterone levels
For the determination of plasma GOAT levels, a commercial ELISA was used following the manufacture's instructions (MyBioSource, San Diego, USA), as previously reported. 10 GOAT ELISA kit exhibits a detection limit lower than 0.31 ng/mL and a detection range between 0.78 and 50 ng/mL. The intra-and interassay accuracy showed a CV lower than 10% and 12%, respectively. Samples were diluted 1:100 before performing the assay. Levels of PSA and testosterone were measured using technology of Chemiluminescent Microparticle Immunoassays (References 7k70 and 7k73, respectively; Abbott) following the manufacturer's instructions.

| Variables and statistical analysis
A descriptive study was performed by calculating the median and interquartile ranges for the quantitative variables and the absolute frequencies and percentages for the qualitative variables. One of the primary endpoints of the study was the presence of a clinically Sig PCa on biopsy. The tumours with a GS ≥ 7 were considered clinically Sig PCa. Student's t test was used for analysis of the quantitative data in case of two groups and ANOVA with Bonferroni's post hoc test in case of comparison between the three groups. A chisquare test was used for the qualitative variables. To study the correlation between GOAT levels and other clinical variables, a Pearson test was used. To address the diagnostic value of both PSA and GOAT measures, their associated ROC curves were built, showing the performance (specificity and sensitivity) for the different risk thresholds. The performance was then compared using DeLong tests over the respective areas under the curves (AUC). Then, the performance of multivariate models based on these measures, when complemented with additional clinical variables (age, DRE, BMI, testosterone, number of biopsies and family history) was investigated. These models were built using logistic regression, preceding the model construction with a feature selection step, using like-hood ratio test to discard variables that do not contribute to diagnostic performance. The performance of these models was then evaluated using 10-fold cross-validation, including the variable selection step to avoid selection bias. Similar to the case of univariate models, ROC curves and DeLong tests were used to compare the different models. An exploratory analysis for the association and prognosis value of GOAT was carried out. For this purpose, data from the follow-up and treatment with hormonotherapy according to clinical practise were also collected. A univariate Cox Regression analysis was carried out to explore the association of GOAT levels with the development of castration resistant disease (CRPC). A 5% level of significance (after adjusting for multiple comparisons, if specified) was used to decide statistically significant differences to make our conclusions T A B L E 1 Demographic/clinical data and anatomopathological characteristics of the three cohorts of patients included in this study

| Capacity of plasma GOAT and PSA levels to predict the presence of PCa and Sig PCa
Plasma levels of GOAT were statistically higher in patients with PCa compared to patients with negative biopsy and healthy patients (Figure 1A, left panel). In contrast, PSA levels were higher in patients with PCa compared to healthy patients but not with patients at risk of PCa but with negative biopsy ( Figure 1A, right panel). When patients with PCa were divided in two subgroups, with and without Sig PCa, we found that, although both plasma GOAT and PSA levels were significantly elevated in patients with Sig PCa (GS ≥ 7) compared to patients with non-Sig PCa (GS = 6), these differences were statistically more significant for GOAT vs PSA levels (P = 0.002 vs P = 0.0145; Figure 1B).
Additionally, plasma GOAT and PSA levels were also found to be higher in patients with Sig PCa compared to the combined group of patients at risk of PCa but with a negative biopsy together with patients with non-Sig PCa ( Figure 1C), being these differences again statistically more significant for GOAT vs PSA levels. Importantly, when the patients with a PSA range between 3 and 20 ng/mL (the most ambiguous region of PSA levels, which leads to a high falsepositive rate and, therefore, to a high number of unnecessary prostate biopsies) were analysed in more detail, we found that plasma GOAT, but not PSA, levels were significantly higher in patients with Sig PCa compared to the combined group of patients with negative biopsy and with non-Sig PCa ( Figure 1D).
Interestingly, plasma GOAT levels positively correlated with plasma PSA levels ( Figure 1E), but not with testosterone levels (r = −0.044; P = 0.49; data not shown), in this cohort of patients, which is consistent with our previous study using a different cohort of patients. 10 Moreover, a positive correlation was found between plasma GOAT or PSA levels with age ( Figure 1F). To explore the potential capacity of prediction of plasma GOAT levels compared to PSA levels, patients from cohorts 2 and 3 with a PSA range between 3 and 20 ng/mL were analysed. This analysis revealed that GOAT was a better biomarker than PSA for the diag-  Figure 2B].  This analysis was also applied to assess the predictive capacity of plasma GOAT levels, compared to PSA levels, in patients with a more restricted range of PSA, of 3-10 ng/mL, the so-called PSA grey zone ( Figure 2C,D). The results clearly indicated that GOAT levels are a significantly better indicator than those of PSA to predict PCa in these patients [n = 117 PCa patients; GOAT levels: AUC = 0.586 (0.497-0.674) vs PSA levels: AUC = 0.417 (0.330-0.504), P < 0.01), Figure 2C]. Likewise, as illustrated in Figure 2D, was implemented to determine whether this combination could improve the accuracy of detection of PCa in patients with PSA levels between 3-20 ng/mL ( Figure 2E) and 3-10 ng/mL ( Figure 2F). This analysis revealed that the combination of this panel of clinical variables with plasma GOAT levels is significantly more efficient in detecting Sig PCa than when combined to plasma PSA levels [P < 0.001 in both cases; Figures 2E,F].

| Association of plasma GOAT levels with aggressiveness features of PCa patients
Association between aggressiveness features of the cohort of patients with PCa revealed that plasma GOAT levels showed a significant correlation with GS (r = 0.24; P = 0.001; Figure 3A). Remarkably, high GOAT levels were associated with the presence of metastasis at the time of diagnosis, as evaluated by computerized tomography and bone scan (P = 0.03; Figure 3B)  Figure 3C).

| DISCUSSION
PCa is a major health problem and a leading cause of mortality and morbidity globally. 1 PSA has been used as the gold standard biomarker for the diagnosis of PCa since the 1990s, although its use remains controversial because of its lack of specificity. Specifically, although the proportion of men with metastatic PCa at the time of diagnosis have decreased dramatically with the introduction of PSA as a screening test, more men are being diagnosed with PCa, with the majority having early stage, clinically indolent disease, the majority of which may never have led to harm. 12 In addition, many men with benign conditions such as inflammation or hyperplasia are also being diagnosed and biopsied based on the results of the PSA test. 3 Moreover, it has been proposed that treatment of indolent cancer may cause a patient more harm than good as biopsies and PCa treatments have been associated with psychological distress, loss of bodily function, pain, suffering for patients and with a decrease in the patient QoL. 13 Consequently, these data have led to widespread criticism that PCa is now an "overdiagnosed" and "overtreated" cancer based on the PSA test. Therefore, there is an urgent need for the identification of new diagnostic and prognostic biomarkers for PCa, especially for Sig PCa, in order to improve the clinical management of PCa and to reduce the elevated number of biopsies and the over- In line with this, our group and others have recently demonstrated that GOAT enzyme is overexpressed (at the mRNA and/or protein level) in PCa tissues and PCa cell lines compared to healthy prostate tissues and normal cell lines, 10,21 and, most importantly, we also reported that GOAT is oversecreted in PCa cells compared to normal prostate cells. 10 In fact, this initial, pilot study from our group revealed that plasma GOAT levels could discriminate between PCa and healthy subjects, suggesting that this enzyme might be used as a potential novel non-invasive biomarker of PCa. 10   In sum, the present report provides the first comparative analysis to determine the potential utility of plasma levels of GOAT, in combination with other traditional clinical variables (i.e. age, DRE and/or testosterone), as diagnostic tools for the detection of PCa, using an ample cohort of patients (n = 312) and defining clinically relevant subgroups (e.g. Sig PCa vs non-Sig PCa). Our results show, for the first time, that the measurement of plasma GOAT levels might represent a significantly better diagnostic marker than plasma PSA levels, exhibiting higher AUC, particularly on those individuals with PSA levels ranging 3-10 ng/mL (the PSA grey-zone) or 3-20 ng/mL. Moreover, as plasma GOAT levels showed a significant better AUC than plasma PSA levels for the detection of Sig PCa and its levels were associated with aggressiveness features of PCa, we propose that the measurement of plasma GOAT levels, in combination with PSA and/or an additional panel of clinical variables measured in PCa (i.e. age, DRE and testosterone levels), might be considered as a novel, complementary, non-invasive tool to provide a better diagnosis of PCa, especially for Sig PCa and for patients with grey-zone PSA levels, as well as a putative tool for the prediction of PCa aggressiveness.