Screening for oncogenic AF1q expression predicts disease recurrence in gastric cancer patients

AF1q associates with tumor progression and metastases upon WNT signaling. The downstream WNT target CD44 has demonstrated prognostic significance in gastric cancer (GC). This study evaluates the impact of AF1q on tumor stage and survival in GC patients. Immunohistochemical marker expression was analyzed and data were processed to correlation and survival analysis. Out of 182 GC samples, 178 (97.8%) showed moderate to high AF1q expression (p < 0.001), these samples correlated with positive lymph node stage (p = 0.036). In a subgroup analysis of patients with nodal-positive GC (n = 129, 70.9%), enhanced tumoral AF1q expression resulted in impaired recurrence-free survival (RFS, p = 0.030). Enhanced tumoral CD44 expression resulted in impaired disease-specific survival (DSS) in the subgroup of patients with nodal-positive GC (p = 0.031) as well as in the overall GC group (p = 0.005). AF1q demonstrated as an independent prognostic marker for RFS (p = 0.035) and CD44 for DSS (p = 0.036). AF1q has shown potential for prognostication of RFS in GC patients and is predominantly expressed in nodal-positive GC. Testing AF1q provides a possibility of identifying patients with locoregional (and advanced) disease, particularly at risk for disease recurrence. Implementing AF1q into the diagnostic process may facilitate screening, prognosis estimation as well as consideration of preoperative multimodal treatment in patients qualifying for elective upfront surgery.

In patients with AF1q-positive GC, median RFS was 26.8 months (19-34 months) in patients with moderate AF1q expression and 18.2 months (10-26 months) in patients with high AF1q expression compared to 22.5 (10-35 months) in patients with AF1q-negative GC; median DSS was 46.1 months (34-58 months) in patients with moderate AF1q expression and 37.0 months (22-52 months) in patients with high AF1q expression compared to 37.3 months (16-58 months) in patients with AF1q-negative GC.In the subgroup of nodal-positive GC, median RFS was 26.1 months (17-35 months) in patients with moderate AF1q expression and 13.1 months (7-19 months) in patients with high AF1q expression compared to 22.5 months (10-35 months) in patients with AF1q-negative GC; median DSS was 39.4 months (27-52 months) in patients with moderate AF1q expression and 26.8 months (23-42 months) in patients with high AF1q expression compared to 37.3 months (16-58 months) in patients with AF1q negative-GC.Enhanced tumoral AF1q expression resulted in significantly impaired RFS (Kaplan Meier/log rank; p = 0.030; Fig. 3), but not DSS in the subgroup of patients with nodal-positive GC.No significant impact was found on RFS or DSS in the overall GC group.In patients with CD44 positive GC, median RFS was 21.2 months (13-30 months) in patients with moderate CD44 expression and 6.0 months (0-12 months) in patients with high CD44 expression compared to 25.3 months (18-32 months) in patients with CD44 negative GC; median DSS was 28.1 months (17-39 months) in patients with moderate AF1q expression and 7.0 months (0-15 months) in patients with high AF1q expression compared to 51.8 months (38-65 months) in patients with AF1q negative GC.In the subgroup of nodal-positive GC, median RFS was 20.4 months (10-31 months) in patients with moderate CD44 expression and 6.0 months  in patients with CD44 negative GC. Enhanced tumoral CD44 expression resulted in significantly impaired DSS in the subgroup of patients with nodal-positive GC (DSS: Kaplan Meier/log rank; p = 0.031; Fig. 4) as well as in the overall GC group (DSS: Kaplan Meier/log rank; p = 0.005; Fig. 5), but no impact was found on RFS.

Discussion
Gastric cancer (GC) is the fifth most common cancer and the fourth common cause of cancer death worldwide, responsible for a total of 10 million deaths in 2020 3 .Though Out of those patients treated with curative intent, a considerable number relapse within two years after surgery (Asia: 60% vs. Europe: 36.8%) 19,20.Exploring predictive markers to identify risk of recurrence is key to optimize follow-up and improve survival of GC patients.Whilst the role of the multifaceted oncogene AF1Q in GC is widely unknown, CD44 has proven prognostic potential for both diagnosis and treatment of GC 18 .We here focused on exploring the expression of AF1q in samples of patients operated on for GC as well as AF1q's relation to tumoral CD44 expression; secondly, we aimed to elicit AF1q's potential as a prognostic marker for GC survival.By analyzing 182 GC samples, we found AF1q to be significantly enhanced, especially in nodal-positive GC.In this subgroup of patients, those with AF1qpositive GC relapsed earlier, whilst those with CD44-positive tumors died earlier from their disease compared to marker-negative GC.Survival analysis revealed AF1q as an independent prognostic marker for RFS and solidifies the role of CD44 as an independent prognostic marker for DSS in GC patients.In this GC cohort, 153 patients (84.1%) qualified for upfront surgery with postoperative chemotherapy, 115 patients (63.2%) suffered from disease recurrence and from these, 111 patients (96.5%) had AF1q-positive GC.Studies in human breast cancer samples demonstrated that the cooperation of AF1q with TCF7 is involved in the transcription of CD44 9 , a WNT target gene that is highly expressed in GC 18 and known to drive tumor progression and epithelial-to-mesenchymal transition 21,22 , the basis for enhanced migratory capacity of cells and hence tumor spread.Additionally, other groups reported that AF1q associated with metastatic spread in colorectal, breast and lung cancer [23][24][25][26] .Although we had expected a possible association of AF1q with CD44 in GC since our recent study in esophageal cancer patients 12 , no association was found in the samples of this GC patient cohort, where AF1q was abundantly expressed compared to CD44 (AF1q: n = 178, 97.8% vs. CD44: n = 64, 35.2%), which is likely the reason for the lacking correlation between these markers.However, this association in terms of AF1q driving CD44 transcription would have been the assumed explanation that AF1q associates with nodal-positive and recurrent GC and since this rather low CD44 expression in GC is contradictive to the literature, the task of how AF1q is involved in tumor spread mechanistically now still remains open for further analysis.Nonetheless, patients with CD44-positive GC in this cohort died earlier from their disease, which underlines the potency of CD44 as a tumor modulator-given the fact that downregulation of CD44 inhibits proliferation, invasion and metastasis of GC 13,15,17 , rendering it a potential therapeutic target for GC.
Until the year 2040, Asia will account for the highest number of GC-related deaths (Asia 1.01 million vs. Europe 124,000 2 ; geographical variations expected due to Helicobacter pylori infection, smoking and consumption of salt and salt-preserved foods [27][28][29] ).Especially in GC, where screening is not recommended routinely, assessment of appropriate marker expression might facilitate the diagnostic process and optimize treatment decisions as well as follow-up.Patients with AF1q-positive GC showed significantly impaired RFS (Fig. 3) and as for this, including marker analysis into the diagnostic process might prompt multimodal treatment in terms of escalating to preoperative therapy in selected cases of de-facto resectable GC; this "pseudo-neoadjuvant" regimen would aim to combat locoregional disease up to dormant metastases and ultimately prevent disease recurrence and consequently death from disease.Patients with CD44-positive GC showed significantly impaired DSS in the subgroup of patients with nodal-positive GC (Fig. 4) as well as in the overall GC group (Fig. 5), but no impact was found on RFS.This finding and the correlation with metastatic GC raises the assumption that CD44 might operate through a more aggressive tumor behavior, which results in earlier death from disease.
In conclusion, this study provides evidence that AF1q has a potential as a negative and independent prognostic marker for RFS in patients with GC.The expression especially in patients with nodal-positive GC justifies considering including AF1q into the diagnostic process and through this to prompt preoperative multimodal treatment even in upfront resectable GC to limit locoregional as well as recurrent disease.This treatment regimen would be a step towards earlier diagnosis, better prognosis estimation and lower socio-economic burden of this still fatal disease.

Patient cohort
Patients operated on for gastric adenocarcinoma between 1992 and 2011 at the Medical University of Vienna were included in the study.Ethical approval for the study was obtained from the institutional review board ('Ethikkommission' of the Medical University of Vienna, protocol #1197/2019) and informed consent was waived off.All methods were carried out in accordance with relevant guidelines and regulations.
Histopathological staging was conducted according to the AJCC/UICC staging system 1 .Surgical specimens were carefully selected by an experienced, board-certified pathologist.Selected tumor samples were represented by triplicate core biopsies to construct a tissue microarray (TMA), which were then cut and processed to immunohistochemical staining.

Immunohistochemistry
The expression of AF1q as well as CD44 was evaluated in resected human GC samples.Immunostaining was performed using a standard protocol with the following antibodies: AF1q (Abcam, ab109016; 1:200), CD44 (Santa Cruz, sc-9960; 1:200).Paraffin sections were de-waxed, and for the antigen retrieval, a citrate buffer pH 6 (CD44) or a Tris/Ethylenediaminetetraacetic acid (EDTA) buffer pH 9 (AF1q) was used.After endogenous peroxidase blocking, avidin and biotin blocking steps were performed.The antibodies were incubated overnight at 4 °C in PBS + 1% bovine serum albumin (BSA).Slides were washed with phosphate buffered saline (PBS) the following day and incubated with polyvalent-secondary antibody (IDetect Super Stain System HRP, ID laboratories) and horseradish peroxidase (HRP; IDetect Super Stain System HRP, ID laboratories).Signals were visualized with www.nature.com/scientificreports/3-amino-9-ethylcarbazole (ID laboratories).After counterstaining with hemalaun, the slides were mounted.The samples were analyzed by an experienced, board-certified pathologist.A specimen was considered as positive when at least 50% of tumor cells showed moderate or strong cytoplasmic (AF1q) or membranous (CD44) marker expression.Antibody specificity has been confirmed in previous studies 9,12 .

Statistical analysis
To evaluate AF1q expression in relation to patient and tumor characteristics, the Chi 2 test and the Spearman rank correlation coefficient (r s ) were used as appropriate.The Kaplan-Meier method was used to estimate survival using the log-rank test for group comparisons.Survival times were defined as follows: RFS-time from surgery until disease recurrence, DSS-time from surgery until death from disease.To evaluate the prognostic potential of AF1q, a multivariable Cox Regression model was calculated including AJCC/UICC tumor staging and resection margin 1,30 .For statistical computing IBM ® SPSS ® Statistics Version 28.0.0.0 (IBM Cooperation, USA) was used.For all analyses, a two-sided p-value less than 0.05 was considered statistically significant.

Figure 1 .
Figure 1.Representative examples of immunohistochemical (IHC) staining of AF1q in gastric adenocarcinoma.Samples with (A) enhanced AF1q expression versus samples with (B) hematoxylin-eosin staining of patients that underwent surgery.Scale bar 100 μm.

Figure 2 .
Figure 2. Representative examples of immunohistochemical (IHC) staining of CD44 in gastric adenocarcinoma.Samples with (A) enhanced CD44 expression versus samples with (B) hematoxylin-eosin staining of patients that underwent surgery.Scale bar 100 μm.

Table 1 .
AF1q expression in relation to patient and tumor characteristics.

Table 2 .
Uni Vol.:(0123456789) Scientific Reports | (2024) 14:15988 | https://doi.org/10.1038/s41598-024-67058-xwww.nature.com/scientificreports/Cox regression analysis Patients with AF1q positive RC showed a trend towards a higher risk for disease recurrence (p = 0.071), but no impact on disease-specific death.Further prognostic factors for RFS and DSS are depicted in Table 2.In a multivariable Cox Regression model, AF1q proved to be an independent factor for RFS (p = 0.035) next to prognostic factors like lymph node stage, distant metastases, tumor grade and resection margin.With regard to DSS, AF1q showed a potential trend as an independent prognostic factor (p = 0.057) next to prognostic factors like lymph node stage, distant metastases, tumor grade (trend) and resection margin.Univariate and multivariable Cox Regression data are compiled in Table 2. Figure 5. Kaplan-Meier analysis for disease-specific survival in gastric cancer (GC) patients.Patients with moderate (1) and high (2) CD44 expression die earlier compared to patients with CD44-negative (0) GC (log rank: p = 0.005).-and multivariate cox regression analysis for recurrence-free and disease-specific survival in GC patients.pN and pM according to the AJCC/UICC staging system.G: tumor grade; R: resection margin.Significant values are in bold.Factor Univariate p-