Association between SMAD4 Mutations and GATA6 Expression in Paired Pancreatic Ductal Adenocarcinoma Tumor Specimens: Data from Two Independent Molecularly-Characterized Cohorts

Several molecular biomarkers have been identified to guide induction treatment selection for localized pancreatic ductal adenocarcinoma (PDAC). SMAD4 alterations and low GATA6 expression/modified “Moffitt” basal-like phenotype have each been associated with inferior survival uniquely for patients receiving 5-FU-based therapies. SMAD4 may directly regulate the expression of GATA6 in PDAC, pointing to a common predictive biomarker. To evaluate the relationship between SMAD4 mutations and GATA6 expression in human PDAC tumors, patients with paired SMAD4 mutation and GATA6 mRNA expression data in the TCGA and CPTAC were identified. In 321 patients (TCGA: n = 180; CPTAC: n = 141), the rate of SMAD4 alterations was 26.8%. The rate of SMAD4 alteration did not vary per tertile of normalized GATA6 expression (TCGA: p = 0.928; CPTAC: p = 0.828). In the TCGA, SMAD4 alterations and the basal-like phenotype were each associated with worse survival (log rank p = 0.077 and p = 0.080, respectively), but their combined presence did not identify a subset with uniquely inferior survival (p = 0.943). In the CPTAC, the basal-like phenotype was associated with significantly worse survival (p < 0.001), but the prognostic value was not influenced by the combined presence of SMAD4 alterations (p = 0.960). SMAD4 alterations were not associated with poor clinico-pathological features such as poor tumor grade, advanced tumor stage, positive lymphovascular invasion (LVI), or positive perineural invasion (PNI), compared with SMAD4-wildtype. Given that SMAD4 mutations were not associated with GATA6 expression or Moffitt subtype in two independent molecularly characterized PDAC cohorts, distinct biomarker-defined clinical trials are necessary.


Introduction
Pancreatic ductal adenocarcinoma (PDAC) has one of the highest case-specific mortality rates of all cancers [1].Although resection remains the only curative therapy for PDAC, improvements in long-term survival are attributable to advances in systemic treatment [2][3][4][5].Currently, 5-fluorouracil-based (i.e., with irinotecan and oxaliplatin as FOLFIRINOX) or gemcitabine-based (with Nab-paclitaxel) chemotherapies are both used, with selection largely driven by patient-related factors such as age, comorbidity, and performance status.Amid the expanding options for systemic therapy and the mounting emphasis on administering such agents in the neoadjuvant setting, identification of biomarkers to guide first-line treatment selection remains a critical unmet need.
Previous work has identified genomic alterations in SMAD4 as predictive of unique resistance to FOLFIRINOX induction.SMAD4 alterations, primarily loss-of-function mutations [6], are present in approximately 20% of localized PDAC patients and may be linked to increased rates of metastatic progression and lower rates of surgical resection in patients receiving induction FOLFIRINOX but not receiving gemcitabine/nab-paclitaxel [7,8].Separately, the modified Moffitt "basal-like" phenotype, marked by the loss of expression of GATA6, may also confer resistance to FOLFIRINOX chemotherapy [9][10][11][12][13].In an exploratory analysis of the ESPAC-3 trial, low GATA6 expression was associated with inferior clinical outcomes uniquely for patients receiving 5-FU/LV, but not gemcitabine [9].Similarly, in the COMPASS trial of patients with locally advanced or metastatic PDAC receiving FOLFIRI-NOX or gemcitabine/nab-paclitaxel, those with a modified "basal-like" phenotype had uniquely fewer responses and worse overall survival when treated with FOLFIRINOX [13].Moreover, GATA6-low cell lines derived from patient-derived xenografts were particularly resistant to 5-FU but not gemcitabine [9].
Understanding the relationship between SMAD4 and GATA6 will be important for informing future studies of molecular biomarkers to guide induction chemotherapy selection and rational clinical trial design.SMAD4 is on chromosome 18q11, and 28.7 Mb from GATA6.GATA6 and SMAD4 are frequently co-altered [9].Moreover, SMAD4 may directly regulate the expression of GATA6 [10].Together, these data raise the possibility that SMAD4 alterations at the genome level and loss of GATA6 expression may represent a single molecular pathway that confers unique resistance to 5-FU-based therapies.In this study, representing the largest cohort of PDAC patients with paired DNA and RNA expression data, we examined the association between SMAD4 alterations and GATA6 expression to further characterize the translational potential of this molecular relationship.We hypothesized that SMAD4 alterations would be positively correlated with GATA6 expression and the basal-like subtype.

Methods
The study was deemed IRB exempt with use of publicly available and deidentified datasets.The cBioPortal platform for Cancer Genomics database is an open-access resource for exploration of multidimensional cancer genomics data.The platform was queried for PDAC samples with paired mutation and mRNA expression data, with identification of two datasets: the Cancer Genome Atlas (TCGA) and the Clinical Proteomic Tumor Analysis Consortium (CPTAC) [14].
SMAD4 was considered altered when there was either a mutation or copy number deletion.The modified Moffitt phenotypes were assigned as previously described [15], where the R package ConsensusClusterPlus27 [16] was employed to subtype PDAC samples according to the expression signatures defined in Moffitt et al. [12].Briefly, the number of clusters was confirmed by examining cumulative distribution function, with the existence of well-separated clusters for Moffitt et al. classification based on tumor (basal-like and classical) and stroma signatures.GATA6 mRNA expression z-scores, as a surrogate biomarker of the modified Moffitt "basal-like" phenotype [13,17], were downloaded along with SMAD4 alteration calls and paired clinical data from the FireBrowse data portal (http://firebrowse.org)accessed on 12 August 2022 (TCGA data version 2016_01_28) and from cbioportal (https://www.cbioportal.org)(CPTAC data) [14].

Statistical Analysis
Descriptive statistics are presented as frequencies for categorical variables and median interquartile range (IQR) for continuous variables.Pearson's χ 2 and Wilcoxon rank-sum test were used to analyze categorical and continuous variables, respectively.GATA6 mRNA expression z-scores were analyzed by tertiles in each study to evaluate low, medium, and high expression as previously described [18].The primary outcomes assessed were (1) rates of SMAD4 alterations for classical vs. basal-like subtypes and (2) rates of SMAD4 alterations for GATA6-low vs. GATA6-medium/high.The secondary outcomes were the impact of SMAD4 and molecular subtype on overall survival (OS), which was evaluated by Kaplan-Meier estimates.The variables associated with OS on univariable analysis (p < 0.10) were entered into a multivariable Cox regression model (TCGA: AJCC stage, nodal status, and tumor grade; CPTAC: AJCC stage).p-values ≤ 0.05 were considered statistically significant; all tests were two-sided.Analyses were carried out using SPSS v27.0 (IBM Corp., Armonk, NY, USA).

Discussion
Improving patient outcomes in pancreatic cancer hinges on precise chemotherapy selection to match tumor responsiveness.FOLFIRINOX has emerged as one of the most effective chemotherapeutic regimens for managing PDAC, demonstrating efficacy in both metastatic and adjuvant therapy settings [4,19].As its application in the neoadjuvant setting continues to evolve, the integration of predictive biomarkers will be critical in refining treatment selection.SMAD4 mutations and the basal-like expression subtype (or GATAT6 low expression) have each emerged as potential biomarkers in this context.Herein, we present the largest cohort of PDAC patients evaluating the relationship between SMAD4 mutation, the modified Moffitt phenotype, and GATA6 expression in the context of clinical outcomes.Contrary to hypothesis, there was no correlation between SMAD4 alterations and the modified Moffitt phenotype or GATA6 expression.While in vitro data suggested that SMAD4 may regulate the expression of GATA6, these results are not confirmed in two human PDAC cohorts.Moreover, the combined presence of both biomarkers did not identify a patient subset with uniquely inferior outcomes.These data highlight the need for distinct biomarker-driven clinical trials and independent investigation of each biomarker in its potential mechanisms of treatment resistance.
SMAD4 serves as a mediator in the TGFB1 (TGF-β) signaling pathway and is recognized as a driver of the progression of pancreatic intraepithelial neoplasia to invasive adenocarcinoma [20,21].Together with KRAS, TP53, and CDKN2A, SMAD4 is recognized as one of the four driver mutations in PDAC [17,20,22].In these data, the rate of SMAD4 alterations ranged from 18.4% to 33.3%.The rate of SMAD4 in this study similarly compares to the previous literature s rates of 20-33% [23,24].This range likely represents inherent differences in study populations, where rates of SMAD4 alterations increase with greater tumor burden.Iacobuzio-Donahue et al. previously observed that locally

Discussion
Improving patient outcomes in pancreatic cancer hinges on precise chemotherapy selection to match tumor responsiveness.FOLFIRINOX has emerged as one of the most effective chemotherapeutic regimens for managing PDAC, demonstrating efficacy in both metastatic and adjuvant therapy settings [4,19].As its application in the neoadjuvant setting continues to evolve, the integration of predictive biomarkers will be critical in refining treatment selection.SMAD4 mutations and the basal-like expression subtype (or GATAT6 low expression) have each emerged as potential biomarkers in this context.Herein, we present the largest cohort of PDAC patients evaluating the relationship between SMAD4 mutation, the modified Moffitt phenotype, and GATA6 expression in the context of clinical outcomes.Contrary to hypothesis, there was no correlation between SMAD4 alterations and the modified Moffitt phenotype or GATA6 expression.While in vitro data suggested that SMAD4 may regulate the expression of GATA6, these results are not confirmed in two human PDAC cohorts.Moreover, the combined presence of both biomarkers did not identify a patient subset with uniquely inferior outcomes.These data highlight the need for distinct biomarker-driven clinical trials and independent investigation of each biomarker in its potential mechanisms of treatment resistance.
SMAD4 serves as a mediator in the TGFB1 (TGF-β) signaling pathway and is recognized as a driver of the progression of pancreatic intraepithelial neoplasia to invasive adenocarcinoma [20,21].Together with KRAS, TP53, and CDKN2A, SMAD4 is recognized as one of the four driver mutations in PDAC [17,20,22].In these data, the rate of SMAD4 alterations ranged from 18.4% to 33.3%.The rate of SMAD4 in this study similarly compares to the previous literature's rates of 20-33% [23,24].This range likely represents inherent differences in study populations, where rates of SMAD4 alterations increase with greater tumor burden.Iacobuzio-Donahue et al. previously observed that locally advanced PDAC without metastatic disease uncommonly showed loss of SMAD4 (22%) as compared with carcinomas with extensive metastatic burden, where the rates of SMAD4 alteration approached 75% [25].
SMAD4 alteration has been reported to be an independent prognostic factor for recurrence-free survival and overall survival [26][27][28][29].In a meta-analysis of eight studies with available data on SMAD4 status and patient survival, SMAD4 alterations conferred a pooled 40% increased risk of death [30].Several studies have observed that the loss of SMAD4 is linked to distant metastases, which may explain its prognostic significance [25,31].In our data, there was a trend to inferior survival in patients with SMAD4 alterations in the TCGA.However, a similar relationship was not observed in the CPTAC cohort.Notably, some studies have not identified SMAD4 as a prognostic biomarker [32,33].Winter et al. reported that loss of SMAD4 expression showed no association with either recurrence or early mortality in resected PDAC patients [23].These conflicting data highlight that disease-related survival is a complex interplay not solely driven by the function (or loss of) SMAD4, but likely driven by additional genetic, epigenetic, and environmental factors.
The lack of in vivo correlation between SMAD4 and the basal-like expression subtype (or GATAT6 low expression) warrants further discussion.Preclinical data suggested that SMAD4 can regulate the expression of GATA6.Using hTERT immortalized pancreatic ductal epithelial cells, suppressed SMAD4 (via small interfering RNA) reduced GATA6 expression.Conversely, FLAG-SMAD4 overexpression in PSN1 cells (which are SMAD4 deleted) resulted in re-establishment of GATA6 [10].However, these findings did not translate to the human PDAC specimens included in our study.The absence of a direct molecular correlation between these two biomarkers diminishes the likelihood of a single, druggable target to address treatment resistance.Moreover, the clinical significance of the basallike subtype, regardless of SMAD4 mutational status, further emphasizes the divergence of these molecular pathways.In these data, the basal-like subtype was associated with trends to inferior survival in the TCGA and significantly inferior survival in the CPTAC, and such associations were not impacted by the loss of SMAD4.While the association between basal-like subtype or low GATA6 and inferior survival are well established in the literature [7,8,10,13], the impact (or lack thereof) of concurrent SMAD4 alterations represents an additional complexity to this search for precision oncology.
GATA6, a member of the transcription factor family, binds to the (A/T)GATA(A/G) consensus sequence, influencing gene expression [34].Crucial for cell differentiation, GATA6 is vital for maintaining the exocrine pancreas [35].Recent studies propose a tumorsuppressive role of GATA6 in PDAC mouse models, influencing both differentiation and cancer-related transcriptional programs [36,37].In human PDAC cells, GATA6 plays a pivotal role in inhibiting de-differentiation and epithelial-mesenchymal transition (EMT).The sequential regulation of EMT and mesenchymal-epithelial transition (MET) is crucial for effective tumor spreading [38,39].In human PDAC samples, the loss of GATA6 in PDAC primary samples correlates with altered differentiation and shorter overall patient survival [40,41].
Our study is not without limitations.Conducting a retrospective analysis of clinical outcomes using a database inherently carries the risk of selection bias and potential inaccuracies in data reporting.Additionally, there is the possibility of sampling error in tumor biopsies, which could lead to mischaracterizations of SMAD4 mutational status or GATA6 mRNA expression.This is a particular limitation of genomic classification of PDAC given the large stromal component of many tumors.Third, certain analyses may have been underpowered given the limited sample sizes.The basal-like phenotype was associated with statistically inferior survival in the CPTAC cohort but not in the TCGA cohort; we believe this represents a type II error in the TCGA cohort.We methodologically avoided combining the TCGA and CPTAC datasets given the differences in tumor processing, which may have also contributing to some conflicting results.Fourth, the survival analyses may have been influenced by heterogeneity in the use of adjuvant therapy, for which data were not available.Additionally, analyses of recurrence-free survival were limited by the lack of such data in these cohorts.
Nonetheless, our study represents the largest cohort of PDAC patients with direct evaluation of the relationship between SMAD4 mutation and the basal-like subtype/GATA6 expression.Given the lack of correlation, distinct biomarker-driven clinical trials and individualized studies exploring the mechanistic basis of each biomarker are necessary for the advancement of precision oncology for this disease.

Figure 1 .
Figure 1.Kaplan-Meier estimates for the impact of SMAD4 and molecular subtype on overall survival in the TCGA cohort.(A) Association of SMAD4 alterations with overall survival; (B) association of Moffitt subtype with overall survival; (C) association of SMAD4 alterations with overall survival in the TCGA subset defined by the classical subtype; (D) association of SMAD4 alterations with overall survival in the TCGA subset defined by the basal-like subtype.

Figure 1 .
Figure 1.Kaplan-Meier estimates for the impact of SMAD4 and molecular subtype on overall survival in the TCGA cohort.(A) Association of SMAD4 alterations with overall survival; (B) association of Moffitt subtype with overall survival; (C) association of SMAD4 alterations with overall survival in the TCGA subset defined by the classical subtype; (D) association of SMAD4 alterations with overall survival in the TCGA subset defined by the basal-like subtype.

Figure 2 .
Figure 2. Kaplan-Meier estimates for the impact of SMAD4 and molecular subtype on overall survival in the CPTAC cohort.(A) Association of SMAD4 alterations with overall survival; (B) association of Moffitt subtype with overall survival; (C) association of SMAD4 alterations with overall survival in the CPTAC subset defined by the classical subtype; (D) association of SMAD4 alterations with overall survival in the CPTAC subset defined by the basal-like subtype.

Figure 2 .
Figure 2. Kaplan-Meier estimates for the impact of SMAD4 and molecular subtype on overall survival in the CPTAC cohort.(A) Association of SMAD4 alterations with overall survival; (B) association of Moffitt subtype with overall survival; (C) association of SMAD4 alterations with overall survival in the CPTAC subset defined by the classical subtype; (D) association of SMAD4 alterations with overall survival in the CPTAC subset defined by the basal-like subtype.
Abbreviations: IQR interquartile range; AJCC American Joint Committee on Cancer.a Race data unavailable for 4 patients.b Ethnicity data unavailable for 43 patients.c AJCC stage data unavailable for 2 patients.
Abbreviations: IQR interquartile range; AJCC American Joint Committee on Cancer; LVI lymphovascular invasion; PNI perineural invasion.a AJCC stage data unavailable for 7 patients.b LVI data unavailable for 11 patients.c PNI data unavailable for 8 patients.

Table 3 .
Association between SMAD4 alterations and GATA6 expression and Moffitt subtype.
a Moffitt subtype unavailable for 28 patients in TCGA cohort.bMoffittsubtype unavailable for 16 patients in CPTAC cohort.