Increasing Use of Germline Genetic Testing in Pancreatic Ductal Adenocarcinoma and Relationship to Clinical Outcome: A Single-Institution Study

Abstract

PURPOSE

Precision therapies have improved outcomes in pancreatic ductal adenocarcinoma (PDAC) and highlight the utility of germline genetic testing (GGT) in clinical decision making. Here, we investigate the association of GGT of individuals diagnosed with PDAC over time to identify changes in testing patterns and the relationship of testing with clinicopathologic parameters and overall survival (OS).

METHODS

The frequency of GGT of 979 patients with PDAC treated at the Roswell Park Comprehensive Cancer Center (RPCCC) between 2017 and 2024 was determined. Factors associated with performing testing and having a germline pathogenic variant (GPV) were assessed. OS and the impact of precision treatment were evaluated.

RESULTS

Patients with PDAC being treated at RPCCC that had GGT increased from 16.5% in 2017 to 57.6% in 2024. Patients with testing had improved OS compared with patients who did not (HR, 0.46). Individuals of older age, without a personal history of other cancers, without a family history of pancreatic cancer, with advanced disease, and with poorly differentiated tumors were less likely to have testing performed. Individuals with a personal history of other cancers were more likely to have a GPV. Of the 425 patients diagnosed with PDAC that had testing, 58 (13.6%) had a GPV detected. Of patients with testing, 12 had actionable mutations in BRCA1/2, seven of whom were subsequently treated with olaparib. Three patients harbored actionable mutations in MSH6, with one patient subsequently treated with pembrolizumab. Patients who received therapy informed by testing results had improved OS (HR, 0.09).

CONCLUSION

GGT prevalence has increased at RPCCC and informed treatment decisions. Universal point-of-care testing is being implemented with the goal of completing testing for all patients with PDAC seen at RPCCC.

INTRODUCTION

Pancreatic ductal adenocarcinoma (PDAC) carries a poor prognosis as the disease is often diagnosed at an advanced state and treatment options are limited.^1–3 Until recently, few genetic features of pancreatic cancer were considered druggable. This was due to the challenge of targeting the Kirsten rat sarcoma viral oncogene homolog and tumor suppressor pathways that are commonly dysregulated in PDAC. As a result, the majority of patients with PDAC have not had access to therapies directed at common somatic mutations. Furthermore, targeting pathways that would be inferred as actionable in PDAC (eg, mitogen-activated protein kinase kinase) are yet to yield promising results.^4–6

By contrast, it is known that select germline risk factors for PDAC and other cancers create vulnerabilities to specific drugs.⁷ Thus, the National Comprehensive Cancer Network (NCCN) Guidelines for PDAC were updated in 2018, recommending that germline genetic testing (GGT) be considered for all patients with PDAC regardless of family history.⁸

Identification of germline mutations was historically used to define at-risk family members who may benefit from surveillance as approximately 10% of patients with PDAC carry a germline pathogenic variant (GPV) that predisposes them to the disease.^9–12 Thus, germline testing was largely driven by clinical history. However, with the realization that certain genetic dysfunctions create a therapeutic vulnerability, testing guidelines were expanded. Today, these inform both therapy for a diagnosed patient and also cascade testing and prevention and early detection interventions for at-risk family members. NCCN has identified ATM, BRCA1/2, CDKN2A, STK11, MLH1, MSH2, MSH6, EPCAM, PALB2, and TP53,¹³ along with APC and PMS2 as identified by additional guidelines from the American College of Gastroenterology¹⁴ as genes to be tested that predispose an individual to the development of PDAC if harboring a pathogenic variant.^15–17 Thus, patients with PDAC harboring a GPV in any of these genes provide the impetus to test close relatives.¹⁸

Cancer patients with a GPV in the BRCA1/2 genes are particularly responsive to platinum agents and subsequently to poly(ADP-ribose) polymerase (PARP) inhibitors such as olaparib.^19–23 Similarly, patients with pancreatic cancer harboring PALB2 GPVs have shown sensitivity to PARP inhibitors,^22,24 with larger clinical trials currently underway (ClinicalTrials.gov identifier: NCT04858334). In patients with PDAC harboring germline or somatic PVs in DNA mismatch repair genes (including MLH1, MSH2, MSH6, and PMS2), individuals had significantly longer progression-free and overall survival and fewer adverse effects to therapy when treated with the immune checkpoint inhibitor pembrolizumab,^25,26 as has been observed across cancer types.²⁷

Here, we review the prevalence of GGT of patients with PDAC treated at the Roswell Park Comprehensive Cancer Center (RPCCC) from 2017 to 2024. We assess the OS of patients who did or did not have testing performed and factors associated with completing testing and having a GPV. We also provide an overview of frequently mutated genes in the germline of patients with PDAC and review patients in which genetic testing informed treatment strategy and the subsequent response to therapy. These findings highlight the importance of continued efforts to increase germline testing for patients with PDAC.

METHODS

Patient Selection

Through an RPCCC institutional review board–approved study, chart review was performed on 979 patients with PDAC 18 years or older who received treatment for PDAC at RPCCC or were seen and referred to hospice between 2017 and 2024. All patients included were evaluated by the medical oncology team; no patients who simply received a second opinion were included. Detailed information, including demographics, histopathologic features, treatments, and genetic testing results were recorded in a single REDCap database.

Genetic Testing

Genetic testing was performed for clinical purposes using germline panels from Ambry Genetics (Aliso Viejo, CA), Baylor Genetics (Houston, TX), Invitae (San Francisco, CA), and Myriad Genetics (Salt Lake City, UT). All records were read to identify if testing had been conducted at RPCCC or elsewhere.

Statistical Analysis

OS was measured in months from the date of primary diagnosis to the date of death or last known follow-up. The Wilcoxon test was used to compare time to GGT ordering and completion from 2017 to 2024 as the data were found to be not normally distributed via Shapiro-Wilk testing. Kaplan-Meier analysis using log-rank comparisons and Cox proportional hazards were used to perform multivariable analysis and compare the OS of patients receiving GGT and those not receiving testing. All statistical analyses, oncoprints, and swimmer plots were completed using the survival, ComplexHeatmap, and swimplot packages in R, version 4.0.5. Univariable analysis was conducted to determine the association of individual factors with the completion of testing and having a GPV. Continuous variables were dichotomized at the median for univariable and multivariable analyses. Significance was set at P < .05.

RESULTS

Patients treated for PDAC at RPCCC from 2017 to 2024 were tracked to assess the prevalence of GGT. During that time, 979 patients were treated for PDAC and GGT was completed for 425 (43.4%) of those patients (Table 1). The reasons noted for why testing was not completed consisted of the following: (1) the clinician never offered (43.1%), (2) unknown/not documented (24.0%), (3) poor performance status (15.7%), (4) the patient declined (12.8%), (5) the clinician never ordered (4.0%), and (6) the patient was lost to follow-up (0.4%; Appendix, Table A1). Of all patients in this study who did or did not have testing performed, 52.3% were male, 47.7% were female, 80.5% were White, 8.3% were African American, 0.6% were Asian, 0.2% were Hispanic or Latino, 0.1% were American Indian/Alaskan Native, and 10.3% were unknown/not reported (Table 1). The median age at diagnosis was 68 years (range, 31–95). The median OS was 8.8 months. Time from diagnosis to ordering GGT varied, with a median of 5 days and a median time from diagnosis to GGT completion of 59 days. Eastern Cooperativity Oncology Group (ECOG) scoring of the patient’s functional status at time of diagnosis varied in 40.6% of patients with grade 0 (fully functional), 38.2% with grade 1, 10.6% with grade 2, 5.2% with grade 3, and 0.6% with grade 4 (completely disabled). Disease status at time of diagnosis was evenly split, with 46.3% of patients having localized disease and 53.7% of patients presenting with metastatic disease; 16.5% of patients presented with stage I/II disease, and 53.8% presented with stage III/IV disease. The majority of patients presented with moderately differentiated tumors (55.6%), followed by poorly differentiated (24.9%), with only 0.8% of study participants presenting with well-differentiated tumors. A total of 20.2% of study participants had a personal history of any other cancers (including nonmelanoma skin cancer), and 9.7% had a family history of pancreatic cancer, defined as a parent, sibling, or other immediate family member having the disease. The median BMI was 26.4 kg/m², and most study participants were never (41.3%) or former (38.8%) smokers, with 18.1% being current smokers. Of the 425 patients in this study who had germline testing, a GPV was detected in 58 (13.6%) individuals. Mutation frequencies for all genes identified as harboring a pathogenic variant are listed in Appendix, Table A2.

TABLE 1.

Patient Demographics

Characteristic	Cohort, No. (%)
Germline testing done	Yes, before diagnosis: 11 (1.1)
	Yes, after diagnosis: 414 (42.3)
	No: 554 (56.6)
Sex at birth	Male: 512 (52.3)
	Female: 467 (47.7)
Race	White (not Hispanic): 788 (80.5)
	African American: 81 (8.3)
	Hispanic or Latino: 2 (0.2)
	Asian: 6 (0.6)
	American Indian/Alaskan native: 1 (0.1)
	Unknown/not reported: 101 (10.3)
Age at diagnosis (years)	Median: 68
	Range: (31–95)
	Mean: 68.1
Current status	Alive: 203 (20.7)
	Deceased: 776 (79.3)
OS (months)	Median: 8.8
	Range: (0.2–91.3)
Days from diagnosis to germline order	Median: 5
	Range: (0–1,910)
	Mean: 49.3
Days from diagnosis to germline testing	Median: 59
	Range: (0–2,032)
	Mean: 117.0
Pathogenic variant detected	Yes: 58 (13.6)
	No: 367 (86.4)
Personal history of other cancers	Yes: 198 (20.2)
	No: 781 (79.8)
Family history of pancreatic cancer	Yes: 95 (9.7)
	No: 806 (82.3)
	Unknown/not reported: 78 (8.0)
Smoking status	Never: 404 (41.3)
	Former: 380 (38.8)
	Current: 177 (18.1)
	Unknown/not reported: 18 (1.8)
BMI (kg/m2)	Median: 26.4
	Range: (14.1–60.4)
	Mean: 27.3
ECOG at presentation	0: 397 (40.6)
	1: 374 (38.2)
	2: 104 (10.6)
	3: 51 (5.2)
	4: 6 (0.6)
	Unknown/not reported: 47 (4.8)
Regional status at diagnosis	Localized disease: 453 (46.3)
	Metastatic disease: 526 (53.7)
Stage at diagnosis	Stage I/II: 161 (16.5)
	Stage III/IV: 527 (53.8)
	Unknown/not reported: 291 (29.7)
Tumor grade	G1—well-differentiated: 8 (0.8)
	G2—moderately differentiated: 544 (55.6)
	G3—poorly differentiated: 244 (24.9)
	Unknown/not reported: 183 (18.7)

Abbreviations: ECOG, Eastern Cooperative Oncology Group; OS, overall survival.

Germline testing prevalence by year of diagnosis was then assessed. For context, in November 2018, ASCO published a provisional clinical opinion that GGT be discussed with patients diagnosed with PDAC, even if the patient had no pertinent familial history of cancer.²⁸ The same month, NCCN published updated guidelines stating that clinicians should offer germline testing for all patients presenting with PDAC regardless of family history.⁸ Accordingly, the overall percentage of patients at RPCCC seen for PDAC who had GGT increased from 16.5% in 2017 to 23.9% in 2018 to 43.6% in 2019 and was 57.6% in 2024 (Fig 1A). Similarly, the median time to order and complete GGT declined after 2018 (Fig 1B, Appendix, Table A3). We then investigated why testing was not performed, stratified by year. The number of patients who did not have GGT because of the clinician never offering declined substantially from 2017 to 2023 (71 to 14, respectively) but doubled from 2023 to 2024 (from 14 to 28; Fig 1C, Appendix, Table A1). Patients who had GGT had a significantly better 5-year OS rate of 12.1% as compared with patients who did not receive GGT (1.8%, HR, 0.48 Fig 1D). Multivariable analysis was performed to correct for confounding factors and found that germline testing was indeed associated with longer OS (HR, 0.46; Table 2). The multivariable analysis further revealed that an ECOG performance status of 3+ was associated with shorter OS (HR, 3.15) as expected. Among the patients who were offered germline testing, patients with metastatic disease, patients with poorly differentiated tumors, patients with advanced-stage disease, patients not having a personal history of other cancers, patients not having a family history of pancreatic cancer, and older age patients were less likely to complete testing as compared with other patients (Table 3). We then investigated factors associated with the presence of a GPV and found that individuals with a personal history of other cancers (including nonmelanoma skin cancer) were found to be significantly more likely to harbor a GPV (OR, 2.57; Table 4).

FIG 1.

Prevalence of PDAC patient GGT at Roswell Park Comprehensive Cancer Center and its association with OS. (A) Percent of patients with PDAC who had GGT performed from 2017 to 2024. (B) Average time to order and complete GGT before and after clinical guidelines were updated in 2018. (C) Quantification of reasons why patients did not receive GGT from 2017 to 2024. (D) Kaplan-Meier plot displaying OS of patients with PDAC who did or did not receive GGT. GC, GGT completed; GGT, germline genetic testing; HR, hazard ratio; NG, no GGT completed; OS, overall survival; PDAC, pancreatic ductal adenocarcinoma.

TABLE 2.

Overall Survival Univariable and Multivariable Analyses

	Univariable		Multivariable
Characteristic	HR (95% CI)	P	HR (95% CI)	P
Germline testing
No	Reference	—	Reference	—
Yes	0.48 (0.42 to 0.56)	<.001	0.46 (0.38 to 0.57)	<.001
Age at diagnosis (years)
>68	Reference	—	Reference	—
≤68	0.74 (0.65 to 0.86)	<.001	0.85 (0.70 to 1.04)	.123
ECOG at presentation
0/1/2	Reference	—	Reference	—
3+	4.32 (3.22 to 5.80)	<.001	3.15 (2.07 to 4.79)	<.001
Regional status at diagnosis
Local	Reference	—	Reference	—
Metastatic	2.45 (2.12 to 2.84)	<.001	1.50 (0.21 to 10.81)	.687
Stage at diagnosis
I/II	Reference	—	Reference	—
III/IV	5.22 (4.08 to 6.68)	<.001	3.56 (0.49 to 26.09)	.211
Tumor grade
Poorly differentiated	Reference	—	Reference	—
Well-/moderately differentiated	0.71 (0.60 to 0.84)	<.001	0.84 (0.68 to 1.04)	.111

Abbreviations: ECOG, Eastern Cooperative Oncology Group; HR, hazard ratio.

TABLE 3.

Factors Associated With Completion of Germline Genetic Testing

Characteristic	OR (95% CI)	P
Sex at birth
Male	Reference	—
Female	0.89 (0.69 to 1.15)	.385
Racial background
White	Reference	—
Not White	0.81 (0.52 to 1.26)	.350
Age at diagnosis (years)
>68	Reference
≤68	1.66 (1.29 to 2.14)	<.001
Personal history of other cancers
Yes	Reference	—
No	0.43 (0.31 to 0.59)	<.001
Family history of pancreatic cancer
Yes	Reference	—
No	0.44 (0.28 to 0.67)	<.001
Smoking status
Never	Reference	—
Former/current	0.78 (0.61 to 1.02)	.07
BMI (kg/m2)
>26.4	Reference
≤26.4	1.27 (0.97 to 1.67)	.08
ECOG at presentation
0/1/2	Reference	—
3+	0.60 (0.33 to 1.04)	.074
Regional status at diagnosis
Localized	Reference	—
Metastatic	0.77 (0.60 to 1.00)	.047
Stage at diagnosis
I/II	Reference	—
III/IV	0.44 (0.30 to 0.62)	<.001
Tumor grade
Poorly differentiated (G3)	Reference	—
Well-/moderately differentiated (G1/2)	1.60 (1.18 to 2.19)	.003

Abbreviations: ECOG, Eastern Cooperative Oncology Group; OR, odds ratio.

TABLE 4.

Factors Associated With Having a Germline Pathogenic Variant

Characteristic	OR (95% CI)	P
Age at diagnosis (years)
>68	Reference	—
≤68	1.24 (0.67 to 2.32)	.496
Personal history of other cancers
Yes	Reference	—
No	0.39 (0.20 to 0.75)	.005
Family history of pancreatic cancer
Yes	Reference	—
No	1.43 (0.59 to 3.83)	.448
Stage at diagnosis
I/II	Reference	—
III/IV	0.92 (0.45 to 1.91)	.816

Abbreviation: OR, odds ratio.

Of genes associated with PDAC predisposition, seven patients each were found to harbor GPVs in ATM and BRCA2, five each in BRCA1 and PALB2, three in MSH6, two in CDKN2A, and one each in TP53 and STK11 (Appendix, Table A2, Fig 2A). Of the 17 patients with an actionable GPV in BRCA1/2 or PALB2 variants that are strongly correlated with high microsatellite instability/mismatch repair–deficient (MSI-H/MMR-D) tumors, seven were subsequently treated with the PARP inhibitor olaparib based on the results of their genetic testing (Figs 2B and 2C). Similarly, of the three patients with MSH6 variants correlated with MSI-H/MMR-D tumors, one patient subsequently received pembrolizumab as first-line therapy instead of standard chemotherapy (Figs 2B and 2D). Survival for the eight patients treated with targeted therapies based on the presence of GPVs was significantly improved as compared with patients with actionable GPVs but who received standard-of-care therapy (HR, 0.09, Fig 2E).

FIG 2.

Detection of actionable germline mutations and subsequent informed treatment strategy. (A) Oncoprint of the most prevalent germline mutations detected in patients with PDAC seen at Roswell Park Comprehensive Cancer Center from 2017 to 2024. (B) Actionable germline mutations identified in four genes (BRCA1/2, PALB2, and MSH6) with the number of patients identified with the mutation and the number of patients subsequently treated with olaparib or pembrolizumab. Treatment strategy and timeline for patients with an identified germline mutation in either the (C) BRCA1/2 or (D) MSH6 gene. DX, diagnosis; Gem + Cis, gemcitabine plus cisplatin. The red triangle denotes death. (E) OS of patients identified as having actionable BRCA1/2, PALB2, or MSH6 mutations from GGT and subsequently receiving either personalized or standard-of-care therapy. GGT, germline genetic testing; HR, hazard ratio; OS, overall survival; PDAC, pancreatic ductal adenocarcinoma; SOC, standard of care.

DISCUSSION

Here, we review the impact that the updated guidelines from ASCO²⁸ and NCCN⁸ published in late 2018 had on the incidence of GGT and outcomes in patients with PDAC at RPCCC. We also report the factors associated with completing genetic testing, the factors associated with having a GPV, and the genes within the germline that are most frequently identified as harboring a pathogenic variant and how these findings altered the therapeutic strategy for the patient.

Consistent with other studies, we found that patients with increased age and advanced disease had a lower likelihood of completing GGT.^29,30 However, we also uniquely found that patients without a personal history of other cancers and without a family history of PDAC were also less likely to complete testing. This study also reveals an association between completion of GGT and increased OS even after correcting for confounding variables such as older age and advanced disease, as mentioned above. These findings highlight the need to educate patients on the benefits of GGT and increase testing in these at-risk populations and in all patients.

The size of our study compared favorably with other similar reports with the inclusion of 979 patients, overshadowed only by the report from Johns Hopkins that included 992 patients.³⁰ Our GGT completion rate of 57.6% in 2024 at RPCCC fell within the range of completion rates reported by other cancer centers that implemented various approaches to increase testing, such as Mayo Clinic (29.7%),²⁹ Dana-Farber (38%),³¹ Mount Sinai (61%),³² Markey Cancer Center (81.5%),¹⁶ and Johns Hopkins (82.2%).³⁰ It was surprising that the number of patients who did not have GGT offered to them by their clinician slightly increased from 2023 to 2024 in our study; this likely has to do with test offering variability between the clinicians on service during that time frame. Nevertheless, our GPV detection rate of 13.6% was consistent with the frequency of detection in other studies of patients with PDAC.^16,29–33

Importantly, our study shows improved survival for the eight patients treated with targeted therapies based on the identification of a GPV, suggesting that testing can improve survival for patients with poor performance status or other adverse prognostic features, although larger prospective studies matched for these characteristics will be required to prove this tantalizing hypothesis. In addition, two patients from our study with actionable GPVs identified in MSH6 and BRCA1 also had a family history of PDAC. This would have credentialed them for PDAC surveillance if germline testing had been performed for their family members with PDAC, highlighting the power of testing to improve outcomes beyond the immediate individual being tested.

A limitation of our study is that several factors previously reported to be associated with a lower probability of completing genetic testing were not assessed, such as lower socioeconomic status,³⁴ not being married,^30,35,36 and living further from the clinic.²⁹ Another limitation is that the general catchment area of RPCCC is predominantly White, limiting study representation of other race/ethnicities.^37–39 Finally, the current study encapsulated the entire arc of the COVID-19 pandemic that drastically altered clinical operations and decreased the willingness of immunocompromised individuals to be exposed to infected individuals, which likely decreased testing rates, as is reflected in our data (Fig 1A). It would also be interesting in future studies to capture data related to the prevalence of de novo germline mutations.

While the overall trends are promising, 42.4% of patients in 2024 still did not have GGT and 50% of those patients’ reason for not completing testing was that the clinician never offered. Similarly, the median time to GGT completion was 59 days and the median OS for patients in this study was 8.8 months. This highlights a need to increase prevalence of GGT and consider new consultation workflows that have been shown to improve timely genetic testing completion at other cancer centers in the United States.^16,30 As we and others demonstrate, testing is not offered to patients or completed by patients for multiple reasons. Based on the importance of timely germline testing for guiding therapeutic decision making and improving outcomes, we argue that it is critical to ensure that all patients with PDAC across all populations are aware that genetic testing is available and could be of significant benefit both to them and their family members as surveillance improves outcomes.⁴⁰ To that end, point-of-care testing has been implemented by the Department of Clinical Genomics for all patients diagnosed with or treated for PDAC at RPCCC, which should substantially increase testing prevalence. The hope is that with improved education of patients and clinicians alike regarding the critical importance of GGT for both the patient and their family, we can improve outcomes for all patients with PDAC seen at RPCCC and their families.

CONTEXT.

Key Objective

Has germline genetic testing (GGT) prevalence in individuals with pancreatic ductal adenocarcinoma (PDAC) progressed over time and improved patient outcomes?

Knowledge Generated

While GGT prevalence of individuals with PDAC has increased from 16.5% in 2017 to 57.6% in 2024, testing remains underutilized particularly among older individuals and individuals without a family history of PDAC, without a personal history of other cancers, and with advanced disease. A pathogenic variant was detected in 13.6% of individuals who completed testing. Individuals who received therapy informed by testing results as compared with standard of care displayed improved overall survival (HR, 0.09).

Relevance

Increased GGT through universal point-of-care protocols can reveal therapeutic vulnerabilities and improve outcomes for both the individual and at-risk family members through surveillance and early detection.

DATA SHARING STATEMENT

Data are available from the corresponding author upon reasonable request.