Preparing for the inevitable: early comprehensive genomic profiling for patients with localized pancreas ductal adenocarcinoma

Abstract

PURPOSE:

Pancreatic ductal adenocarcinoma (PDAC) is a highly lethal disease, even for patients diagnosed at a radiographically localized stage. Despite disease recurrence in most patients who undergo curative-intent therapy, national guidelines do not currently recommend comprehensive genomic profiling (CGP) until progression occurs following treatment. As the number of approved and investigational therapies expands for PDAC, ensuring access to early CGP data is critical for implementing molecularly informed treatment strategies.

PATIENTS AND METHODS:

We performed a single-institution, retrospective analysis of CGP utilization and treatment patterns in patients with localized PDAC who underwent curative-intent resection in the contemporary period during which CGP was recommended for all patients with advanced/recurrent disease (2019–2024).

RESULTS:

Among 380 patients, 215 (56.5%) developed recurrence at a median follow-up time of 19.4 months. Of patients with documented recurrence, we found that 43.2% never underwent CGP, and among those who did have CGP performed, only 61.5% had access to these data prior to initiating post-recurrence systemic therapy. In terms of barriers to CGP, we found that earlier testing (within two months of surgery) improved access to CGP prior to initiation of post-recurrence treatment, while deferred testing was associated with a greater chance of molecularly uninformed treatment. We also found that patients whose surgery and systemic therapy was fragmented across centers were less likely to receive CGP (65.6% vs. 28.8%, P<0.0001). CGP data was successfully obtained using several modalities, and receipt of CGP was associated with higher rates of molecularly tailored treatment (10.6% vs 1.1%, P=0.004), including 10 patients (8.2%) who received treatment as part of a clinical trial.

CONCLUSION:

A uniform strategy for ensuring resected PDAC patients receive CGP may improve access to clinical trials and molecularly tailored treatment.

Introduction

For patients diagnosed with radiographically localized pancreatic ductal adenocarcinoma (PDAC), post-treatment recurrence rates remain exceedingly high, and five-year survival is persistently low. In modern clinical trials evaluating surgery and multiagent systemic therapy approaches, fewer than one-third of patients remain recurrence-free five years following treatment.^1,2 Real-world outcomes among patients undergoing curative-intent resection suggest that 10-year survival rates are <5%.³ Once recurrence is detected, clinical trial enrollment is recommended, as second-line therapy has a response rate of only 10–20%, and median survival is less than one year.^4–6 Nonetheless, fewer than 5% of all patients with PDAC enroll in a clinical trial.⁷

Beginning in 2019, comprehensive genomic profiling (CGP) was recommended for all patients with advanced PDAC.⁸ The initial motivation behind universal CGP was the significant rates of targetable gene fusions and BRAF alterations in KRAS wild type PDAC, the known efficacy of platinum-based and PARP inhibitor regimens for patients with BRCA1/2 and PALB2 altered tumors, and the rare identification of tumors with microsatellite instability responsive to immune checkpoint inhibition.^9–12 More recently, several promising therapeutic targets have emerged in molecularly defined subgroups such as KRAS mutations (~90% of patients), MTAP loss (~20% of patients), and claudin 18.2 overexpression (~50–70% of patients).^13,14 Agents with proven clinical efficacy in these molecular subgroups are currently under investigation in the metastatic, recurrent, and adjuvant/neoadjuvant settings, underscoring the critical need for access to CGP at the time of treatment planning. Notably, national guidelines do not currently recommend CGP for patients with localized PDAC amenable to surgical resection, despite the emergence of trials across disease stages and the fact that most patients experience disease recurrence following curative-intent treatment.

Here, we present our contemporary institutional experience in obtaining CGP for patients with recurrent PDAC treated with curative-intent. Despite national guidelines supporting CGP in the post-recurrence setting and readily available tissue, we found that failure to obtain CGP was common, and that deferring CGP beyond the immediate post-operative period was often associated with molecularly uninformed first-line treatment decisions. We explore clinicodemographic factors that may influence CGP access. We also evaluate how the timing of CGP influences the availability of results at the time of post-recurrence treatment as well as access to molecularly tailored treatment and clinical trial enrollment.

Methods

A retrospective analysis of an IRB-approved (Pro00111472), prospectively maintained database was performed in order to identify all patients with documented recurrence of PDAC following curative-intent surgery at Duke Health from January 1, 2019 to June 6, 2024. All data, including CGP utilization, was obtained via electronic health record review up to a data cutoff of May 1, 2025. Availability of complete patient records, including treating oncologist documentation and CGP reports, was required for study inclusion. Patients with partial documentation of CGP receipt, but without complete records of the CGP report available were excluded from analysis (N=3). All patients received surgical care and post-surgical follow up at Duke. Patients were considered to have fragmented care if they received at least a portion of their post-recurrence treatment at a setting outside of Duke Health. Prior to analysis, these data were logged in a protected computer environment (PACE). All data analysis and visualization was performed in R (version 4.3.2).

Patients were categorized by receipt of CGP, and then further stratified by timing of CGP relative to initiation of post-recurrence systemic therapy. Patients were classified as receiving CGP if an FDA approved multi-gene sequencing-based panel was performed on tumor tissue or circulating tumor DNA. Patients with single-gene testing or germline sequencing only were not considered to have undergone CGP. Data from CGP were collated from test reports, including those generated by FoundationOne, Caris, Guardant, and PGDx. Homologous Recombination and Repair (HRR) pathway mutations were defined as BRCA1, BRCA2 and PALB2.

Descriptive statistics are presented as raw numbers and proportions for categorical variables and as median/interquartile range (IQR) for continuous variables. Comparisons between groups were performed using Fisher’s exact test for categorical variables and Mann-Whitney U test for continuous variables. Survival estimates were performed using the Kaplan-Meier method.

Results

During the study period, a total of 380 patients with PDAC underwent curative-intent surgery at Duke Health, and at a median follow-up of 19.4 months (IQR 11.2–34.8), 218 (55.6%) had developed post-treatment recurrence (Fig. 1A). Median time-to-recurrence among all patients was 17.3 (IQR 7.6–65.6) months from resection (Fig. 2B), and median overall survival was 31.4 (IQR 14.0-NR) months from resection (Fig. 2C). Among patients who experienced disease recurrence, median time-to-recurrence was 9.6 months (IQR 4.8–15.7), and 62.8% went on to receive post-recurrence systemic therapy prior to our data cutoff. Three patients with disease recurrence had insufficient CGP documentation for inclusion in further analyses.

Figure 1.

The PDAC cohort included for analysis. A) Flow diagram describing patients included in our study. B) Kaplan-Meier curve showing time from surgery to recurrence for PDAC patients with localized disease. C) Kaplan-Meier curve showing overall survival from time of surgery for PDAC patients with localized disease. The dashed lines on the Kaplan-Meier curves represent 95% confidence intervals.

Figure 2.

Timing of CGP relative to recurrence and treatment. A) Swimmers plot with black lines indicating the time to death (X) or last follow (O) for each patient who received CGP (red square), recurred (blue square), and subsequently received systemic therapy (yellow square). B) Line plot showing the cumulative proportion of patients with no CGP results available prior to post-recurrence systemic therapy as a function of the delay in CGP post-surgery. All patients (N=15) who received CGP within 2 months had results available prior to post-recurrence treatment.

Of the 215 evaluable patients with documented recurrence during the follow-up period, 93 (43.2%) did not undergo CGP. There were several clinicodemographic and treatment features associated with CGP utilization (Table 1). Older patients were less likely to undergo CGP, as were patients with an ECOG performance status ≥2 at the time of recurrence. CGP receipt was also less common in patients who did not receive adjuvant chemotherapy. Patients who underwent fragmented care across health systems were significantly less likely to undergo CGP (28.8 vs. 65.6%, P<0.0001). Notably, there were no differences in race or ethnicity, nor in insured status. No pathologic features were significantly associated with CGP use (Supplementary Table 1). Patients who underwent CGP were more likely to receive post-recurrence systemic therapy (P<0.0001); however, nearly half (46.2%) of patients without CGP still received systemic therapy post-recurrence.

Table 1.

Clinical and demographic features based on receipt of CGP

Variable	Total	Underwent CGP	No CGP	P Value
	(n=215)	(n=122)	(n=93)
Age, years, median [IQR]	69 [62,75]	68 [60,74]	71 [65,76]	0.0194
Gender, n (%)
Male	105 (48.8)	59 (48.4)	46 (49.5)	0.891
Female	110 (51.2)	63 (51.6)	47 (50.5)
Race, n (%)
Caucasian/White	164 (76.3)	95 (77.9)	69 (74.2)	0.872
Black/African American	35 (16.3)	18 (14.7)	17 (18.3)
Other	8 (3.7)	4 (3.3)	4 (4.3)
Not Reported/Unknown	8 (3.7)	5 (4.1)	3 (3.2)
Ethnicity, n (%)
Non-Hispanic	189 (87.9)	108 (88.5)	81 (87.1)	0.657
Hispanic	4 (1.9)	3 (2.5)	1 (1.1)
Not Reported/Unknown	22 (10.2)	11 (9.0)	11 (11.8)
ECOG PS at time of recurrence, n (%)
0	78 (36.3)	48 (39.3)	30 (32.2)	0.0365
1	108 (50.2)	65 (53.3)	43 (46.2)
2	16 (7.4)	6 (4.9)	10 (10.8)
≥3	7 (3.3)	1 (0.8)	6 (6.5)
Not Reported/Unknown	6 (2.8)	2 (1.7)	4 (4.3)
Primary health insurance, n (%)
Private	104 (48.4)	60 (49.2)	44 (47.3)	0.939
Medicare	88 (40.9)	48 (39.3)	40 (43.0)
Medicaid	4 (1.8)	3 (2.5)	1 (1.1)
Veterans Affairs	1 (0.5)	1 (0.8)	0
None	18 (8.4)	10 (8.2)	8 (8.6)
Received neoadjuvant chemotherapy, n (%)	116 (54.0)	67 (54.9)	49 (52.7)	0.783
Received adjuvant chemotherapy, n (%)	138 (64.2)	86 (70.5)	52 (55.9)	0.0317
Site of post-recurrence treatment, n (%)
Duke Health	163 (75.8)	107 (87.7)	56 (60.2)	<0.0001
Outside hospitals	52 (24.2)	15 (12.3)	37 (39.8)
Preoperative CA 19–9, median [IQR]	34 [15,83.3]	29 [18,73]	39 [12.8,120.5]	0.599
Time to recurrence, days, median [IQR]	292 [146.5,480.5]	274.5 [133.8,464]	346 [198,499]	0.21

For patients who underwent CGP (N=122, 55.6%), testing was obtained before starting systemic therapy in 47 patients (38.5%) and after initiating systemic therapy in 47 patients (38.5%), while 28 (23.0%) underwent CGP without going on to receive post-recurrence systemic therapy. There were no statistically significant demographic or clinicopathologic differences between patients who received CGP before versus after initiation of post-recurrence systemic therapy (Table 2 and Supplementary Table 2). Importantly, earlier CGP acquisition relative to surgical resection was associated with a higher likelihood of having results available prior to initiation of post-recurrence therapy (Fig. 2A). For example, two patients underwent CGP prior to resection and 21 underwent CGP within two months of resection. For all patients in this early testing cohort, CGP results were available prior to initiation of treatment post-recurrence. In contrast, when patients underwent CGP more than two months after resection, we observed a steep increase in the likelihood of initiating post-recurrence therapy without CGP results available (Fig. 2B).

Table 2.

Clinical and demographic features based on timing of CGP

Variable	CGP Before Chemotherapy (n=47)	CGP After Chemotherapy (n=47)	P Value
Age, years, median [IQR]	67 [60.5,72.5]	66 [58.5,74.5]	0.882
Gender, n (%)			0.68
Male	24 (51.1)	21 (44.7)
Female	23 (48.9)	26 (55.3)
Race, n (%)			0.492
Caucasian/White	37 (78.7)	40 (85.1)
Black/African American	7 (14.9)	5 (10.6)
Other	2 (4.3)	0
Not Reported/Unknown	1 (2.1)	2 (4.3)
Ethnicity, n (%)			>0.999
Non-Hispanic	43 (91.5)	42 (89.4)
Hispanic	1 (2.1)	1 (2.1)
Not Reported/Unknown	3 (6.4)	4 (8.5)
ECOG PS at time of recurrence, n (%)			0.081
0	22 (46.8)	15 (31.9)
1	24 (51.1)	29 (61.7)
2	0	3 (6.4)
≥3	0	0
Not Reported/Unknown	1 (2.1)	0
Primary health insurance, n (%)			0.492
Private	19 (40.4)	26 (55.3)
Medicare	21 (44.7)	16 (34.0)
Medicaid	2 (4.3)	1 (2.1)
Veterans Affairs	0	1 (2.1)
None	5 (10.6)	3 (6.4)
Received neoadjuvant chemotherapy, n (%)	27 (57.4)	26 (55.3)	>0.999
Received adjuvant chemotherapy, n (%)	35 (74.5)	31 (65.9)	0.499
Site of post-recurrence treatment, n (%)			0.677
Duke Health	43 (91.5)	45 (95.7)
Outside hospitals	4 (8.5)	2 (4.3)
Preoperative CA 19–9, median [IQR]	34 [19,75.8]	29 [19.8,79.3]	0.789
Time to recurrence, days, median [IQR]	248 [182.5,460.5]	281 [112,465.5]	0.621

Results from CGP demonstrated genomic findings similar to other large PDAC sequencing efforts (Fig. 3A).¹⁵ The vast majority of patients had KRAS-mutated tumors (95.1%), which is an emerging therapeutic target with multiple inhibitors either approved or under investigation. Aside from KRAS, 16.4% had tumors with MTAP loss, which may be sensitive to PRMT5 inhibitors currently under investigation. An additional 12 (9.8%) patients had tumors with relevant findings for existing FDA approved therapies, including alterations in BRCA1/2 or PALB2 (olarparib, platinum chemotherapy), mismatch repair genes (pembrolizumab), ERBB2 amplification (trastuzumab), BRAF^V600E (dabrafenib plus trametinib), and KRAS^G12C (sotorasib or adagrasib). A variety of CGP approaches were used, including samples from resection specimens (82.2%), fine-needle aspirations/biopsies prior to resection (13.9%), and blood-based circulating tumor DNA (4.9%) (Fig. 3B). The mutation profiles observed were similar across all CGP methods; however, no MTAP deletions or FDA approved targets were found in the limited number of ctDNA tests. Once CGP was ordered, median time to reporting was 15 days (IQR 12–20 days), with 23.6% of patients having a delay greater than three weeks. There were no significant differences in the time interval from provider order and result reporting by CGP method, though there was a trend toward shorter turnaround times in blood-based CGP (median of 11 days) relative to FNA-based (median of 13 days) and resected specimen-based (median of 15 days) CGP.

Figure 3.

Actionable CGP results in resected PDAC patients. A) An Oncoprint displaying alterations identified by CGP in patients with resected PDAC (N=123). Genes with alterations in >5% of our resected cohort or with corresponding investigational or approved therapies (red) were included. B) Table detailing actionable targets identified by CGP. FDA approved therapies include those approved or have NCCN guideline support for use in PDAC. Only clinical trials currently active at Duke are included. C) Donut chart with inner ring showing the proportion of patients who received CGP via blood-based ctDNA (red), fine needle aspiration (blue), or resected tumor (yellow). The outer ring shows the proportion of patients within each testing approach with FDA approved targets (black), clinical trial targets (green), both trial and approved targets (gray), or no targets (white).

While use of molecularly tailored treatment was overall rare in our cohort, there was a statistically significant difference based on patient access to CGP (Fig. 3C). For example, 13 (10.6%) patients who underwent CGP received molecularly tailored therapy, including those treated as part of a clinical trial (N=10, 8.2%) or with an FDA approved regimen (N=3, 2.4%). In contrast, only one of 93 patients without CGP (1.1%) received a molecularly tailored therapy (olarparib) after germline testing revealed a BRCA2 mutation. Furthermore, earlier CGP was associated with access to these molecularly tailored therapies as first-line therapy post-recurrence. Amongst the subset of patients who underwent CGP prior to initial post-recurrence treatment (N=47), 6.4% entered a clinical trial as first-line therapy, and an additional 2.1% received first-line platinum chemotherapy after an HRR pathway mutation was found on testing. For patients who received CGP after initiating first line therapy (N=47), all molecularly tailored therapies were given in the second line (N=6, 4.9%) and of the three patients with a documented HRR pathway mutation, only one received a platinum chemotherapy in the second line and two never received a platinum chemotherapy.

Discussion

In this retrospective analysis of patients with early stage PDAC, we found that most subjects experiencing disease recurrence either did not receive CGP or did not have results available prior to initiating post-recurrence systemic therapy. Consistent with nearly all prior studies, we found that the majority of patients who underwent resection with curative intent nonetheless experienced disease recurrence, typically within one year of surgery, and that most of these patients went on to receive systemic therapy. Because diagnostic or resected tissue is available for all of these patients far in advance of their recurrence, this cohort represents a population for whom universal access to early CGP is feasible and practical. In fact, we found that CGP could be performed on either FNA biopsies or resected tissue with comparable resulting molecular profiles and turnaround times.

At present, national guidelines do not recommend CGP for patients with localized PDAC until post-resection recurrence is documented. However, we found that obtaining CGP early—ideally within two months of resection—ensured available CGP results prior to initiation of post-recurrence systemic therapy for all patients in our cohort, while deferring testing beyond this point was associated with molecularly uninformed treatment in nearly half of patients. In conflict with this observation is the fact that multiple national consensus guidelines for management of recurrent PDAC recommend pursuit of a first-line clinical trial, most of which require CGP, and several existing FDA-approved systemic therapies require knowledge of a patient’s molecular profile.^4,16 Thus, a lack of CGP data at the time of recurrence may compromise the ability of providers to uniformly follow these guidelines. In our cohort, a small but meaningful proportion of patients with CGP data available were treated with a molecularly tailored approach, many as part of a clinical trial. We acknowledge that earlier CGP use may add to the financial burden of PDAC treatment, and we encourage future studies exploring the cost-effectiveness of CGP, which we expect to become increasingly favorable as the cost of sequencing lowers and the number of approved therapeutic targets inevitably increases as has been observed in other cancer types.¹⁷ Additional risks of CGP that should be considered include threats to genetic privacy, feelings of guilt in the setting of results with hereditary implications, and patient mistrust when goals of testing are not effectively communicated.¹⁸ In weighing these risks versus benefits, it is extremely important to consider that PDAC patients facing disease recurrence are at an extremely high risk of rapid progression and compromised performance status, ultimately limiting therapeutic options.⁶ In fact, nearly half of the patients in our cohort who were treated post-recurrence received only a single line of therapy, indicating that selection of initial therapy is critical. Early CGP provides these patients the best opportunity to inform therapeutic decisions based on potentially actionable features.

As CGP becomes more widely used in PDAC and other cancer types, it is important to identify barriers to implementation. We found that care fragmentation across institutions was one notable barrier to CGP, which is likely a product of logistical challenges in testing resected tissue residing at a different hospital from where patients are undergoing postoperative therapy and surveillance. There are likely numerous additional barriers to CGP access, such as pathology workflows and provider comfort ordering and interpreting CGP results, that are difficult to fully account for based on the treatment records available for our analysis. Uniform testing guidelines could ensure that CGP data are available for all patients at the time of recurrence, and we would advocate for universal testing on all patients undergoing curative-intent surgery, either utilizing diagnostic or resected tissue. The former approach will become particularly important as molecularly defined clinical trials inevitably expand in the perioperative setting, as standard CGP of FNA specimens will streamline screening processes and enable trial access. Rather than deferring CGP until documented recurrence, early testing will prepare for recurrence in most patients, and consensus guidelines supporting this approach will ensure that reimbursement practices do not act as a further barrier to obtaining these critical data.

Our study has several important limitations. First, this is a retrospective analysis from a single institution, thus these data may not generalize to all patient populations. We represent a high-volume academic program with an established CGP pipeline that extends to our community partners, thus we suspect that our results likely overestimate rates of CGP in this patient population. As community clinics treat the majority of cancer patients nationally, direction from our national guidelines on uniform testing strategies that can be adopted by all providers is critical for ensuring broad patient access. Second, we intentionally limited the scope of this analysis to the contemporary period during which CGP was recommended for patients with advanced/recurrent disease (2019 to present), thus our median follow-up time of 19.4 months does not capture all patients who will ultimately experience recurrence. As such, our documented 55.6% recurrence rate over this period likely underestimates the population at-risk for a failure to obtain early CGP results. Third, no patients in our cohort received CGP with an RNA-based assay, which is a more sensitive approach to detect gene fusions; this likely explains the relative lack of fusions detected, even in KRAS wild type patients, and further suggests that we may be underestimating the number of actionable targets when CGP is universally performed in this setting.¹⁹ Finally, we were unable to uniformly capture failed attempts at CGP, which might occur due to sample insufficiency, although we anticipate this to occur at an exceedingly low event rate given that resected specimens are available for analysis in this patient population.²⁰

In summary, we found that CGP was feasible and actionable in patients treated for localized PDAC, yet failure to obtain these data prior to disease recurrence was common. This is a disease where most patients are at exceedingly high risk for recurrence even when diagnosed and treated at an early stage, and progress in advancing therapies has been historically stagnant. Recently, a number of promising molecularly targeted strategies have emerged that foster optimism as they enter the first-line clinical trial setting, and inevitably patients will require timely CGP to access these therapies. We would encourage guidelines to reflect the need for molecularly informed treatment planning: namely that CGP be performed early in all patients with PDAC, including those with localized disease undergoing treatment with curative intent.

Context Summary.

Key objective:

What proportion of Pancreas Ductal Adenocarcinoma (PDAC) patients who recur after curative-intent treatment have access to Comprehensive Genomic Profiling (CGP) and molecularly tailored therapy?

Knowledge generated:

We found that 43.2% of patients who experienced disease recurrence never underwent CGP and 21.8% received CGP after initiation of post-recurrence systemic therapy. Receipt of CGP was associated with a higher likelihood of molecularly tailored post-recurrence treatment (10.6% vs 1.1%) and participation in a clinical trial (8.2% vs 0%).

Relevance:

PDAC patients who do not receive CGP may have limited access to clinical trials and the rapidly expanding repertoire of molecularly tailored treatment.