Research

JAMA Oncology | Original Investigation

Pancreatic Cancer Surveillance and Survival of High-Risk Individuals

Amanda L. Blackford, ScM; Marcia Irene Canto, MD, MHS; Mohamad Dbouk, MD; Ralph H. Hruban, MD;
Bryson W. Katona, MD, PhD; Amitabh Chak, MD; Randall E. Brand, MD; Sapna Syngal, MD; James Farrell, MBChB;
Fay Kastrinos, MD; Elena M. Stoffel, MD; Anil Rustgi, MD; Alison P. Klein, PhD, MHS; Ihab Kamel, MD;
Elliot K. Fishman, MD; Jin He, MD; Richard Burkhart, MD; Eun Ji Shin, MD;
Anne Marie Lennon, MBBCh, PhD; Michael Goggins, MBBCh, MD

Supplemental content
IMPORTANCE Pancreatic ductal adenocarcinoma (PDAC) is a deadly disease with increasing
incidence. The majority of PDACs are incurable at presentation, but population-based
screening is not recommended. Surveillance of high-risk individuals for PDAC may lead to
early detection, but the survival benefit is unproven.

OBJECTIVE To compare the survival of patients with surveillance-detected PDAC with US

national data.

DESIGN, SETTING, AND PARTICIPANTS This comparative cohort study was conducted in
multiple US academic medical centers participating in the Cancer of the Pancreas Screening
program, which screens high-risk individuals with a familial or genetic predisposition for
PDAC. The comparison cohort comprised patients with PDAC matched for age, sex, and year
of diagnosis from the Surveillance, Epidemiology, and End Results (SEER) program. The
Cancer of the Pancreas Screening program originated in 1998, and data collection was done
through 2021. The data analysis was performed from April 29, 2022, through April 10, 2023.

EXPOSURES Endoscopic ultrasonography or magnetic resonance imaging performed annually

and standard-of-care surgical and/or oncologic treatment.

MAIN OUTCOMES AND MEASURES Stage of PDAC at diagnosis, overall survival (OS), and PDAC
mortality were compared using descriptive statistics and conditional logistic regression, Cox
proportional hazards regression, and competing risk regression models. Sensitivity analyses
and adjustment for lead-time bias were also conducted.

RESULTS A total of 26 high-risk individuals (mean [SD] age at diagnosis, 65.8 [9.5] years; 15
female [57.7%]) with PDAC were compared with 1504 SEER control patients with PDAC
(mean [SD] age at diagnosis, 66.8 [7.9] years; 771 female [51.3%]). The median primary tumor
diameter of the 26 high-risk individuals was smaller than in the control patients (2.5 [range,
0.6-5.0] vs 3.6 [range, 0.2-8.0] cm, respectively; P < .001). The high-risk individuals were
more likely to be diagnosed with a lower stage (stage I, 10 [38.5%]; stage II, 8 [30.8%]) than
matched control patients (stage I, 155 [10.3%]; stage II, 377 [25.1%]; P < .001). The PDAC
mortality rate at 5 years was lower for high-risk individuals than control patients (43% vs
86%; hazard ratio, 3.58; 95% CI, 2.01-6.39; P < .001), and high-risk individuals lived longer
than matched control patients (median OS, 61.7 [range, 1.9-147.3] vs 8.0 [range, 1.0-131.0]
months; 5-year OS rate, 50% [95% CI, 32%-80%] vs 9% [95% CI, 7%-11%]).

CONCLUSIONS AND RELEVANCE These findings suggest that surveillance of high-risk

individuals may lead to detection of smaller, lower-stage PDACs and improved survival.

Author Affiliations: Author

affiliations are listed at the end of this
article.
Corresponding Author: Marcia Irene
Canto, MD, MHS, Division of
Gastroenterology, Johns Hopkins
Medicine, 1800 Orleans St,
JAMA Oncol. 2024;10(8):1087-1096. doi:10.1001/jamaoncol.2024.1930 Blalock 407, Baltimore, MD 21287
Published online July 3, 2024. Corrected on September 12, 2024. ([email protected]).

(Reprinted) 1087

Downloaded from jamanetwork.com by University of Gent / UZGent Kenniscentrum user on 10/10/2024

 Research Original Investigation Pancreatic Cancer Surveillance and Survival

P
ancreatic ductal adenocarcinoma (PDAC) is a highly
lethal disease.1 In the US, it is the eighth most com- Key Points
mon cancer but ranks third as a cause of cancer death.2
Question What is the association of selective pancreatic cancer
While the overall survival (OS) for individuals with PDAC has surveillance with survival?
improved in recent years, it remains poor, with approxi-
Findings In this cohort study comparing 26 individuals with a
mately 11% alive 5 years after diagnosis.2,3 The majority of
genetic or familial predisposition undergoing pancreatic cancer
PDACs are diagnosed at an advanced, often locally unresect-
surveillance and 1504 matched control patients, surveillance led to
able or metastatic stage, which in turn directly affects a significantly higher proportion of stage I cancers (31% vs 10%),
survival. However, population-based screening for PDAC in av- longer 5-year survival rate (50% vs 9%), and lower pancreatic
erage-risk, asymptomatic individuals is not recommended.4 cancer–specific mortality rate (43% vs 86%).
For the past 2 decades, there has been increasing interest
Meaning These findings suggest that selective surveillance of
and accumulating data on the diagnostic yield of selective sur- individuals at high risk for pancreatic cancer may improve clinical
veillance in high-risk individuals with a family history of PDAC outcomes.
or with a genetic predisposition due to an inherited patho-
genic variant.5-9 The Cancer of the Pancreas Screening (CAPS)
program, consisting of high-risk individuals with inherited
genetic variants or familial pancreatic cancer kindreds, re- Table 1. Characteristics of 26 High-Risk Individuals Diagnosed
ported 10 PDACs diagnosed in 1461 study participants, with With PDAC During Screening, 1998-2021
78% diagnosed during surveillance at stage I using endo- No. of high-risk individuals
Characteristic with PDAC (%)
scopic ultrasonography (EUS) and/or magnetic resonance
Age at diagnosis, mean (SD), y 65.8 (9.5)
imaging (MRI).10 The 20-year Dutch program reported its ex-
Sex
perience with 347 carriers of deleterious germline CDKN2A
Female 15 (57.7)
variants who underwent surveillance primarily with MRI.
Among study participants who developed PDAC, 33.3% were Male 11 (42.3)

diagnosed with stage I disease, and 83% had resectable White race 26 (100)

disease.11 Familial pancreatic cancer 16 (61.5)

Although these data suggest that surveillance can detect Germline variant carriers 10 (38.5)
earlier disease, the effectiveness of surveillance for improve- Follow-up, median (range), y 2.5 (0.2-13.0)
ment of PDAC survival in high-risk individuals is unproven.12 Time from baseline to diagnosis for 5.8 (1.1-11.2)
nonprevalent cases, median (range), ya
We aimed to determine whether there was a shift to lower-
PDAC diagnosed at baseline, prevalent cases 6 (23.1)
stage PDAC and an associated improved long-term survival in
high-risk individuals compared with matched control pa- Abbreviation: PDAC, pancreatic ductal adenocarcinoma.
a
tients after considering factors known to be associated with Number of nonprevalent cases was 20.
survival.
workshop. The CAPS cohort studies were approved by the
Johns Hopkins institutional review board. This study was
exempt from review because it used prospectively collected
Methods data from the prior CAPS studies that obtained patients’
Study Population informed consent. High-risk individuals eligible for surveil-
This comparative cohort study examined high-risk lance provided written informed consent. This study fol-
individuals enrolled in the CAPS program, 13 an ongoing, lowed the Strengthening the Reporting of Observational
multicenter, prospective, institutional review board– Studies in Epidemiology (STROBE) reporting guideline.
approved cohort study. CAPS originated in 1998 as a single-
center clinical program and observational study at the Johns High-Risk Individuals
Hopkins Hospital.14 CAPS1 was a pilot study involving 38 High-risk individuals were eligible for surveillance if they
high-risk individuals14; CAPS2, a prospective cohort study were asymptomatic at baseline. Participants underwent
of 78 high-risk individuals and 149 control participants15; baseline evaluation with a comprehensive questionnaire,
and CAPS3 (2007-2010), a multicenter study that compared EUS, and abdominal contrast-enhanced CT or 1.5T or 3T
the diagnostic yield of computed tomography (CT), MRI, contrast-enhanced MRI with magnetic resonance cholangio-
and EUS in 225 high-risk individuals. CAPS4 (2008-2013) pancreatography according to study protocol (1998-2013) or
was a Johns Hopkins single-center study that evaluated the international CAPS Consortium consensus guidelines (since
yield of pancreatic surveillance and biomarkers for early 2013).16,17 After baseline, participants continued annual sur-
detection. The ongoing CAPS5 (since 2014) is a multicenter veillance unless there was a clinical need for a shortened
study of 8 academic centers.5,10 Experts performed EUS and interval follow-up based on detected abnormalities.
MRI at each site. The CAPS imaging protocols and reporting The high-risk individuals were enrolled based on either
have been standardized since CAPS3 and have followed a family history criteria, which included individuals who had at
2004 CAPS consensus of a joint National Cancer Institute least 1 first-degree relative with PDAC and who were part of a
and Specialized Program of Research Excellence–sponsored kindred with at least 1 pair of affected first-degree relatives

1088 JAMA Oncology August 2024 Volume 10, Number 8 (Reprinted) jamaoncology.com

Downloaded from jamanetwork.com by University of Gent / UZGent Kenniscentrum user on 10/10/2024

 Pancreatic Cancer Surveillance and Survival Original Investigation Research

Table 2. Characteristics of High-Risk Individuals With PDAC, Matched SEER Control Patients With PDAC,
and the Full SEER Cohort Eligible for Matching

No. (%)
Matched SEER
High-risk control patients
individuals with PDAC Eligible SEER cohort
Characteristic (n = 26) (n = 1504) (n = 66 987) P valuea
Age at diagnosis, mean (SD), y 65.8 (9.5) 66.8 (7.9) 67.9 (11.3) NA
Sex
Female 15 (57.7) 771 (51.3) 32 242 (48.1)
NA
Male 11 (42.3) 733 (48.7) 34 745 (51.9)
Year of diagnosis
1998-2010 6 (23.1) 325 (21.6) 23 929 (35.7)
2011-2015 7 (26.9) 378 (25.1) 22 985 (34.3) NA
2016-2021b 13 (50.0) 801 (53.3) 20 073 (30.0)
Tumor location
Head 12 (50.0) 767 (51.0) 35 117 (52.4)
Body or tail 12 (50.0) 448 (29.8) 18 691 (27.9) .002
Other 0 289 (19.2) 13 179 (19.7)
Missing 2 0 0
Type of surgery
Whipple 9 (50.0) 243 (63.0) 9713 (60.9)
Distal 8 (44.4) 108 (28.0) 4493 (28.2) .03
Total pancreatectomy 1 (5.6) 35 (9.1) 1754 (11.0)
Missingc 8 1118 51 027
Tumor grade
1 (Well differentiated) 1 (5.3) 74 (14.5) 4012 (17.0)
2 (Moderately differentiated) 15 (78.9) 247 (48.5) 9941 (42.0)
.03
3 or 4 (Poorly differentiated, 3 (15.8) 188 (36.9) 9699 (41.0)
undifferentiated, or anaplastic)
Missing 7 995 43 335

(continued)

(familial pancreatic cancer kindred), or germline gene status We used the MatchIt package in R, version 4.2.3
criteria (carriers of a germline deleterious variant in a gene as- (R Project for Statistical Computing) to select SEER control pa-
sociated with hereditary pancreatic cancer [ATM, BRCA1, tients with PDAC who matched exactly on age, sex, and year
BRCA2, CDKN2A, PALB2, or STK11]). A total of 1731 high-risk of diagnosis to high-risk individuals with PDAC. All the high-
individuals were enrolled in the CAPS studies between 1998 risk individuals with PDAC were White, so control patients were
and June 2021.5,10,14,15,18 The current study includes the 26 matched on White race. Participants self-reported race and eth-
high-risk individuals diagnosed with PDAC during this pe- nicity information on the baseline survey. The matching ap-
riod (hereafter, high-risk individuals with PDAC) (eFigure 1, proach identified all unique combinations of covariates to form
eTable 1 in Supplement 1). A more detailed description of this subclasses, and only subclasses with both high-risk individu-
study cohort is available in Dbouk et al.19 als with PDAC and SEER control patients with PDAC were in-
When an high-risk individual developed a suspected cluded in the final cohort.
pancreatic neoplasm (solid mass, cyst with worrisome fea-
tures, or lesion with suspicious or positive cytology), surgi- Statistical Analysis
cal treatment was performed after multidisciplinary discus- Data were collected from 1998 through 2021. The data analysis
sion and shared decision-making with the participant. was performed from April 29, 2022, through April 10, 2023. The
Surgery was performed by expert pancreaticobiliary sur- primary analyses were comparisons of American Joint
geons, and final diagnoses were made by an expert patholo- Committee on Cancer PDAC tumor stage at diagnosis (using the
gist at each site. 6th, 7th, and 8th editions of the American Joint Committee on
Cancer Staging Manual, depending on year of diagnosis), OS, and
Matched Control Patients With PDAC PDAC mortality between high-risk individuals with PDAC and
The high-risk individuals with PDAC were matched to control SEER control patients. Clinical and demographic characteris-
patients with PDAC selected from the Sur veillance, tics, including stage at diagnosis, were summarized using de-
Epidemiology, and End Results (SEER) 18 registry grouping scriptive statistics and tested for differences between cohorts
(November 2020 submission). The SEER data are considered using conditional logistic regression models with matching
nonhuman participant research. Data are anonymized, and no weights and clustered on the covariate subclasses. Overall
patient contact is involved in their collection. survival was calculated as the time from diagnosis to last fol-

jamaoncology.com (Reprinted) JAMA Oncology August 2024 Volume 10, Number 8 1089

Downloaded from jamanetwork.com by University of Gent / UZGent Kenniscentrum user on 10/10/2024

 Research Original Investigation Pancreatic Cancer Surveillance and Survival

Table 2. Characteristics of High-Risk Individuals With PDAC, Matched SEER Control Patients With PDAC,
and the Full SEER Cohort Eligible for Matching (continued)

No. (%)
High-risk Matched SEER control
individuals patients with PDAC Eligible SEER cohort
Characteristic (n = 26) (n = 1504) (n = 66 987) P valuea
Tumor size, median (range), cm 2.5 (0.6-5.0) 3.6 (0.2-8.0) 3.6 (0.0-8.0)d <.001
T stage
1 6 (27.3) 57 (4.4) 2977 (5.2)
2 10 (45.5) 425 (32.8) 16 563 (29.0)
<.001
3 5 (22.7) 530 (40.9) 25 144 (44.0)
4 1 (4.5) 283 (21.9) 12 484 (21.8)
Missing 4 209 9819
N stage
0 12 (57.1) 771 (61.9) 35 328 (61.5)
1 9 (42.9) 446 (35.8) 21 566 (37.5) .56
2 0 29 (2.3) 590 (1.0)
Missing 5 258 9503
No. of positive nodes per total 0.06 (0.10) 0.15 (0.19) 0.15 (0.20) .004
nodes examined, mean (SD)
M stage
0 19 (73.1) 695 (46.2) 32 665 (48.8)
.01
1 7 (26.9) 809 (53.8) 34 322 (51.2)
AJCC stage
I 10 (38.5) 155 (10.3) 7508 (11.2)
II 8 (30.8) 377 (25.1) 18 056 (27.0)
<.001
III 1 (3.8) 163 (10.8) 7101 (10.6)
IV 7 (26.9) 809 (53.8) 34 322 (51.2)
AJCC stage
Ia 6 (23.1) 37 (2.5) 2187 (3.3)
Ib 4 (15.4) 118 (7.8) 5321 (7.9)
IIa 2 (7.7) 154 (10.2) 7622 (11.4)
<.001
IIb 6 (23.1) 223 (14.8) 10 434 (15.6)
III 1 (3.8) 163 (10.8) 7101 (10.6)
IV 7 (26.9) 809 (53.8) 34 322 (51.2)
Summary stage
Localized 12 (46.2) 309 (20.5) 15 130 (22.6)
Regional 7 (26.9) 386 (25.7) 17 535 (26.2) .01
Distant 7 (26.9) 809 (53.8) 34 322 (51.2)
Received neoadjuvant
chemotherapy
Yes 3 (11.5) 985 (65.5) 39 998 (59.7)
No, unknown, no surgery 23 (88.5) 519 (34.5) 26 989 (40.3) <.001
performed

Abbreviations: AJCC, American Joint Committee on Cancer; NA, not applicable; PDAC, pancreatic ductal adenocarcinoma;
SEER, Surveillance, Epidemiology, and End Results.
a
P values for differences between high-risk individuals with PDAC and matched SEER control patients with PDAC
estimated from conditional logistic regression models. P values were not computed for age, sex, and year of diagnosis, as
they were the variables included in the matching algorithm.
b
SEER data were available up through 2019. Each high-risk individual with PDAC diagnosed in 2020 or 2021 was matched
to a patient in SEER diagnosed in 2019.
c
The 8 high-risk individuals with PDAC did not undergo surgery. In the SEER cohorts, either patients did not undergo
surgery or information about surgery was not available.
d
Tumor sizes of 0 indicate that no mass or tumor was found, for example, when a tumor of a stated primary site is not
found, but the tumor has metastasized.

1090 JAMA Oncology August 2024 Volume 10, Number 8 (Reprinted) jamaoncology.com

Downloaded from jamanetwork.com by University of Gent / UZGent Kenniscentrum user on 10/10/2024

 Pancreatic Cancer Surveillance and Survival Original Investigation Research

low-up or death and estimated using the Kaplan-Meier method.

Figure 1. Tumor Stage at Diagnosis
The OS was compared between groups using a Cox propor-
tional hazards regression model that included model weights and 60
High-risk individuals with PDAC
a cluster-robust standard error estimation accounting for the
Matched SEER control patients
matching. Mortality specifically from PDAC was compared be- 50
Eligible SEER control patients
tween high-risk individuals with PDAC and SEER control pa-
40
tients using competing-risks regression for clustered data, with

Patients, %
death from other causes considered a competing event (pack-
30
age crrSC in R, version 4.2.3). A 2-sided P < .05 was considered
significant for all analyses. 20

Sensitivity Analyses 10
Three sets of sensitivity analyses for the primary analyses were
conducted after sequentially excluding specific subgroups of 0
I II III IV
high-risk individuals with PDAC (eFigure 1 in Supplement 1). Tumor stage
These analyses included (1) rematching on primary tumor lo-
cation after excluding 2 high-risk individuals with PDAC who Frequency distribution of tumor stage at diagnosis for 26 high-risk individuals
with pancreatic ductal adenocarcinoma (PDAC); 1504 matched Surveillance,
had dropped out of surveillance and presented at an outside
Epidemiology, and End Results (SEER) control patients with PDAC; and the pool
institution with metastatic disease but for whom primary tu- of 66 987 eligible SEER patients with PDAC from which the matched case
mor location within the pancreas was unknown, (2) exclud- patients were drawn. P < .001 for high-risk individuals with PDAC vs matched
ing 4 more high-risk individuals who presented with PDAC af- SEER control patients.
ter having dropped out of surveillance (over approximately 3
months beyond the recommended annual surveillance), and thus, the type of surgery performed compared with matched
(3) limiting analyses to the subgroup of 18 high-risk individu- SEER control patients (Table 2). The high-risk individuals with
als with screen-detected PDAC without metastatic disease who PDAC had fewer well-differentiated cancers compared with
underwent pancreatic resection. matched SEER control patients (1 of 19 [5.3%] vs 74 of 509
[14.5%]; P = .03). Median tumor size was significantly smaller
Adjustment for Lead-Time Bias in high-risk individuals with PDAC compared with matched
To adjust for potential lead-time bias in the high-risk individu- SEER control patients (2.5 [range, 0.6-5.0] vs 3.6 [range, 0.2-
als with PDAC, we included an additional sensitivity analysis 8.0] cm, respectively; P < .001).
using the method described in Duffy et al.20 We considered
mean sojourn times of 3, 6, and 12 months and reestimated Stage Shift Analysis
survival probabilities and hazard ratios (HRs) using the ad- There was a significant shift in the American Joint Commit-
justed survival times. Further details about the SEER cohort tee on Cancer tumor stage of high-risk individuals with PDAC
and the sensitivity analysis are provided in the eMethods in toward lower disease stages compared with the matched SEER
Supplement 1. control patients (stage I, 10 of 26 [38.5%] vs 155 of 1504 [10.3%];
stage II, 8 of 26 [30.8%] vs 377 of 1504 [25.1%]; P < .001)
(Figure 1). Similarly, T stage for high-risk individuals with PDAC
was lower compared with matched SEER control patients, with
Results 6 of 22 (27.3%) vs 57 of 1295 (4.4%) with T1 stage, respectively
Baseline Characteristics (Table 2). Furthermore, fewer high-risk individuals with PDAC
The characteristics of 26 high-risk individuals with PDAC (mean had distant metastases at diagnosis (M1 stage) compared with
[SD] age at diagnosis, 65.8 [9.5] years; 15 female [57.7%] and matched SEER control patients (7 of 26 [26.9%] vs 809 of 1504
11 male [42.3%]; all White race) are detailed in Table 1. Twenty [53.8%]; P = .01).
of the 26 (76.9%) high-risk individuals with PDAC were diag-
nosed after a median of 5.8 years (range, 1.1-11.2 years) from Survival Comparison, Sensitivity Analyses,
study entry, while 6 (23.1%) were diagnosed at baseline and Adjustment for Lead-Time Bias
(Table 1). Nearly one-half (12 of 26 [46.2%]) of the PDACs in Overall survival was higher in the 26 high-risk individuals with
high-risk individuals were located in the pancreatic head PDAC (median, 61.7 months; range, 1.9-147.3 months) com-
(Table 2). Of the high-risk individuals with PDAC, 3 (11.5%) un- pared with matched SEER controls (median, 8.0 months; range,
derwent neoadjuvant therapy. Patient and tumor character- 1.0-131.0 months; HR, 4.19; 95% CI, 2.32-7.56; P < .001) (Figure 2).
istics of both the matched cohort of 1504 White patients (mean One- and 5-year survival probabilities were 84% (95% CI, 70%-
[SD] age at diagnosis, 66.8 [7.9] years; 771 female [51.3%] and 100%) and 50% (95% CI, 32%-80%), respectively, for high-risk
733 male [48.7%]) and entire eligible SEER PDAC cohort of individuals with PDAC and 38% (95% CI, 36%-41%) and 9% (95%
66 987 individuals (mean [SD] age at diagnosis, 67.9 [11.3] years; CI, 7%-11%), respectively, for SEER control patients (Table 3).
32 242 female [48.1%] and 34 745 male [51.9%]) are included Moreover, median OS in the 20 high-risk individuals whose PDAC
for comparison (Table 2). There were small differences in the was screen detected was 144 months (range, 3.6-147.3 months),
distribution of the primary tumor within the pancreas and, while matched SEER control patients had a median OS of 9

jamaoncology.com (Reprinted) JAMA Oncology August 2024 Volume 10, Number 8 1091

Downloaded from jamanetwork.com by University of Gent / UZGent Kenniscentrum user on 10/10/2024

 Research Original Investigation Pancreatic Cancer Surveillance and Survival

Figure 2. Overall Survival for Pancreatic Ductal Adenocarcinoma (PDAC) in High-Risk Individuals and Matched
Surveillance, Epidemiology, and End Results (SEER) Control Patients

100
HR, 4.19; 95% CI, 2.32-7.56

80

Overall survival, %
60

High-risk individuals with PDAC

40

20
Matched SEER control patients

0
0 1 2 3 4 5 6 7 8 9 10 11 12 13
Time since diagnosis, y
No. at risk
High-risk individuals with PDAC 26 19 15 11 9 7 5 5 2 2 2 2 1 0
Matched SEER control patients 1504 472 210 97 41 36 19 14 10 10 7 0 0 0
HR indicates hazard ratio.

months (range, 1.0-123.0 months). One- and 5-year survival prob- Screening of average-risk individuals for PDAC is not rec-
abilities were 95% (95% CI, 86%-100%) and 61% (95% CI, 40%- ommended by the US Preventive Services Task Force.4 Be-
93%), respectively, in the high-risk individuals with screen- cause PDAC is a rare, albeit deadly disease, a randomized clini-
detected PDAC and 41% (95% CI, 36%-46%) and 9% (95% CI, 5%- cal trial comparing surveillance vs no surveillance in high-risk
14%), respectively, in matched SEER control patients (eTable 3 individuals is not feasible. Selective surveillance of high-risk
in Supplement 1). Similar results were observed when match- individuals has been 1 proposed alternative to population-
ing was limited to the high-risk individuals with PDAC with based screening in the past 2 decades. Using EUS, MRI, and/or
known primary tumor location (eTable 1 in Supplement 1) and CT, 25 surveillance studies involving more than 3000 high-risk
for the 18 surgically treated high-risk individuals with screen- individuals showed that early detection of asymptomatic PDAC
detected PDAC (eTable 2, eFigure 2 in Supplement 1). Addi- is feasible.23 In a 2019 clinical practice update, the American
tional survival analysis and results are described in the eMethods Gastroenterological Association endorsed selective surveil-
in Supplement 1. lance in high-risk individuals.24 In a 2022 clinical guideline, the
American Society for Gastrointestinal Endoscopy recom-
Tumor-Specific Mortality mended selective surveillance of high-risk individuals based on
The cumulative probability of death from PDAC at 1 and 5 years scientific review and a Grading of Recommendations Assess-
was 16% and 43% for high-risk individuals with PDAC and 59% ment, Development, and Evaluation methodology.23,25
and 86% for matched SEER control patients, respectively (HR, The findings of our cohort study show that high-risk
3.58; 95% CI, 2.01-6.39; P < .001). For the surgically treated individuals who underwent annual or semiannual surveil-
high-risk individuals with screen-detected PDAC, the corre- lance EUS and MRI and who were diagnosed with PDAC
sponding cumulative probabilities were 6% and 29% at 1 and had a greater likelihood of having smaller, lower-stage
5 years, respectively. tumors; lower PDAC-specific mortality; and better OS
(5-year rate, 50%) compared with a national cohort (5-year
rate, 9%) matched for age, sex, race, and year of diagnosis
and with similar tumor location and type of surgery. The
Discussion adjusted HRs showed a 4-fold higher chance of being alive
Incidence rates for PDAC over the past decade have contin- for screened high-risk individuals.
ued to increase, with minimal improvement in survival world- One study reported improved outcomes for surveillance-
wide despite advances in health care.1,4 Unlike common can- detected PDAC in a cohort with the Dutch CDKN2A/p16 founder
cers, such as breast, colorectal, and lung cancer, screening for variant compared with matched patients with PDAC from the
early detection and prevention of pancreatic cancer remains Dutch National Cancer Registry after adjusting for lead-time
a challenge because it is a relatively rare disease that can be bias.26 In this Dutch study, surveillance led to a shift to a lower
difficult to diagnose, even when symptomatic.12 The overall stage (38.7% stage I) and higher 5-year survival rate (32.4%).
and cancer-specific 5-year survival for patients with PDAC are Similarly, in the current study, we also found that the im-
11% and 13%, respectively.21 However, patients with resect- proved effects of early detection and treatment of PDAC in high-
able stage Ia PDAC have a 5-year survival rate of approxi- risk individuals undergoing surveillance may be due to lead-
mately 84%,22 which underscores the potential for improved time bias since disease-specific mortality was also decreased
outcomes through early detection. and sensitivity analyses showed similar findings. We also re-

1092 JAMA Oncology August 2024 Volume 10, Number 8 (Reprinted) jamaoncology.com

Downloaded from jamanetwork.com by University of Gent / UZGent Kenniscentrum user on 10/10/2024

 Pancreatic Cancer Surveillance and Survival Original Investigation Research

Table 3. Comparison of OS in High-Risk Individuals With PDAC and Matched SEER Control Patients
With Additional Sensitivity Analysis Accounting for Potential Lead-Time Bias

Lead-time bias
Survival probability
(range), % Observed 3 mo 6 mo 12 mo
OS, median HR HR HR
Group (range), mo 1y 5y (95% CI)a P value HR (95% CI)a P value (95% CI)a P value (95% CI)a P value
High-risk 61.7 84 50 1 1 1 1 [Reference]
individuals (1.9-147.3) (70-100) (32-80) [Reference] [Reference] [Reference]
with PDAC
(n = 26)
Matched 8.0 38 9 (7-11) 4.19 <.001 3.91 <.001 3.69 <.001 3.34 <.001
SEER control (1.0-131.0) (36-41) (2.32-7.56) (2.12-7.21) (1.97-6.91) (1.74-6.40)
patients
with PDAC
(n = 1504)
Abbreviations: HR, hazard ratio; OS, overall survival; PDAC, pancreatic ductal cluster-robust standard errors clustered on the matched subclass and
adenocarcinoma; SEER, Surveillance, Epidemiology, and End Results. including subclass-specific weights. The high-risk individuals with PDAC were
a
The HRs from Cox proportional hazards models were estimated using matched to SEER control patients on age, sex, and year of diagnosis.

port a reduction in PDAC mortality that supports selective sur- cancers detected in surveillance programs of high-risk indi-
veillance for a rare but deadly cancer, which compares favor- viduals highlight the continuing need to develop more sensi-
ably with screening for more common cancers arising from the tive diagnostic tools and standardize or modify surveillance
breast27 or colon.28-30 methods.
Survival after a PDAC diagnosis among high-risk individu- Another potential reason for variability in the survival ben-
als undergoing surveillance in the CAPS program is higher than efit from surveillance of high-risk individuals is variability in how
reported in other established programs.7,11,26 Although PDAC patients with pancreatic cancer are treated. For example, some
surveillance programs7,10,11 have shown a greater proportion patients in the Dutch cohort with resectable PDAC declined sur-
of stage I cancers (33%-40% of cases) and improved resect- gery, and the more recently diagnosed high-risk individuals in
ability (60%-83%) for all patients with screen-detected PDAC our cohort were more likely to have undergone neoadjuvant
compared with patients with PDAC diagnosed in the general therapy (11.5% in our cohort vs 0% in the Rotterdam cohort32 and
population or outside of surveillance, the reported survival 11% in the CDKN2A cohort). Targeted chemotherapy with poly-
benefit is highly variable, with median survival ranging from ADP ribose polymerase inhibitors and platinum-based agents for
18 months in the other studies to 61.7 months in our study. high-risk individuals with germline BRCA/PALB2 variants can
What accounts for these differences in survival among potentially improve progression-free survival.34,35 Finally, dif-
high-risk surveillance cohorts is an important unanswered ferences in tumor biology may contribute to the wide range of
question. One of the main differences between the 2 cohorts reported PDAC survival. Germline gene status may affect the
is the number of stage I cases, which may be the result of dif- timeline from preclinical invasive cancer, to a detectable local-
ferences in the surveillance method used or in disease biol- ized tumor, to metastatic or locally advanced PDAC.
ogy. Regarding surveillance tests, EUS is known to have a higher While the current study provides encouraging results and
resolution compared with MRI and CT and better accuracy for underscores the potential of surveillance and early detection
detecting small solid pancreatic cancers than MRI, although to improve survival of patients with PDAC in a well-defined
the available data are limited.5,31,32 The Dutch surveillance pro- high-risk cohort, it is unclear whether surveillance outside
gram that reported PDAC outcomes of individuals with del- specialized centers can be recommended. The benefits of sur-
eterious germline CDKN2A variants primarily relied on MRI for veillance have to be weighed against potential harms. Al-
surveillance, with optional EUS.11,33 Seventeen percent of though morbidity of pancreatic surgery may be substantial,
incident PDACs in the Dutch cohort presented as interval morbidity after pancreatic resection of surveillance-detected
cancers.11 Furthermore, a retrospective review of imaging lesions in high-risk individuals in high-volume surgical cen-
abnormalities in 28 of 2552 high-risk individuals who devel- ters (19.9%) 25 has been lower or comparable to known
oped PDAC while participating in 1 of 16 surveillance morbidity rates after PDAC resection (30%-60%),36 with no
programs from 7 countries (Australia, Germany, Israel, Italy, operative mortality to date.25,37 However, screening for a low-
the Netherlands, Spain, and the US) found that almost one- prevalence disease may lead to false-positive results.
half of the high-grade dysplasia or PDAC lesions were de- The diagnostic yield (finding PDAC or high-grade dysplasia at
tected before the next scheduled surveillance visit, at a me- resection) of surveillance-detected lesions is low (29% of those
dian of approximately 11 months after an unremarkable imaging undergoing surgical resection, 3 8 or 8.6% of high-risk
examination.32 Differences between surveillance programs32 individuals23), despite expert multidisciplinary recommen-
with respect to how the imaging tests are used (magnetic reso- dation and shared decision-making. False-negative test re-
nance/magnetic resonance cholangiopancreatography, EUS, or sults are also still a problem, and radiologists frequently
both at each visit); surveillance intervals (6 or 12 months); and (50%-62%) miss pancreatic cancers presenting with atypical
other aspects of their radiologic, surgical, and oncologic care or subtle signs on CT or MRI,39-41 which may lead to delayed
may have contributed to the variability in outcomes. Interval diagnosis and advanced disease. In a retrospective review of

jamaoncology.com (Reprinted) JAMA Oncology August 2024 Volume 10, Number 8 1093

Downloaded from jamanetwork.com by University of Gent / UZGent Kenniscentrum user on 10/10/2024

 Research Original Investigation Pancreatic Cancer Surveillance and Survival

prior negative MRI findings for carriers of germline CDKN2A Strengths and Limitations
pathogenic variants who developed PDAC, mild pancreatic duc- Our study has several strengths, including multiple centers,
tal dilatation was an unrecognized subtle abnormality.42 None- a well-defined cohort of high-risk individuals and control
theless, nearly one-half of high-risk individuals who devel- patients, standardized surveillance methods combining EUS
oped neoplastic progression while under surveillance had no and MRI, a long-standing clinical and research surveillance
prior pancreatic lesions detected by imaging studies.7 In ad- program with multidisciplinary expert teams, and analytic
dition, the psychological burden of annual surveillance of high- methods that address the potential effects of bias. Our study
risk individuals should be considered, but it appears to be low,43 also has several limitations, including enrollment at aca-
and surveillance may be associated with decreased cancer demic referral centers and limited generalizability, limited
worry after initiating it. Finally, the cost of selective surveil- racial and ethnic diversity, a relatively small number of
lance of high-risk individuals with EUS and MRI are high, and high-risk individuals who progressed to PDAC, and lack of a
access to surveillance programs is currently limited. Four stud- control arm of unscreened high-risk individuals. The
ies consistently found that surveillance of high-risk individu- National Cancer Institute SEER database provides a wealth
als was cost-effective compared with no surveillance,44-47 but of national cancer data, but it has limitations that include
there is substantial variability in cost.23 incomplete neoadjuvant and adjuvant therapy data and
A multimodal approach to PDAC surveillance that estimates variations in data reporting, among others. Furthermore, it
individualized risk using patient demographics, clinical tests, is possible that patients who participate in pancreas surveil-
imaging, genomics,48,49 biomarkers, and artificial intelligence is lance may be otherwise healthier (with respect to diet, exer-
an as-yet unrealized “holy grail” for achieving better outcomes cise, and smoking) and so may have better clinical outcomes
with the highest benefit and least cost and harm. Serum biomark- than those in the SEER database.
ers using proteins or circulating tumor DNA have the potential to
improve surveillance.50-57 To improve on current pancreas sur-
veillance with EUS and MRI, these biomarkers must be low cost,
highly specific, and sufficiently sensitive to detect early-stage
Conclusions
PDAC. In the future, these noninvasive tests might be incorpo- In this comparative cohort study, surveillance of high-risk
rated into a surveillance strategy for high-risk individuals between individuals for PDAC using EUS and MRI within established
or at the time of scheduled MRI and/or EUS visits or widely ap- programs at academic centers was observed to lead to the
plied for screening average-risk individuals. Furthermore, arti- detection of smaller pancreatic cancers, a greater number of
ficial intelligence is gaining momentum in improving the sensi- patients with stage I disease, lower mortality, and a much
tivity and specificity of surveillance CT, MRI, and EUS.58,59 Finally, higher likelihood of long-term survival than unscreened
artificial intelligence can also help to estimate PDAC risk through patients in the general population diagnosed with PDAC.
electronic health records,60 refine the identification of high-risk These findings suggest that selective surveillance of indi-
individuals, and potentially enable a broader targeted surveillance viduals at high risk for pancreatic cancer may improve clini-
program in much larger populations for wider implementation. cal outcomes.

ARTICLE INFORMATION Pancreatic Cancer Research Center, Johns Hopkins Author Contributions: Ms Blackford had full access
Accepted for Publication: February 5, 2024. Medical Institutions, Baltimore, Maryland (He, to all of the data in the study and takes
Burkhart, Lennon); Division of Gastroenterology, responsibility for the integrity of the data and the
Published Online: July 3, 2024. Department of Medicine, Abramson Cancer Center, accuracy of the data analysis. Ms Blackford and Dr
doi:10.1001/jamaoncol.2024.1930 University of Pennsylvania Perelman School of Canto contributed equally as co–first authors.
Correction: This article was corrected on Medicine, Philadelphia (Katona); Division of Concept and design: Blackford, Canto, Hruban,
September 12, 2024, to add a missing middle initial Gastroenterology and Liver Disease, University Syngal, Farrell, Rustgi, Klein, Kamel,
to Elliot K. Fishman’s name in the byline. Hospitals Cleveland Medical Center, Case Western Burkhart, Goggins.
Open Access: This is an open access article Reserve University, Cleveland, Ohio (Chak); Division Acquisition, analysis, or interpretation of data:
distributed under the terms of the CC-BY License. of Gastroenterology, Hepatology and Nutrition, Blackford, Canto, Dbouk, Katona, Chak, Brand,
© 2024 Blackford AL et al. JAMA Oncology. University of Pittsburgh Medical Center, Syngal, Farrell, Kastrinos, Stoffel, Rustgi, Klein,
Pennsylvania (Brand); Cancer Genetics and Kamel, Fishman, He, Burkhart, Shin,
Author Affiliations: Department of Oncology, The Prevention, Medical Oncology, Dana-Farber Cancer Lennon, Goggins.
Sol Goldman Pancreatic Cancer Research Center, Institute, Boston, Massachusetts (Syngal); Division Drafting of the manuscript: Blackford, Canto,
Johns Hopkins Medical Institutions, Baltimore, of Gastroenterology, Brigham and Women’s Hruban, Farrell, Kamel, Fishman, Goggins.
Maryland (Blackford, Canto, Hruban, Klein, Lennon, Hospital and Harvard Medical School, Boston, Critical review of the manuscript for important
Goggins); Department of Pathology, The Sol Massachusetts (Syngal); Yale Center for Pancreatic intellectual content: Blackford, Canto, Dbouk,
Goldman Pancreatic Cancer Research Center, Johns Disease, Section of Digestive Disease, Katona, Chak, Brand, Syngal, Farrell, Kastrinos,
Hopkins Medical Institutions, Baltimore, Maryland Yale University, New Haven, Connecticut (Farrell); Stoffel, Rustgi, Klein, Kamel, Fishman, He, Burkhart,
(Dbouk, Hruban, Goggins); Department of Division of Digestive and Liver Diseases, Herbert Shin, Lennon, Goggins.
Medicine (Gastroenterology), The Sol Goldman Irving Comprehensive Cancer Center, Vagelos Statistical analysis: Blackford, Canto, Dbouk, Farrell,
Pancreatic Cancer Research Center, Johns Hopkins College of Physicians and Surgeons, Columbia Klein, Kamel.
Medical Institutions, Baltimore, Maryland (Canto, University Irving Medical Center, New York, Obtained funding: Canto, Syngal, Rustgi, Klein,
Shin, Lennon, Goggins); Department of Radiology, New York (Kastrinos, Rustgi); Division of Burkhart, Goggins.
The Sol Goldman Pancreatic Cancer Research Gastroenterology and Hepatology, Department of Administrative, technical, or material support:
Center, Johns Hopkins Medical Institutions, Internal Medicine, University of Michigan, Canto, Syngal, Rustgi, Klein, Kamel, Fishman, He,
Baltimore, Maryland (Kamel, Fishman, Lennon); Ann Arbor (Stoffel). Burkhart, Shin, Lennon, Goggins.
Department of Surgery, The Sol Goldman Supervision: Blackford, Canto, Syngal, Goggins.

1094 JAMA Oncology August 2024 Volume 10, Number 8 (Reprinted) jamaoncology.com

Downloaded from jamanetwork.com by University of Gent / UZGent Kenniscentrum user on 10/10/2024

 Pancreatic Cancer Surveillance and Survival Original Investigation Research

Conflict of Interest Disclosures: Dr Canto reported Gastroenterology. 2012;142(4):796-804. doi:10. Consortium. Gut. 2020;69(1):7-17. doi:10.1136/
receiving royalties from UpToDate and consultant 1053/j.gastro.2012.01.005 gutjnl-2019-319352
fees from BlueStar Genomics and Cernostics (Castle 6. Corral JE, Mareth KF, Riegert-Johnson DL, Das A, 18. Canto MI, Almario JA, Schulick RD, et al. Risk of
Biosciences) outside the submitted work. Dr Brand Wallace MB. Diagnostic yield from screening neoplastic progression in individuals at high risk for
reported receiving grants from the National Cancer asymptomatic individuals at high risk for pancreatic pancreatic cancer undergoing long-term
Institute during the conduct of the study and grants cancer: a meta-analysis of cohort studies. Clin surveillance. Gastroenterology. 2018;155(3):740-
from Immunovia and Freenome outside the Gastroenterol Hepatol. 2019;17(1):41-53. doi:10. 751.e2. doi:10.1053/j.gastro.2018.05.035
submitted work. Dr Syngal reported receiving 1016/j.cgh.2018.04.065
grants from Biological Dynamics, personal fees 19. Dbouk M, Brewer Gutierrez OI, Lennon AM,
from GlaxoSmithKline and Natera, and 7. Overbeek KA, Goggins MG, Dbouk M, et al; et al. Guidelines on management of pancreatic
noncompensated service on a scientific advisory International Cancer of the Pancreas Screening cysts detected in high-risk individuals: an
board for the PanCan Foundation outside the Consortium. Timeline of development of pancreatic evaluation of the 2017 Fukuoka guidelines and the
submitted work and a patent for the Prediction cancer and implications for successful early 2020 International Cancer of the Pancreas
Model for Gene Mutations, with royalties paid from detection in high-risk individuals. Gastroenterology. Screening (CAPS) consortium statements.
Myriad Genetics. Dr Klein reported receiving grants 2022;162(3):772-785.e4. doi:10.1053/j.gastro.2021.10. Pancreatology. 2021;21(3):613-621. doi:10.1016/j.pan.
from the National Cancer Institute during the 014 2021.01.017
conduct of the study. Dr Fishman reported 8. Signoretti M, Bruno MJ, Zerboni G, Poley JW, 20. Duffy SW, Nagtegaal ID, Wallis M, et al.
receiving grants from the Lustgarten Foundation, Delle Fave G, Capurso G. Results of surveillance in Correcting for lead time and length bias in
nonfinancial support from Microsoft, and personal individuals at high-risk of pancreatic cancer: estimating the effect of screen detection on cancer
fees from Exact Sciences and Imaging Insights a systematic review and meta-analysis. United survival. Am J Epidemiol. 2008;168(1):98-104. doi:
outside the submitted work. Dr Shin reported European Gastroenterol J. 2018;6(4):489-499. doi: 10.1093/aje/kwn120
receiving consultant fees from Boston Scientific 10.1177/2050640617752182 21. Hur C, Tramontano AC, Dowling EC, et al. Early
outside the submitted work. Dr Lennon reported 9. Bartsch DK, Slater EP, Carrato A, et al. pancreatic ductal adenocarcinoma survival is
receiving grants from the National Institute of Refinement of screening for familial pancreatic dependent on size: positive implications for future
Health and donations from the Intraductal Papillary cancer. Gut. 2016;65(8):1314-1321. doi:10.1136/ targeted screening. Pancreas. 2016;45(7):1062-1066.
Mucinous Neoplasms Global Research Foundation gutjnl-2015-311098 doi:10.1097/MPA.0000000000000587
during the conduct of the study and consultant fees
from Exact Science outside the submitted work. No 10. Dbouk M, Katona BW, Brand RE, et al. The 22. Blackford AL, Canto MI, Klein AP, Hruban RH,
other disclosures were reported. Multicenter Cancer of Pancreas Screening Study: Goggins M. Recent trends in the incidence and
impact on stage and survival. J Clin Oncol. 2022;40 survival of stage 1A pancreatic cancer:
Funding/Support: This work was supported by (28):3257-3266. doi:10.1200/JCO.22.00298 a Surveillance, Epidemiology, and End Results
grants U01210170, R01CA176828, and CA62924 analysis. J Natl Cancer Inst. 2020;112(11):1162-1169.
from the National Institutes of Health; Susan 11. Klatte DCF, Boekestijn B, Wasser MNJM, et al.
Pancreatic cancer surveillance in carriers of a doi:10.1093/jnci/djaa004
Wojcicki and Dennis Troper; Pancreatic Cancer
Interception Translational Cancer Research Grant germline CDKN2A pathogenic variant: yield and 23. Calderwood AH, Sawhney MS, Thosani NC,
SU2C-AACR-DT25-17 from the Stand Up to outcomes of a 20-year prospective follow-up. J Clin et al. American Society for Gastrointestinal
Cancer-Lustgarten Foundation (a program of the Oncol. 2022;40(28):3267-3277. doi:10.1200/JCO. Endoscopy guideline on screening for pancreatic
Entertainment Industry Foundation administered 22.00194 cancer in individuals with genetic susceptibility:
by the American Association for Cancer Research); 12. Mazer BL, Lee JW, Roberts NJ, et al. Screening methodology and review of evidence. Gastrointest
the Pancreatic Cancer Action Network; the Rolfe for pancreatic cancer has the potential to save lives, Endosc. 2022;95(5):827-854.e3. doi:10.1016/j.gie.
Pancreatic Cancer Foundation; the V Foundation; but is it practical? Expert Rev Gastroenterol Hepatol. 2021.12.002
the Hooven Memorial Fund; and the Victor Family 2023;17(6):555-574. doi:10.1080/17474124.2023. 24. Aslanian HR, Lee JH, Canto MI. AGA clinical
Pancreatic Cancer Fund. 2217354 practice update on pancreas cancer screening in
Role of the Funder/Sponsor: The funders had no 13. The Cancer of the Pancreas Screening-5 high-risk individuals: expert review. Gastroenterology.
role in the design and conduct of the study; (CAPS5) Study. ClinicalTrials.gov identifier: 2020;159(1):358-362. doi:10.1053/j.gastro.2020.03.
collection, management, analysis, and NCT02000089. Updated March 18, 2024. 088
interpretation of the data; preparation, review, or Accessed May 17, 2024. https://clinicaltrials.gov/ 25. Sawhney MS, Calderwood AH, Thosani NC,
approval of the manuscript; and decision to submit study/NCT02000089 et al; ASGE Standards of Practice Committee. ASGE
the manuscript for publication. 14. Canto MI, Goggins M, Yeo CJ, et al. Screening guideline on screening for pancreatic cancer in
Data Sharing Statement: See Supplement 2. for pancreatic neoplasia in high-risk individuals: an individuals with genetic susceptibility: summary
EUS-based approach. Clin Gastroenterol Hepatol. and recommendations. Gastrointest Endosc. 2022;
REFERENCES 2004;2(7):606-621. doi:10.1016/S1542-3565(04) 95(5):817-826. doi:10.1016/j.gie.2021.12.001

1. Huang J, Lok V, Ngai CH, et al. Worldwide burden 00244-7 26. Klatte DCF, Boekestijn B, Onnekink AM, et al;
of, risk factors for, and trends in pancreatic cancer. 15. Canto MI, Goggins M, Hruban RH, et al. Dutch Pancreatic Cancer Group. Surveillance for
Gastroenterology. 2021;160(3):744-754. doi:10.1053/ Screening for early pancreatic neoplasia in high-risk pancreatic cancer in high-risk individuals leads to
j.gastro.2020.10.007 individuals: a prospective controlled study. Clin improved outcomes: a propensity score-matched
Gastroenterol Hepatol. 2006;4(6):766-781. doi:10. analysis. Gastroenterology. 2023;164(7):1223-1231.e4.
2. Siegel RL, Miller KD, Wagle NS, Jemal A. Cancer doi:10.1053/j.gastro.2023.02.032
statistics, 2023. CA Cancer J Clin. 2023;73(1):17-48. 1016/j.cgh.2006.02.005
doi:10.3322/caac.21763 16. Canto MI, Harinck F, Hruban RH, et al; 27. Hendrick RE, Helvie MA. Mammography
International Cancer of Pancreas Screening (CAPS) screening: a new estimate of number needed to
3. Cancer stat facts: pancreatic cancer. National screen to prevent one breast cancer death. AJR Am
Cancer Institute. Accessed June 3, 2024. https:// Consortium. International Cancer of the Pancreas
Screening (CAPS) Consortium summit on the J Roentgenol. 2012;198(3):723-728. doi:10.2214/
seer.cancer.gov/statfacts/html/pancreas.html AJR.11.7146
management of patients with increased risk for
4. Owens DK, Davidson KW, Krist AH, et al; US familial pancreatic cancer. Gut. 2013;62(3):339-347. 28. Rembold CM. Number needed to screen:
Preventive Services Task Force. Screening for doi:10.1136/gutjnl-2012-303108 development of a statistic for disease screening. BMJ.
pancreatic cancer: US Preventive Services Task 1998;317(7154):307-312. doi:10.1136/bmj.317.7154.
Force reaffirmation recommendation statement. 17. Goggins M, Overbeek KA, Brand R, et al;
International Cancer of the Pancreas Screening 307
JAMA. 2019;322(5):438-444. doi:10.1001/jama.2019.
10232 (CAPS) Consortium. Management of patients with 29. Burke CA. Number needed to screen to detect
increased risk for familial pancreatic cancer: adenomas, advanced adenomas and colorectal
5. Canto MI, Hruban RH, Fishman EK, et al; updated recommendations from the International cancer is higher in women than in similarly aged
American Cancer of the Pancreas Screening (CAPS) Cancer of the Pancreas Screening (CAPS) men. Evid Based Med. 2012;17(5):159-160. doi:10.
Consortium. Frequent detection of pancreatic 1136/ebmed-2011-100440
lesions in asymptomatic high-risk individuals.

jamaoncology.com (Reprinted) JAMA Oncology August 2024 Volume 10, Number 8 1095

Downloaded from jamanetwork.com by University of Gent / UZGent Kenniscentrum user on 10/10/2024

 Research Original Investigation Pancreatic Cancer Surveillance and Survival

30. Rothe M. Eight years of colonoscopic bowel detect pancreatic cancer on prediagnostic thrombospondin-2 and CA19-9 blood markers. Sci
cancer screening in Germany: initial findings and computed tomography scans at a substantial lead Transl Med. 2017;9(398):eaah5583. doi:10.1126/
projections. “number needed to screen”. Dtsch time before clinical diagnosis. Gastroenterology. scitranslmed.aah5583
Arztebl Int. 2011;108(10):170. 2022;163(5):1435-1446.e3. doi:10.1053/j.gastro.2022. 51. Abe T, Koi C, Kohi S, et al. Gene variants that
31. Harinck F, Konings IC, Kluijt I, et al; Dutch 06.066 affect levels of circulating tumor markers increase
Research Group on Pancreatic Cancer Surveillance 41. Kang JD, Clarke SE, Costa AF. Factors associated identification of patients with pancreatic cancer.
in High-Risk Individuals. A multicentre comparative with missed and misinterpreted cases of pancreatic Clin Gastroenterol Hepatol. 2020;18(5):1161-1169.e5.
prospective blinded analysis of EUS and MRI for ductal adenocarcinoma. Eur Radiol. 2021;31(4): doi:10.1016/j.cgh.2019.10.036
screening of pancreatic cancer in high-risk 2422-2432. doi:10.1007/s00330-020-07307-5 52. Fahrmann JF, Schmidt CM, Mao X, et al.
individuals. Gut. 2016;65(9):1505-1513. doi:10.1136/ 42. Vasen HFA, Boekestijn B, Ibrahim IS, et al. Lead-time trajectory of CA19-9 as an anchor marker
gutjnl-2014-308008 Dilatation of the main pancreatic duct as first for pancreatic cancer early detection.
32. Overbeek KA, Levink IJM, Koopmann BDM, manifestation of small pancreatic ductal Gastroenterology. 2021;160(4):1373-1383.e6. doi:10.
et al; Dutch Familial Pancreatic Cancer Surveillance adenocarcinomas detected in a hereditary 1053/j.gastro.2020.11.052
Study Group. Long-term yield of pancreatic cancer pancreatic cancer surveillance program. HPB (Oxford). 53. Song CX, Yin S, Ma L, et al.
surveillance in high-risk individuals. Gut. 2022;71 2019;21(10):1371-1375. doi:10.1016/j.hpb.2019.02. 5-Hydroxymethylcytosine signatures in cell-free
(6):1152-1160. doi:10.1136/gutjnl-2020-323611 013 DNA provide information about tumor types and
33. Vasen H, Ibrahim I, Ponce CG, et al. Benefit of 43. Konings IC, Sidharta GN, Harinck F, et al. stages. Cell Res. 2017;27(10):1231-1242. doi:10.1038/
surveillance for pancreatic cancer in high-risk Repeated participation in pancreatic cancer cr.2017.106
individuals: outcome of long-term prospective surveillance by high-risk individuals imposes low 54. Cohen JD, Li L, Wang Y, et al. Detection and
follow-up studies from three European expert psychological burden. Psychooncology. 2016;25(8): localization of surgically resectable cancers with a
centers. J Clin Oncol. 2016;34(17):2010-2019. doi: 971-978. doi:10.1002/pon.4047 multi-analyte blood test. Science. 2018;359(6378):
10.1200/JCO.2015.64.0730 44. Corral JE, Das A, Bruno MJ, Wallace MB. 926-930. doi:10.1126/science.aar3247
34. Wattenberg MM, Asch D, Yu S, et al. Platinum Cost-effectiveness of pancreatic cancer surveillance 55. Lennon AM, Buchanan AH, Kinde I, et al.
response characteristics of patients with pancreatic in high-risk individuals: an economic analysis. Feasibility of blood testing combined with PET-CT
ductal adenocarcinoma and a germline BRCA1, Pancreas. 2019;48(4):526-536. doi:10.1097/MPA. to screen for cancer and guide intervention. Science.
BRCA2 or PALB2 mutation. Br J Cancer. 2020;122 0000000000001268 2020;369(6499):eabb9601. doi:10.1126/science.
(3):333-339. doi:10.1038/s41416-019-0582-7 45. Joergensen MT, Gerdes AM, Sorensen J, abb9601
35. Rosen MN, Goodwin RA, Vickers MM. BRCA Schaffalitzky de Muckadell O, Mortensen MB. Is 56. Liu MC, Oxnard GR, Klein EA, Swanton C,
mutated pancreatic cancer: a change is coming. screening for pancreatic cancer in high-risk groups Seiden MV; CCGA Consortium. Sensitive and
World J Gastroenterol. 2021;27(17):1943-1958. doi: cost-effective?—experience from a Danish national specific multi-cancer detection and localization
10.3748/wjg.v27.i17.1943 screening program. Pancreatology. 2016;16(4):584- using methylation signatures in cell-free DNA. Ann
36. Vollmer CM Jr, Sanchez N, Gondek S, et al; 592. doi:10.1016/j.pan.2016.03.013 Oncol. 2020;31(6):745-759. doi:10.1016/j.annonc.
Pancreatic Surgery Mortality Study Group. A 46. Kumar S, Saumoy M, Oh A, et al. Threshold 2020.02.011
root-cause analysis of mortality following major analysis of the cost-effectiveness of endoscopic 57. Klein EA, Richards D, Cohn A, et al. Clinical
pancreatectomy. J Gastrointest Surg. 2012;16(1):89- ultrasound in patients at high risk for pancreatic validation of a targeted methylation-based
102. doi:10.1007/s11605-011-1753-x ductal adenocarcinoma. Pancreas. 2021;50(6):807- multi-cancer early detection test using an
37. Canto MI, Kerdsirichairat T, Yeo CJ, et al. 814. doi:10.1097/MPA.0000000000001835 independent validation set. Ann Oncol. 2021;32(9):
Surgical outcomes after pancreatic resection of 47. Rulyak SJ, Kimmey MB, Veenstra DL, Brentnall 1167-1177. doi:10.1016/j.annonc.2021.05.806
screening-detected lesions in individuals at high TA. Cost-effectiveness of pancreatic cancer 58. Kuwahara T, Hara K, Mizuno N, et al. Artificial
risk for developing pancreatic cancer. J Gastrointest screening in familial pancreatic cancer kindreds. intelligence using deep learning analysis of
Surg. 2020;24(5):1101-1110. doi:10.1007/s11605- Gastrointest Endosc. 2003;57(1):23-29. doi:10. endoscopic ultrasonography images for the
019-04230-z 1067/mge.2003.28 differential diagnosis of pancreatic masses.
38. Paiella S, Salvia R, De Pastena M, et al. 48. Roberts NJ, Norris AL, Petersen GM, et al. Endoscopy. 2023;55(2):140-149. doi:10.1055/a-1873-
Screening/surveillance programs for pancreatic Whole genome sequencing defines the genetic 7920
cancer in familial high-risk individuals: a systematic heterogeneity of familial pancreatic cancer. Cancer 59. Marya NB, Powers PD, Chari ST, et al. Utilisation
review and proportion meta-analysis of screening Discov. 2016;6(2):166-175. doi:10.1158/2159- of artificial intelligence for the development of an
results. Pancreatology. 2018;18(4):420-428. doi:10. 8290.CD-15-0402 EUS-convolutional neural network model trained to
1016/j.pan.2018.04.002 49. Tan M, Brusgaard K, Gerdes AM, et al. Whole enhance the diagnosis of autoimmune pancreatitis.
39. Hoogenboom SA, Engels MML, Chuprin AV, genome sequencing identifies rare germline Gut. 2021;70(7):1335-1344. doi:10.1136/gutjnl-2020-
et al. Prevalence, features, and explanations of variants enriched in cancer related genes in first 322821
missed and misinterpreted pancreatic cancer on degree relatives of familial pancreatic cancer 60. Placido D, Yuan B, Hjaltelin JX, et al. A deep
imaging: a matched case-control study. Abdom patients. Clin Genet. 2021;100(5):551-562. doi:10. learning algorithm to predict risk of pancreatic
Radiol (NY). 2022;47(12):4160-4172. doi:10.1007/ 1111/cge.14038 cancer from disease trajectories. Nat Med. 2023;29
s00261-022-03671-6 50. Kim J, Bamlet WR, Oberg AL, et al. Detection of (5):1113-1122. doi:10.1038/s41591-023-02332-5
40. Mukherjee S, Patra A, Khasawneh H, et al. early pancreatic ductal adenocarcinoma with
Radiomics-based machine-learning models can

1096 JAMA Oncology August 2024 Volume 10, Number 8 (Reprinted) jamaoncology.com

Downloaded from jamanetwork.com by University of Gent / UZGent Kenniscentrum user on 10/10/2024