Racial, Ethnic, and Sex-based Disparities among High-risk Individuals Undergoing Pancreatic Cancer Surveillance

The incidence of pancreatic adenocarcinoma (PDAC) is increasing worldwide, and PDAC will soon become the second most deadly cancer in the United States (1, 2). Only a minority of PDACs present at an early, operable stage, and consequently the 5-year survival among all individuals with PDAC remains less than 15% (3). While the majority of PDAC is thought to be sporadic, up to 10% are due to familial risk (4), with germline pathogenic or likely pathogenic variants (PV) associated with hereditary breast and ovarian cancer (BRCA1, BRCA2, PALB2, ATM), Lynch syndrome (MLH1, MSH2, MSH6, PMS2, EPCAM), Peutz–Jeghers syndrome (STK11), familial atypical multiple mole melanoma syndrome (FAMMM; CDKN2A), and hereditary pancreatitis (PRSS1) being the most commonly identified PDAC-risk genes (5). Apart from known genetic risk, PDAC risk is also increased in familial pancreatic cancer (FPC), which can be defined as having two relatives with PDAC who are first-degree relatives of one another without a known genetic predisposition (6).

If PDAC is detected at early stages, survival is substantially improved; individuals with stage IA PDAC have a 5-year survival greater than 80% compared with a 5-year survival of less than 10% for individuals with stage III and IV PDAC (7). Therefore, early detection of PDAC is critical to improving long-term outcomes. While PDAC screening is not recommended for the general population (8), PDAC surveillance is recommended for individuals with a lifetime PDAC risk of 5% or greater, including those from FPC families as well as those from families with a genetic susceptibility (9–13). Multiple studies have reported on longstanding cohorts of high-risk individuals (HRI) undergoing PDAC surveillance, with some of these studies showing that surveillance can downstage PDAC diagnosis and improve PDAC survival (14–18).

While data on the effectiveness of PDAC surveillance in HRIs are promising, these study cohorts lack the racial, ethnic, and sex diversity that would be expected for the population served (14–18). This lack of diversity is important as there are clear disparities in PDAC outcomes based on race, ethnicity, and sex (19). Furthermore the lack of diversity is not specific to PDAC surveillance studies and has in fact been commonly recognized throughout clinical research (20, 21), with multiple efforts proposed to improve inclusion of under-represented populations (22). For PDAC surveillance, existing studies have recruited and followed patients for many years, during which time there might be limited access to genetic testing and/or cancer risk evaluation services that may have affected study enrollment and contributed to the observed disparities.

One of the newest and largest high-risk PDAC surveillance studies is the international, multicenter Pancreatic Cancer Early Detection (PRECEDE) Consortium, which began recruiting in 2020 and has enrolled more than 3,000 participants to date. Herein, we examine recruitment of HRIs to the PRECEDE Consortium since its inception two years ago to examine the characteristics of enrolled participants and highlight areas where improved recruitment is needed.

The PRECEDE Consortium (NCT04970056, https://precedestudy.org/) is a prospective, multicenter study focused on improving survival from PDAC through early detection (23). At the time of this writing, the PRECEDE Consortium had a total of 32 member institutions, with enrollment conducted after receiving approval from the Ethical and Independent Review Services Institutional Review Board (IRB; study number 20043–01B) or local IRB when required, and was conducted in accordance with the U.S. Common Rule. Starting in May of 2020, the PRECEDE Consortium began enrolling patients, where demographic information (including self-reported race/ethnicity and sex at birth), personal and family history, and pancreatic imaging data were collected and entered into a centralized study database maintained by the PRECEDE Data Coordinating Center at Arbor Research Collaborative for Health. In addition to consent for enrollment in PRECEDE, participants were separately asked whether they would provide specific consent for use of imaging data for research as well as for collection of DNA and non-DNA (i.e., serum, plasma) biosamples. Recruitment strategies were not standardized across PRECEDE sites, and there is significant variability in approaches to subject recruitment at each site. Local recruitment efforts were supplemented by a consortium-wide website (precedestudy.org) and clinicaltrials.gov posting (NCT04970056).

This analysis is focused on the cohort of individuals enrolled in PRECEDE with the highest risk of developing PDAC, for whom pancreatic surveillance is typically recommended based on published guidelines including individuals with one of the following (9–13):

• (i) Two or more relatives with PDAC on the same side of family where the 2 affected individuals are first-degree related to each other and at least 1 affected is first-degree related to the subject undergoing surveillance; age of 50 or older, or ≤ 10 years younger than earliest PDAC diagnosis in family.

• (ii) BRCA1, BRCA2, PALB2, ATM, MLH1, MSH2, MSH6, PMS2, or EPCAM PV and 1 first- or second-degree relative with PDAC; age of 50 or older, or ≤ 10 years younger than earliest PDAC diagnosis in family.

• (iii) FAMMM syndrome with a CDKN2A PV; age of 40 or older.

• (iv) Peutz–Jeghers syndrome (PJS) with a STK11 PV; age of 35 or older.

• (v) Hereditary pancreatitis (HP) with a PRSS1 PV and history of pancreatitis; age of 40 or older.

Data on HRIs enrolled between May of 2020 and March of 2022 were collected for this analysis. As entering enrollment data often lags behind date of enrollment, patients with incomplete enrollment data (N = 160) were excluded from additional analysis. Completed enrollment data were defined as reporting a HRI's enrollment criteria, enrollment site, age at time of consent, sex at birth, race, ethnicity, and whether or not a PV was identified. Participants categorized as having completed enrollment data may, however, still have incomplete data entry as there are multiple additional data points collected on participants.

Data are described using counts and percentages or median and range as appropriate. Comparisons between groups were made using χ² or Fisher exact test as appropriate or Kruskal–Wallis tests, respectively. As this is an exploratory analysis, P values were not adjusted for multiple comparisons. Analyses were conducted in SAS 9.4.

Raw data for this study were obtained from the PRECEDELink database, which is managed by Arbor Research Collaborative for Health. Derived data supporting the findings of this study are available from the corresponding author upon request.

During the study period, 1,273 HRIs were enrolled in PRECEDE. After excluding 160 individuals with incomplete enrollment data, 1,113 individuals remained for analysis. Of the 1,113 eligible individuals, 47.2% (n = 525) met familial pancreatic cancer criteria (FPC Group) and 45.4% (n = 505) had a PV in a PDAC-risk gene (BRCA1, BRCA2, PALB2, ATM, MLH1, MSH2, MSH6, PMS2, or EPCAM) in addition to a family history of PDAC (Gene + FHx Group; Table 1). FAMMM (5.7%, n = 63), PJS (1.6%, n = 18), and HP (0.2%, n = 2) made up smaller percentages of overall enrollment. While enrollment was initially low at study initiation in May 2020, from July 2020 through March 2022 enrollment across all centers was at least 30 patients per month in this defined group of HRIs, with similar enrollment of both the FPC and Gene + FHx groups (Fig. 1A and B).

Study participants were predominantly from the United States where the majority of PRECEDE centers are located (19 centers located in the United States enrolled HRIs for this study), accounting for 82.7% (N = 921) of eligible participants, followed by Canada, Italy, Israel, and Spain (Table 1). Among U.S. enrollment, New York (4 centers) and Pennsylvania (2 centers) were the two states with the largest number of HRIs enrolled (Fig. 2A), and there were five sites outside the United States that also enrolled HRIs for this study (Fig. 2B).

The most common age range at enrollment was 60–69 years (37.4%, n = 416) followed by 50–59 years (33%, n = 367; Table 1). The majority of participants were female (65.9%, n = 734) and self-reported non-Hispanic white (87.7%, n = 976), while only 2.9% (n = 32) reported Hispanic ethnicity. Results of germline genetic testing were available for the majority of participants (87.5%, n = 974). Of the 52.8% (n = 588) of individuals with a PV in a gene associated with PDAC, BRCA2 was the most common (37.9%, n = 223), followed by ATM (16.5%, n = 97) and BRCA1 (16.2%, n = 95). There were no significant differences in rate of PVs between individuals identified as white compared with those identified as non-white (Table 2); however, a lower percentage of males had a PV compared with females (Supplementary Table S1).

Of those participants where data were reported, 9.2% (n = 102) had diabetes mellitus, 31.7% (n = 345) had a history of smoking, and 72.3% (n = 725) had consumed alcohol in the 12 months prior to enrollment (Table 1). A history of cancer was present in 32.4% (n = 355) of participants with breast cancer being the most common among those with cancer (50.4%, n = 179). Almost all participants had a first-degree relative (94.9%, n = 1,029) or second-degree relative (SDR; 90.5%, n = 945) with any cancer. After PDAC, breast cancer was the next most common cancer in both FDRs (28.9%, n = 297) and SDRs (39.3%, n = 371).

As more than 90% of participants were enrolled in either the FPC or Gene + FHx groups, baseline characteristics were compared between these two groups (Table 3). There were geographic differences in enrollment with the United States, Israel, and Canada enrolling a higher percentage of participants in the Gene + FHx group, whereas Italy and Spain enrolled more in the FPC group. There were more males enrolled in the FPC group compared with the Gene + FHx group (40.6% vs. 24.9%, P < 0.001). There was also a higher percentage of individuals with a personal history of cancer in the Gene + FHx group compared with the FPC group (47.3% vs. 14.1%, P < 0.001). Further comparison was performed between baseline characteristics of United States sites versus those outside the United States, showing that white participants comprised a larger percentage of enrollees at United States sites compared with non-U.S. sites (95.2% vs. 67.2%, P < 0.001; Supplementary Table S2). However, there was no statistically significant difference in sex and only a small difference in Hispanic ethnicity (2.6% compared with 4.7%) between those enrolled in the United States versus outside the United States. Participants from the United States had higher rates of recent alcohol use (76.2% vs. 54.6%), but lower rates of current smoking (1.8% vs. 12.3%) compared with participants from outside the United States.

Almost all participants who enrolled in PRECEDE provided consent to utilize imaging data for research (99.6%, n = 1,108), collect DNA (97.7%, n = 1,087), and collect non-DNA biosamples including serum and plasma (99.5%, n = 1,107; Table 4). Consent rates were high among all groups even when stratifying by race/ethnicity (Table 4), sex (Supplementary Table S3), risk group (Supplementary Table S4), and age (Supplementary Table S5). Despite the persistence of the COVID-19 pandemic throughout study enrollment, DNA was successfully collected from 74% (n = 804) of consented participants, and serum and plasma were obtained from 69.7% (n = 772) and 69.8% (n = 773) of consented participants, respectively (Table 4). Individuals of Asian race, female sex, and those who met criteria for the Gene + FHx or FAMMM groups had lower rates of DNA and/or serum/plasma collection (Table 4; Supplementary Tables S3 and S4). However, there were no significant differences in biosample collection based on age (Supplementary Table S5).

Of individuals who underwent imaging after enrollment, the majority had EUS (62.2%, n = 408; Table 4). EUS was used at similar rates in most racial groups (Table 4), while it was used less often in those meeting FPC criteria compared with the other high-risk groups (Supplementary Table S4). EUS was also used less often in those under age 50 or over age 80, compared with those between ages 50 and 79 (Supplementary Table S5). There were no sex-based differences in imaging modality used post-enrollment (Supplementary Table S3).

Early detection of PDAC in HRIs is important for reducing long-term PDAC mortality. Although studies have provided early data that PDAC surveillance can downstage PDACs in HRIs and improve overall survival (14–18), study of PDAC surveillance in larger and more diverse cohorts is still needed to be able to generalize these findings to all individuals at increased PDAC risk. We report on the recent, rapid recruitment of more than 1,000 HRIs who meet criteria for PDAC surveillance to the multicenter, international PRECEDE Consortium. Our results highlight that while rapid recruitment of a geographically diverse cohort is feasible, racial, ethnic, and sex-based disparities persist, and strategies to diversify enrollment are warranted.

Enrollment in the PRECEDE Consortium reveals that despite increased awareness of disparities and efforts to reduce disparities in medicine, substantial race/ethnicity- and sex-based disparities still persist. Of HRIs enrolled, the majority were female (65.9%) and white (87.7%), with only a small fraction of enrollees reporting Hispanic ethnicity (2.9%; Table 1). These demographics are similar to other long-standing high-risk PDAC cohorts (14–18); however, they differ from U.S. Census data (50.5% female, 72.9% white, 18.8% Hispanic ethnicity; ref. 24) and SEER data for patients with PDAC (47.9% female, 78.3% white, 11.2% Hispanic ethnicity; ref. 25), and therefore do not reflect the expected demographics of the populations served by PRECEDE centers or the demographics of individuals diagnosed with PDAC. While data are not available to determine whether enrollment disparities result from lower rates of study consent by men and non-white individuals, based on our collective clinical experiences, we believe this is unlikely. Instead, it is more likely that there is less identification of certain at-risk individuals and lower referral rates of these individuals to surveillance programs, as prior work has shown variability among races in rates of previously unknown, newly identified BRCA1/2 PVs among patients with PDAC (26). Females enrolled in PRECEDE carried a PV at significantly higher rates than men (64% vs. 52.3%, P < 0.001; Supplementary Table S1), with large numbers of BRCA1/BRCA2/PALB2/ATM carriers. Identification of these hereditary breast and ovarian cancer genes is not surprising given that more than 50% of all cancers in PRECEDE participants were breast or ovarian cancer, which may explain some of the over-representation of female participants in PRECEDE.

Importantly, once patients enrolled in the PRECEDE Consortium, participation in subsequent research activities, such as biosample collection, was similar across all groups highlighting the importance of enrollment diversity. As equitable access to care is critically important, it is imperative that identification and referral of under-represented groups, in this case non-whites and men, be a major priority of participating PRECEDE centers as well as all centers that perform PDAC surveillance in HRIs.

While multiple, ongoing PDAC surveillance studies have been following participants for many years, the PRECEDE Consortium is unique as it started recruiting simultaneously across centers in 2020. The Consortium now has 32 participating centers and is the quickest recruiting PDAC early detection study to date. Barriers to care for HRIs, such as access to both genetic testing and health insurance, have existed for years (27, 28), with some improvement in these barriers over time (29). As such, a distinct advantage of the PRECEDE Consortium for examining disparities is that all participants were enrolled over the last 2 years, thus providing a unique and current snapshot of the international PDAC surveillance landscape. This recent timeline also affords the benefit of access to multigene panel testing in the majority of enrolled subjects (87.5% having undergone genetic testing), where participants have had a more comprehensive approach to genetic evaluation than in prior years.

One major challenge of the initial enrollment period for the PRECEDE Consortium is that enrollment began in May of 2020, which was the height of the COVID-19 pandemic. While COVID-19 led to substantial disruptions to PDAC surveillance (30), our data demonstrate that HRIs were able to be effectively enrolled throughout the COVID-19 pandemic across multiple institutions and multiple countries, and despite COVID-19 pandemic-related restrictions, biosamples were able to be successfully collected on approximately 70% of enrolled participants (Table 4). Furthermore, it is possible that the same drivers of widespread disparities associated with COVID-19 (31–34) also contributed to disparities observed in recruitment into PRECEDE.

Understanding factors associated with increased uptake of PDAC surveillance is also important. It is possible that having a PV in a PDAC risk gene or a personal history of cancer may influence an individual's decision to undergo PDAC surveillance. Interestingly, among high-risk males enrolled, there was a lower rate of having a PV in a PDAC risk gene as well as a lower rate of having a personal history of cancer compared with females (Supplementary Table S1), indicating that there may be other unknown factors that are driving certain subgroups to participate in PDAC surveillance. Fortunately, no such racially-based differences were observed (Table 2).

Quantifying an individual's PDAC risk relies on multiple upstream factors including awareness of family cancer history as well as access to and sharing of results of germline testing. Multiple upstream barriers may also contribute to the decision to conduct germline testing, including physician knowledge and perceptions of guidelines for genetic testing, individual knowledge of family history (35), as well as perceptions of and access to germline testing. While characterizing these barriers is beyond the scope of this study, addressing these barriers is critical and will impact the downstream identification and subsequent uptake of PDAC surveillance in HRIs. A diversity, equity, and inclusion (DEI) committee within PRECEDE is now actively working on strategies to address disparities identified in this analysis such as expansion to centers serving under-represented populations. Future projects of the DEI committee may also include better quantitation of the demographics of the populations served by PRECEDE centers, as well as spearheading consortium-wide initiatives, potentially through use of a Delphi method. Through these new initiatives we envision that in 5 years we will be able to double the percentage of under-represented groups in the PRECEDE Consortium and by that time, we also envision that recruitment of new individuals into the PRECEDE Consortium will be at a rate that more closely represents the demographics of patients served by PRECEDE centers.

This study does have limitations, including that a subset of patients was excluded due to incomplete data entry as data entry was still in progress for some enrolled subjects. In addition, while self-reported race/ethnicity is the expectation, it is possible that this data may have been derived from the medical record. Also, capturing race/ethnicity data followed traditional census categories; however, race/ethnicity is not monolithic so these categories may not adequately capture this data, and as such the PRECEDE DEI committee is working on adding methods to capture more granular race/ethnicity data. Disparities can result from multiple etiologies (social, biological, and genetic factors among others) and a limitation of our data is we are unable to determine the etiology of the disparities in PRECEDE enrollment. Furthermore, although there are numerous centers throughout the United States and internationally participating in PRECEDE, enrollment of participants through these tertiary academic centers is associated with selection bias. Finally, as enrollment into PRECEDE started at the beginning of the COVID-19 pandemic, pandemic-related restrictions may have introduced ascertainment bias into the individuals enrolled.

PDAC surveillance among individuals at high-risk for PDAC based on familial and/or known germline genetic susceptibility is incredibly important for early detection and the opportunity to downstage PDAC diagnosis for improved survival. While the PRECEDE Consortium was able to rapidly enroll many HRIs within its first 2-year study period, with high consent rates for both imaging and biosample use among all enrollees, race/ethnicity- and sex-based disparities in enrollment were evident. Reducing these disparities will be a major focus of the PRECEDE Consortium and PRECEDE's DEI committee moving forward to ensure that all HRIs have the opportunity to undergo potentially life-saving PDAC surveillance.

B.W. Katona reports other support from Immunovia outside the submitted work. R.E. Brand reports grants from Immunovia and grants from Freenome outside the submitted work. J.N. Everett reports non-financial support from Trovanow (nonprofit foundation) and non-financial support from PRECEDE Inc (nonprofit foundation) outside the submitted work. K. Hawthorne reports grants from REALM IDx and grants from Project Purple during the conduct of the study. V. Kaul reports personal fees from Cook Medical and CDX Diagnostics, and personal fees from STERIS outside the submitted work. S. Paiella reports personal fees from AlphaTau outside the submitted work. A.L. Lucas reports grants and personal fees from Immunovia AB and grants and personal fees from AIOnco outside the submitted work. No disclosures were reported by the other authors.

B.W. Katona: Conceptualization, formal analysis, supervision, investigation, writing–original draft, project administration, writing–review and editing. K. Klute: Conceptualization, investigation, writing–original draft, writing–review and editing. R.E. Brand: Investigation, writing–review and editing. J.N. Everett: Investigation, writing–review and editing. J.J. Farrell: Investigation, writing–review and editing. K. Hawthorne: Data curation, formal analysis, writing–review and editing. V. Kaul: Investigation, writing–review and editing. S.S. Kupfer: Investigation, writing–review and editing. S. Paiella: Investigation, writing–review and editing. D.M. Simeone: Supervision, investigation, project administration, writing–review and editing. D.A. Sussman: Investigation, writing–review and editing. G. Zogopoulos: Investigation, writing–review and editing. A.L. Lucas: Conceptualization, investigation, writing–original draft, writing–review and editing. F. Kastrinos: Conceptualization, investigation, writing–original draft, writing–review and editing.

This work was supported by grants from the Smith Family Research Fund (to B.W. Katona), Basser Center for BRCA (to B.W. Katona), Project Purple (to D.M. Simeone), and Trovanow (to D.M. Simeone).

Members of the PRECEDE consortium: Yan Bi, MD, PhD, Teresa Brentnall, MD, Daniel Chung, MD, Julie Earl, PhD, Srinivas Gaddam, MD, Steve Gallinger, MSc, MD, John J. Graff, PhD, MS, William Greenhalf, PhD, Aaron Grossberg, MD, PhD, Philip Hart, MD, Gregory Idos, MD, MS, Joanne Jeter, MD, James Lindberg, MD, Andrew Lowy, MD, Jennifer Permuth, PhD, MS, Maria Raitses-Gurevich, PhD, Intan Schrader, MBBS, PhD, Jens Siveke, MD, Elena Stoffel, MD, Marlies Vornhuelz, MD, Raymond Wadlow, MD.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.