Polygenic risk prediction in diverticulitis (2024)

Journal List
HHS Author Manuscripts
PMC10874245

As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsem*nt of, or agreement with, the contents by NLM or the National Institutes of Health.
Learn more: PMC Disclaimer | PMC Copyright Notice

Ann Surg. Author manuscript; available in PMC 2024 Jun 1.

Published in final edited form as:

Ann Surg. 2023 Jun 1; 277(6): e1262–e1268.

Published online 2022 Jul 25. doi:10.1097/SLA.0000000000005623

PMCID: PMC10874245

NIHMSID: NIHMS1822528

PMID: 35876359

Ana C De Roo,¹ Yanhua Chen,² Xiaomeng Du,² Samuel Handelman,³ Mary Byrnes,¹ Scott E Regenbogen,¹ Elizabeth K Speliotes,^2,⁴ and Lillias H Maguire⁵

Author information Copyright and License information PMC Disclaimer

The publisher's final edited version of this article is available at Ann Surg

Associated Data

Supplementary Materials

Abstract

Objective:

To derive and validate a polygenic risk score (PRS) to predict the occurrence and severity of diverticulitis and to understand the potential for incorporation of a PRS in current decision-making.

Summary Background Data:

PRS quantifies genetic variation into a continuous measure of risk. There is a need for improved risk stratification to guide surgical decision-making that could be fulfilled by PRS. It is unknown how surgeons might integrate PRS in decision-making.

Methods:

We derived a PRS with 44 SNPs associated with diverticular disease in the United Kingdom Biobank and validated this score in the Michigan Genomics Initiative (MGI). We performed a discrete choice experiment of practicing colorectal surgeons. Surgeons rated the influence of clinical factors and a hypothetical polygenic risk prediction tool.

Results:

Among 2,812 MGI participants with diverticular disease, 1,964 were asymptomatic, 574 had mild disease, and 274 had severe disease. PRS was associated with occurrence and severity. Patients in the highest PRS decile were more likely to have diverticulitis (OR = 1.84 (95%CI 1.42–2.38)) and more likely to have severe diverticulitis (OR = 1.61 (95% CI 1.04–2.51)) than the bottom 50%. Among 213 surveyed surgeons, extreme disease-specific factors had the largest utility (3 episodes in the last year, +74.4; percutaneous drain, + 69.4). Factors with strongest influence against surgery included 1 lifetime episode (−63.3), outpatient management (−54.9), and patient preference (−39.6) PRS was predicted to have high utility, (+71).

Conclusions:

A PRS derived from a large national biobank was externally validated, and found to be associated with the incidence and severity of diverticulitis. Surgeons have clear guidance at clinical extremes, but demonstrate equipoise in intermediate scenarios. Surgeons are receptive to PRS, which may be most useful in marginal clinical situations. Given the current lack of accurate prognostication in recurrent diverticulitis, PRS may provide a novel approach for improving patient counseling and decision-making.

We generated a polygenic risk score (PRS) in the United Kingdom Biobank and applied this to 2,812 diverticular disease patients in the Michigan Genomic Initiative. PRS was strongly associated with incident diverticulitis and severe diverticulitis. This may represent a novel approach to risk stratification.

Introduction

Like most common diseases, diverticulitis has both environmental and genetic risk factors¹. Environmental factors, such as diet, are the primary focus of diverticulitis research, but genetic research may provide insight. Physicians traditionally use family history as a proxy for genetic risk, and in diverticulitis family history has been consistently associated with recurrent and severe disease². However, family histories reflect both shared genetics and shared environment. Twin studies, which epidemiologically parse this gene-environment interaction, estimate 40–52% of individual risk of diverticulitis is heritable^3,4. Genome-wide association studies (GWAS) describe the genetic architecture of diverticulitis^5,6. Our own study of 27,444 patients identified 39 novel genetic loci associated with inpatient admission for diverticular disease⁷. GWAS associations can be translated into clinical relevance through polygenic risk scoring.

Polygenic risk scores (PRSs) quantify genetic risk for common diseases. The individually small effects of multiple single nucleotide polymorphisms (SNPs) are combined to generate a disease- and individual-specific PRS⁶. PRSs have been successfully studied in complex phenotypes such as coronary artery disease, diabetes, and common cancers^8–11 and made affordable and commercially available¹², but have yet to be applied to most surgical diseases. Additionally, it is unknown whether surgeons would use a PRS in surgical decision-making, particularly as diverticulitis is traditionally considered environmental.

PRS has the potential to improve patient counseling and surgical decision-making in diverticular disease by improving the current lack of risk stratification. Currently, physicians cannot predict which patients with diverticulosis will develop symptoms. Similarly, surgeons cannot predict which patients with acute diverticulitis will go on to suffer severe or recurrent disease. Clinical, radiologic, and endoscopic risk prediction tools have previously been proposed and prospectively studied^13–16 but have not been effectively adopted into clinical guidelines or common use. As a result of this lack of risk stratification, some patients with diverticulitis are over-treated with unnecessary surgery and others are under-treated, suffering recurrent episodes before a therapeutic operation. Nationally, local rates of colectomy vary widely¹⁷ and our own qualitative work demonstrated significant opportunity for improvement in decision-making (unpublished). Current guidelines recommend an “individualized approach” to recurrent diverticulitis¹⁸, but how surgeons select, weight, and incorporate patient- and disease-specific factors into clinical decision-making is neither well-studied nor prescribed. To understand the potential of PRS in diverticulitis, we concurrently attempt to identify areas for improvement in current decision-making paradigms and quantify the receptivity of surgeons for genetic risk scores.

In this project, we derive and evaluate a PRS for diverticulitis to determine if baseline genetic risk differs along the spectrum of diverticular disease: asymptomatic diverticulosis, mild diverticulitis, and severe diverticulitis. Additionally, we quantify the relative contributions of specific clinical factors and the potential impact of a PRS to the decision to recommend surgery in a discrete choice experiment, an approach used in medical and, more commonly, marketing research to understand what factors drive decision-making.

Methods

Polygenic Risk Score

Patient Populations

United Kingdom Biobank (UKBB) –

The UKBB has been well-described elsewhere, including its recruitment, informed consent, participant genotyping, data collection, and quality control^19,20. In brief, it consists of 500,000 British individuals recruited between the ages of 40 and 69. Participants’ clinical, demographic, and morphometric data were collected, de-identified, and linked to inpatient admissions in the National Health System (NHS). Participants then had their genotypes linked to clinical records. The UKBB protocols were approved by the National Research Ethics Service Committee. Participants were genotyped on one of two purpose-designed arrays (UK BiLEVE Axiom Array (N=50,520) and UK Biobank Axiom Array (N=438,692)) with 95% marker overlap. The Haplotype Reference Consortium was used as a reference panel to phase and impute the data. Participants were identified as having diverticular disease if they experienced an inpatient admission with corresponding NHS administrative International Classification of Diseases (ICD), Version 10 code, K57. Inpatient admissions data on UKBB participants are available beginning in 1997, 1998, and 1981 for England, Wales, and Scotland, respectively, and updated annually. Over 30 million variants in 408,455 individuals (http://geneatlas.roslin.ed.ac.uk/) were evaluated to identify SNPs independently associated with diverticular disease, after adjusting for age, gender, principal components and relatedness using mixed linear modeling. For full details of the initial GWAS, please see our previously published work⁷.

Michigan Genomics Initiative (MGI) –

The MGI is an institutional biobank of DNA obtained from patients in the University of Michigan Health System²¹. Genomic data are linked to participants’ medical profiles via the electronic medical record at the University of Michigan. The MGI is approved for research use by the Institutional Review Board of the University of Michigan and use of participant data for this project was approved by the University of Michigan’s Institutional Review Board. Following informed consent, individuals (N=35,888) were genotyped using the Illumina HumanCoreExome Array. Genotype analysis was performed with Illumina GenomeStudio (module 1.9.4, algorithm GenTrain 2.0). Variants were clustered and quality-controlled. We excluded variants if: (1) their probes could not be perfectly mapped or mapped perfectly to multiple positions (2) they showed deviations from Hardy Weinberg equilibrium (p-value< 0.0001), (3) had a call rate < 99%, or (4) another variant with higher call rate assayed the same variant (PLINK (v1.90)²²). Imputation was carried out using the Haplotype Reference Consortium (chromosome 1–22: HRC release 1; chromosome X: HRC release 1.1). Excluding variants with low imputation quality (R2 <0.3) resulted in dense mapping at 39,127,678 million quality-imputed genetic markers. We excluded any 1st- or 2nd-degree relative pairs within the cohort using KING (v.1.4.2)²³. In addition, we used principal component analysis to identify ethnically hom*ogeneous groups using individuals from the Human Genome Diversity Project^24,25. For both datasets we included only European-ancestry samples, to reduce heterogeneity and maximize power.

Rationale for use of Two Biobanks

We used two separate biobanks for creation and validation of the PRS, providing some evidence that the PRS is not population-specific. Additionally, the disparate sizes and clinical details of the two biobanks support the work. The UKBB provides the sample size necessary for discovery of disease-associated SNPs at the genome-wide significant level (p < × 10–8). The MGI allows for study of those identified SNPs at the level of clinical detail which is relevant to physicians. Severity of diverticulitis was assessed only within MGI, in which we were able to use identified, encounter-level data. Previous work has shown that in diverticular disease, sub-codes of the ICD-10 K57, which are meant to indicate severity, are not actually reliably associated with disease severity when assessed clinically²⁶.

Diverticulitis Sub-phenotyping

To identify MGI patients with diverticular disease we queried the institutional electronic medical record for billed encounters associated with ICD-9 codes for diverticular disease (562.11 and 562.13 for diverticulitis, 562.10 and 562.12 for diverticulosis) from 1999–2018. We excluded patients <18 years of age, patients with ulcerative colitis, Crohn’s disease, or colorectal cancer, and patients with historical diagnoses of diverticular disease without billed encounters for the diagnosis, due to the potential conflation of diverticulosis and diverticulitis. Within this cohort we defined three sub-phenotypes of diverticular disease: asymptomatic, mild, and severe. Asymptomatic patients were those with a billing code for diverticulosis on the same day as a billed colonoscopy, but no encounters for diverticulitis. Mild diverticulitis comprised patients with either diverticulitis treated only in the outpatient setting or no more than 1 inpatient encounter for diverticulitis, and no procedural codes for colectomy or percutaneous drainage. Severe diverticulitis patients included those with more than 1 inpatient admission for diverticulitis or procedural (ICD or CPT) coding for percutaneous drainage or surgery for diverticulitis.

Polygenic Risk Score Derivation

We used conditional and joint association analysis (COJO)²⁷ to select independent SNPs with which to construct our PRS. In general, a PRS can be considered a weighted sum of the number of SNPs carried by an individual. To construct our polygenic risk score, we used independent SNPs that were associated with inpatient admissions for diverticular disease (K57) in the UKBB at the genome-wide significant level (at p < 5 × 10–8), were relatively common in the population (minor allele frequency >0.0001), which had high imputation quality info score >0.85), and which were shared between the different genotyping chips used by UKBB and MGI. We removed in-del variants, multi-allelic variants, and ambiguous SNPs. This yielded 44 SNPs (Supplementary Table 1). We constructed a PRS for each participant by summing the participant’s unique combination of individual SNPs weighted by the SNP’s effect size (β) as determined from the UKBB. We considered MGI asymptomatic diverticulosis participants to define the spectrum of at-risk patients, rather than the entire MGI population. We used their PRSs to create the risk distribution, divided this into deciles, then scored the diverticulitis participants and mapped them into the risk deciles created by the diverticulosis cohort.

Discrete Choice Experiment

Study setting and participants

We conducted a web-based study of colorectal surgeons in the U.S. to understand the relative importance of factors used in deciding whether or not to offer elective surgery for recurrent diverticulitis. We used a discrete choice experiment to describe the relative importance of components of a decision. Discrete choice experiments are commonly used in marketing research and evaluate a service or good.¹⁴ DCE in healthcare often seek to understand tradeoffs among components in decisions or processes, particularly with the end-user (patient or prescriber) in mind²⁸. In surgery, discrete choice experiments have evaluated which aspects of the treatments drive patients’ preferences for bariatric surgical procedures, surgical management of arthritis, and cataract intervention, among many others.^29–31

Surgeons were recruited via e-mail through the American Society of Colon & Rectal Surgeons (ASCRS) member list in May 2020. Participants were eligible if they were practicing colorectal surgeons. Demographic information was obtained by respondent self-report. We excluded participants with partial response to the survey. Participants received a small monetary gift ($5) to as a token of appreciation. The study was reviewed by the University of Michigan Institutional Review Board and deemed exempt from oversight (HUM00165726).

Development of attributes and levels

The identification of attributes for this discrete choice experiment was based on semi-structured interviews of practicing surgeons, clinical options for diverticulitis treatment, and data regarding a polygenic risk score and its development. The final 5 attributes included: total number of attacks and tempo, previous episode severity, patient preference, patient lifestyle factors, and the outcome of a hypothetical genetic prediction tool. Levels were selected based on data from semi-structured interviews and a panel including practicing colorectal surgeons and the study authors. The survey introduction included an explanation of the discrete choice experiment followed by the clinical scenario of a patient presenting to clinic 8 weeks after a confirmed episode of diverticulitis. The question prompt was “in which case would you be MOST willing to offer an operation?”. We used a mixed logit model to estimate the relative importance surgeons placed on each attribute/level. Each surgeon provides 10 observations and the impact of each attribute on surgeon preference was analyzed by treating the decision as a dependent variable and the levels as independent variables.

Results

Polygenic Risk Score

In MGI we identified 2,812 patients including 1,964 patients with asymptomatic diverticulosis (69.7%), 574 patients with mild diverticulitis (20.4%), and 274 patients with severe diverticulitis (9.7%). The age, sex, and BMI of each group are shown in Table 1. The median polygenic risk score increased between groups (Figure 1).

Open in a separate window

Figure 1:

Total numbers of patients with mild and severe disease by decile of polygenic risk. Numbers above columns indicate percent of patients with severe diverticulitis

Table 1:

Patient Characteristics

	Asymptomatic	Mild	Severe	P
Patients, no. (%)
Female	956 (48.7)	302 (52.6)	139 (50.7)
Male	1008 (51.3)	272 (47.4)	135 (49.3)	0.24
Age, mean (SD), yr	59.7 (8.6)	59.8 (11.7)	57.5 (12.0)	0.0011
BMI, mean (SD), kg/m2	30.9 (6.8)	31.3 (6.6)	30.1 (6.4)	0.034

Open in a separate window

There was a significant increase in the incidence of diverticulitis, and in the proportion of patients with severe disease among those at higher predicted risk based on PRS (p=<0.001) (Figure 1). The rate of severe diverticulitis for patients (38.7%) in the top decile was more than double the rate for those in the lowest decile (17.5%) (OR = 2.99 (95% CI 1.42–6.28, p =0.004).

We then performed multivariable logistic regression incorporating age, sex, and body mass index (BMI). We found that PRS was significantly higher in diverticulitis patients versus diverticulosis patients (Table 2). We then determined whether PRS could identify high risk patients we found that patients in the top decile of polygenic risk were significantly more likely to have diverticulitis when compared to those in the bottom 50% (OR = 1.84 (95%CI 1.42–2.38, p <0.001)).

Table 2:

Association of Polygenic Risk with Incident Diverticulitis Adjusted for Age, Sex, and BMI

Risk Factor	Asymptomatic	Symptomatic	P
	OR	95% CI	P
Top PRS Risk Decile	1.84	1.42–2.38	<0.001
Bottom 50% PRS Risk Distribution	Reference
Age	0.99	0.98–1.01	0.54
Male Sex	1.01	0.81–1.26	0.90
BMI	1	0.98–1.02	0.97

Open in a separate window

Polygenic risk score was also significantly associated with severe diverticulitis (p=0.03) when controlling for age, sex, and BMI (Table Three). Patients in the top decile of PRS were more likely to have severe diverticulitis compared to the bottom 50% (OR = 1.61 (95% CI 1.04–2.51, p =0.03)). However, patients in the top 10% of the PRS were not at significantly higher risk when compared to the entire other 90% of the patients. We then determined whether the PRS could identify patients with low risk of severe disease. Patients in the bottom 10% of polygenic risk were at significantly lower risk for severe disease than the other 90% of patients (OR = 0.44 (95% CI = 0.21–0.82, p =0.015)). Age and sex were not significantly associated with diverticulitis, but both age and BMI were associated with less risk of severe diverticulitis (Table Three).

Table 3:

Association of Polygenic Risk with Severe versus Mild Diverticulitis, Adjusted for Age, Sex, and BMI

Risk Factor	Mild	Severe	P
	n=574	n=274
	OR	95% CI	P
Top PRS Risk Decile	1.61	1.04–2.51	0.033
Bottom 50% PRS Risk Distribution	Reference
Age,	0.97	0.96–0.99	0.004
Male Sex	0.9	0.60–1.35	0.63
BMI	0.99	0.954–1.02	0.39
	OR	95% CI	P
Top PRS Risk Decile	1.36	0.91–2.02	0.13
Bottom 90% PRS Risk Distribution	Reference
Age	0.98	0.97–0.99	0.0011
Male Sex	1.03	0.77–1.37	0.85
BMI	0.96	0.94–0.99	0.0021
	OR	95% CI	P
Bottom PRS Risk Decile	0.44	0.21–0.82	0.015
Top 90% PRS Risk Distribution	Reference
Age	0.98	0.97–0.99	0.002
Male Sex	1.04	0.78–1.40	0.78
BMI	0.96	0.94–0.99	0.002

Open in a separate window

Discrete Choice Experiment

There were 238 respondents out of 1321 emailed surgeons for an 18% response rate. 213 respondents completed the full experiment (89%). Demographics of complete survey participants, whose responses were used for the analysis, are listed in Supplementary Table 2. Most participants were white (72.3%), male (78.9%), and practiced in a major metropolitan area (90.1%). There was a relatively even split between academic and community surgeons and a variety of years of experience. Institutional variation in the specialty caring for inpatients with diverticulitis was evident; 37% and 43% of respondents reported colorectal surgeons cared for uncomplicated and emergent diverticulitis in their practice setting, respectively. The majority of the responding colorectal surgeons reported that diverticulitis made up 10–40% of their practice.

The utilities (relative influences) of attributes and levels are demonstrated in Figure 2. Disease-specific factors were most influential in favor of recommending surgery. The greatest influence on recommending surgery was 3 attacks in the last year and 3 lifetime episodes with a utility of +74.4. Strongest negative utilities were 1 lifetime episode (−63.3) and management with outpatient oral antibiotics (−54.9). Patient-specific factors also influenced decision-making but to a lesser extent than disease-specific factors. Mid-range tempo of episodes and inpatient antibiotic treatment represented “gray zones” with little influence to recommend strongly for or against surgery.

Open in a separate window

Figure 2:

Influence of Clinical and Lifestyle Factors on Recommendation for Surgery (Utilities)

When asked to consider how a polygenic risk prediction tool would affect decision-making, surgeons responded that a prediction of future complicated diverticulitis with high reliability of the test would strongly influence a decision for surgery (+ 71.9). Even with a lower confidence in a risk stratification tool, there was still an influence to recommend surgery based on polygenic testing (+ 23.3). No prediction or prediction of mild disease were both associated with recommendation against surgery.

Discussion

In this multi-stage study, we derived a PRS in the UK Biobank, validated it in the Michigan Genomics Initiative, and evaluated its potential clinical utility with surgeons experienced in the care of patients with diverticulitis. This study is novel in its application of translational genomics to diverticulitis and its integrated assessment of clinical utility of PRS to surgeons. We found that our 44-SNP PRS was associated with the occurrence of diverticulitis among those with diverticulosis. It also associated with higher severity of disease. Patients in the top PRS decile had approximately three times the odds of severe disease compared to those the lowest decile. When controlling for age, sex, and BMI, patients in the highest and lowest deciles of PRS were at significantly different risk of severe disease compared to other patients. If further validated, these findings could have significant influence in clinical care. For those found to have diverticulosis on colonoscopy, the data suggest that PRS might predict their likelihood of future diverticulitis and perhaps guide dietary modification or other interventions. For those who have had uncomplicated diverticulitis, it might predict the likelihood of future complications of diverticulitis, and contribute to decision-making around elective surgical resection.

In the discrete choice experiment, we find that in the context of individualized recommendations for routine colectomy, number and frequency of episodes have the strongest influence on surgeons’ decision-making. Although guidelines no longer recommend decisions based number of episodes alone, very high or low frequencies (>3 in 6 months, 1 in the last year) anchored decisions to operate, and not operate, respectively. These findings suggest that in clinical situations with extremes of severity, surgeons feel that they have clear guidance for recommending for or against surgery. However, in early disease or intermediate-severity scenarios, there is less certainty. In reality, in these scenarios more aggressive surgeons and patients might operate, risking over-treatment, or wait-and-see, risking under-treatment. It is in these marginal scenarios that surgeons might use a PRS to counsel patients and guide decision-making.

PRS has been developed in multiple different common medical diseases, but have been rarely applied to surgical decision-making. PRS in coronary artery disease (CAD) identifies individuals at equivalent CAD risk to patients with familial hypercholesterolemia despite normal clinical risk factors⁹. These data suggest that polygenic scores may measure some disease risk not captured by clinical risk factors or Mendelian syndromes. PRS can also target high risk individuals to preventative therapies: post hoc analysis of randomized clinical trials of PCSK9 inhibition in CAD demonstrated that PRS could identify which trial participants were most likely to benefit from PCSK9 inhibitors and which individuals in the placebo arms were most likely have major cardiac events^34,35. As in our study, PRS has been most frequently studied retrospectively. However, prospective randomized trials of PRS in CAD revealed that personalized medical data may have a significant impact on patients. Individuals randomized to receive a PRS, risk reduction counseling, and clinical risk score had lower cholesterol and greater statin use at 6 months than those who received risk reduction counseling and a clinical risk score alone³⁶. The aforementioned studies provide a roadmap of application of PRS to targeted primary and secondary prevention in diverticular disease. Compared to CAD, one opportunity of application of PRS in diverticulitis is the lack of a pre-existing robust clinical risk score like the Framingham model. However, the preventative intervention, sigmoid colectomy, carries far more risk. Fortunately, practical application for future research is feasible. PRS are created using widely available next-generation sequencing chips using patient DNA samples obtained from blood or cheek swabs. PRS is rapid, affordable, and currently commercially available. If future prospective analysis proves a benefit of PRS to diverticulitis patients, the technological and cost barrier for incorporation into the clinical setting is low.

Both arms of our study have several limitations. In terms of PRS, MGI participants are recruited in peri-procedural settings, tending to enrich the population for severe disease. Second, the severe disease phenotype is likely to be influenced by surgical and medical decision-making as well as disease severity. Younger patients with lower BMI might be perceived as better surgical candidates and therefore be more likely to be captured by the “severe” phenotype. We mitigated this surgical selection bias by including multiple inpatient admissions as a severe disease criterion. Third, potential environmental factors associated with diverticular disease, such as diet and physical activity are poorly captured in medical documentation. Fourth, our PRS was developed within two biobanks primarily compromised of participants with European ancestry. Therefore, our PRS reflects the overrepresentation of this group within the biobanks and may not generalize well to other populations. This is a problem for the entire field of precision medicine due to the small sample size of non-European ancestry populations in most biobanks and the higher genetic diversity in these populations, precluding meaningful genetic analysis^37,38. In this particular case, we do however recognize that diverticulitis is a distinct anatomic and clinical entity in European versus Asian ancestry populations and that there is relatively limited data on disease prevalence in modern African populations. Fifth, our study is limited by reliance on administrative codes which could misclassify disease. In particular, we elected a strict definition of asymptomatic diverticulosis, likely excluding many patients with asymptomatic lesions. However, this defines a clinically relevant at-risk population and will, if anything, bias the findings toward the null rather than exaggerate the estimates. Finally, the degree of discrimination was somewhat limited. Like all complex traits, severe diverticulitis develops as a result of a complex interplay between genetic and environmental risk. As such, it is unsurprising that the polygenic data alone fail to generate extreme risk categories, but may inform marginal cases in which there is equipoise.

Our discrete choice experiment also has limitations. First, the attributes and levels of our discrete choice experiment do not reflect all common clinical scenarios. However, we were able to represent common real-world presentations. Furthermore, we selected these attributes and levels based on in-depth qualitative interviews with patients and surgeons and field-tested them with practicing surgeons for clarity.¹⁹ Next, we did not include specific patient-level factors such as comorbidity, obesity, and age that influence surgical decision-making^39–41. However, these factors are not specific to the considerations for management of recurrent diverticulitis. Future work should include general patient health characteristics that are known to influence risk stratification⁴². Third, our study cohort was a convenience sample of colorectal surgeons and the results may not be generalizable to all surgeons. Finally, we field-tested the concept of a PRS rather than actual metrics in order to determine potential receptivity and utility. This both minimized complexity and recognized our current PRS will likely evolve as prospective data are accrued.

Despite these limitations, our study is novel in its application of PRS to a clinical scenario that could improve surgical decision-making. In this study, we demonstrate the ability to identify individuals within an “at risk” population with diverticulosis, and potentially inform a high-stakes surgical decision. We demonstrate the need for improved decision-making in intermediate clinical scenarios and the receptiveness of surgeons to integrating translational genomics into personalized surgical decision-making.

Supplementary Material

Supplemental Table 1

Click here to view.^{(16K, xlsx)}

Supplemental Table 2

Click here to view.^{(14K, docx)}

Supplemental Figure 1

Click here to view.^{(2.0M, docx)}

Acknowledgements

The authors acknowledge the Michigan Genomics Initiative participants, Precision Health at the University of Michigan, the University of Michigan Medical School Central Biorepository, and the University of Michigan Advanced Genomics Core for providing data and specimen storage, management, processing, and distribution services, and the Center for Statistical Genetics in the Department of Biostatistics at the School of Public Health for genotype data curation, imputation, and management in support of the research reported in this publication.

Funding Sources:

Blue Cross Blue Shield Foundation of Michigan Physician Investigator Award – ADR, LHM
NIH/NIDDK - 5K08DK124687 – LHM
EKS, SKH, XD and YC are supported in part by R01DK106621 (to EKS), R01DK107904 (to EKS) and The University of Michigan Department of Internal Medicine.

References

1. Strate LL, and Morris AM. Epidemiology, pathophysiology, and treatment of diverticulitis. Gastroenterology. 2019; 156: 1282–1298. [PMC free article] [PubMed] [Google Scholar]

2. Lmalki T, Garfinkle R, Kmiotek E, et al. Family history is associated with recurrent diverticulitis after an episode of diverticulitis managed nonoperatively. Diseases of the Colon and Rectum. 2020; 63, 944–954. [PubMed] [Google Scholar]

3. Strate LL, Erichsen Rune, Baron John A., et al. Heritability and Familial Aggregation of Diverticular Disease: A Population-Based Study of Twins and Siblings. Gastroenterology. 2013; 144, 736–742. [PubMed] [Google Scholar]

4. Granlund J, Svensson T, Olén O, Hjern F, Pedersen NL, Magnusson PKE, Thelin Schmidt PThe genetic influence on diverticular disease - A twin study. Alimentary Pharmacology and Therapeutics.2012; 35: 1103–110. [PubMed] [Google Scholar]

5. Reichert MC, Kupcinskas J, Schulz A, et al. Common variation in FAM155A is associated with diverticulitis but not diverticulosis. Sci Rep. 2020; 10: 1658. [PMC free article] [PubMed] [Google Scholar]

6. Sigurdsson S, Alexandersson KF, Sulem P, et al. Sequence variants in ARHGAP15, COLQ and FAM155A associate with diverticular disease and diverticulitis. Nat Commun. 2017; 8: 15789. [PMC free article] [PubMed] [Google Scholar]

7. Maguire LH, Handelman SK, Du X, Chen Y, Pers TH, Speliotes, EK. Genome-wide association analyses identify 39 new susceptibility loci for diverticular disease. Nature Genetics. 2018; 50, 1359–1365. [PMC free article] [PubMed] [Google Scholar]

8. Torkamani Ali, Wineinger Nathan E., and Topol Eric J.. The personal and clinical utility of polygenic risk scores. Nature Reviews Genetics. 2018; 19: 581–590. [PubMed] [Google Scholar]

9. Khera Amit V., et al. Genome-wide polygenic scores for common diseases identify individuals with risk equivalent to monogenic mutations. Nature Genetics. 2018; 50: 1219–1224. [PMC free article] [PubMed] [Google Scholar]

10. Mavaddat N, Michailidou K, Dennis J, Lush, et al. Polygenic risk scores for prediction of breast cancer and breast cancer subtypes. The American Journal of Human Genetics. 2019; 104, 21–34. [PMC free article] [PubMed] [Google Scholar]

11. Garcia-Closas M, Rothman N, Figueroa JD, et al. Common genetic polymorphisms modify the effect of smoking on absolute risk of bladder cancer. Cancer research. 73; 2211–2220. [PMC free article] [PubMed] [Google Scholar]

12. 23andMe will tell you how your DNA affects your diabetes risk. STAT2019. https://www.statnews.com/2019/03/10/23andme-will-tell-you-how-your-dna-affects-your-diabetes-risk-will-it-be-useful/.

13. Sallinen V, Mali J, Leppäniemi A, Mentula P. Assessment of risk for recurrent diverticulitis: a proposal of risk score for complicated recurrence. Medicine (Baltimore). 2015; 94: e557. [PMC free article] [PubMed] [Google Scholar]

14. Shin S, Kim D, Kang UR, Yang CS. Impact of CT imaging on predicting the surgical management of acute diverticulitis. Ann Surg Treat Res. 2018;94(6):322–329. [PMC free article] [PubMed] [Google Scholar]

15. Tursi A, Brandimarte G, Di Mario F, et al. Development and Validation of an Endoscopic Classification of Diverticular Disease of the Colon: The DICA Classification. Dig Dis. 2015; 33: 68–76. [PubMed] [Google Scholar]

16. Tursi Antonio, et al. Prognostic performance of the ‘DICA’ endoscopic classification and the ‘CODA’ score in predicting clinical outcomes of diverticular disease: an international, multicentre, prospective cohort study.” GutPublished Online First: 26October2021. [PubMed] [Google Scholar]

17. Hawkins AT, Samuels LR, Rothman R, Geiger TM, Penson DF, Resnick MJ. National Variation in Elective Colon Resection for Diverticular Disease. Annals of Surgery. 2020 [PMC free article] [PubMed] [Google Scholar]

18. Hall J, Hardiman K, Lee S, et al. The American Society of Colon and Rectal Surgeons clinical practice guidelines for the treatment of left-sided colonic diverticulitis. Diseases of the Colon & Rectum. 2020;63(6):728–747 [PubMed] [Google Scholar]

19. Bycroft C, Freeman C, Petkova Det al. The UK Biobank resource with deep phenotyping and genomic data. Nature. 2018; 562, 203–209. [PMC free article] [PubMed] [Google Scholar]

20. Sudlow C, Gallacher J, Allen N, et al. UK biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age. PLoS medicine. 2015; 12, e1001779. [PMC free article] [PubMed] [Google Scholar]

21. https://precisionhealth.umich.edu/our-research/michigangenomics/ [Google Scholar]

22. Chang CCet al. Second-generation PLINK: rising to the challenge of larger and richer datasets. Gigascience. 2015; 4, 7. [PMC free article] [PubMed] [Google Scholar]

23. Manichaikul Aet al. Robust relationship inference in genome-wide association studies. Bioinformatics. 2010; 26: 2867–2873. [PMC free article] [PubMed] [Google Scholar]

24. Wang K, Li M & Hakonarson HANNOVAR: Functional annotation of genetic variants from next-generation sequencing data. Nucleic Acids Research38, e164 (2010). [PMC free article] [PubMed] [Google Scholar]

25. Cavalli-Sforza LL. The human genome diversity project: past, present and future. Nature Reviews Genetics. 2005. Apr;6(4):333–40. [PubMed] [Google Scholar]

26. Erichsen R, Strate L, Sørensen HT, Baron JA. Positive predictive values of the International Classification of Disease, 10th edition diagnoses codes for diverticular disease in the Danish National Registry of Patients. Clin Exp Gastroenterol. 2010; 3: 139–142. [PMC free article] [PubMed] [Google Scholar]

27. Yang J, Ferreira T, Morris AP, et al. Conditional and joint multiple-SNP analysis of GWAS summary statistics identifies additional variants influencing complex traits. Nat Genet. 2012; 44: 369–S3. [PMC free article] [PubMed] [Google Scholar]

28. Lancsar E, Louviere J. Conducting discrete choice experiments to inform healthcare decision making. Pharmacoeconomics. 2008; 26(8):661–677. [PubMed] [Google Scholar]

29. Rozier MD, Ghaferi AA, Rose A, Simon N-J, Birkmeyer N, Prosser LA. Patient preferences for bariatric surgery: findings from a survey using discrete choice experiment methodology. JAMA surgery. 2019;154(1):e184375–e184375. [PMC free article] [PubMed] [Google Scholar]

30. Rozier MD, Ghaferi AA, Rose AM, Simon N-J, Birkmeyer NJ, Prosser LA. Preferences for bariatric surgery: comparing surgeons and patients using a discrete choice experiment. Surgery for Obesity and Related Diseases. 2020;16(6):738–744. [PubMed] [Google Scholar]

31. Harris CA, Shauver MJ, Yuan F, Nasser J, Chung KC. Understanding patient preferences in proximal interphalangeal joint surgery for osteoarthritis: a conjoint analysis. The Journal of hand surgery. 2018;43(7):615–624. e614. [PubMed] [Google Scholar]

32. Orme BK. Getting started with conjoint analysis: strategies for product design and pricing research. 2006. [Google Scholar]

33. Marston NA, Kamanu FK, Nordio F, et al. Predicting benefit from evolocumab therapy in patients with atherosclerotic disease using a genetic risk score: results from the FOURIER trial. Circulation. 2020; 141:616–623. [PMC free article] [PubMed] [Google Scholar]

34. Damask A, Steg PG, Schwartz GG, et al. Patients with high genome-wide polygenic risk scores for coronary artery disease may receive greater clinical benefit from alirocumab treatment in the ODYSSEY OUTCOMES trial. Circulation. 2020; 141:624–636. [PubMed] [Google Scholar]

35. Kullo IJ, Jouni H, Austin EE, et al. Incorporating a Genetic Risk Score Into Coronary Heart Disease Risk Estimates: Effect on Low-Density Lipoprotein Cholesterol Levels (the MI-GENES Clinical Trial). Circulation. 2016; 133:1181–1188. [PMC free article] [PubMed] [Google Scholar]

36. Martin AR, Kanai M, Kamatani Y, Okada Y, Neale BM, Daly MJ. Clinical use of current polygenic risk scores may exacerbate health disparities. Nat Genet. 2019;51(4):584–591. [PMC free article] [PubMed] [Google Scholar]

37. Landry LG, Ali N, Williams DR, Rehm HL, Bonham VL. Lack Of Diversity In Genomic Databases Is A Barrier To Translating Precision Medicine Research Into Practice. Health Aff (Millwood). 2018;37(5):780–785. [PubMed] [Google Scholar]

38. Hughes TM, Dossett LA, Hawley ST, Telem DA. Recognizing Heuristics and Bias in Clinical Decision-making. Annals of Surgery. 2020;271(5):813–814. [PubMed] [Google Scholar]

39. Hughes TM, Palmer EN, Capers Q, et al. Practices and perceptions among surgical oncologists in the perioperative care of obese cancer patients. Annals of surgical oncology. 2018;25(9):2513–2519. [PubMed] [Google Scholar]

40. Palmer Kelly E, Hyer M, Paredes AZ, et al. Assessing Differences in Cancer Surgeon Approaches to Patient-Centered Decision-Making Using Vignette-Based Methodology. Annals of Surgical Oncology. 2020:1–8. [PubMed] [Google Scholar]

41. Liu Y, Cohen ME, Hall BL, Ko CY, Bilimoria KY. Evaluation and enhancement of calibration in the American College of Surgeons NSQIP Surgical Risk Calculator. Journal of the American College of Surgeons. 2016;223(2):231–239. [PubMed] [Google Scholar]