Pharmacogenomics knowledge within Neurology: A brief survey amongst physicians

doi:10.21203/rs.3.rs-2996973/v1

Pharmacogenomics has emerged as a tool to optimise treatment. Many neurologists encounter drugs with pharmacogenomic associations with unclear guidelines. We therefore aimed to determine the knowledge, attitudes, and barriers to implement pharmacogenomics in clinical practice within a neurology unit. We developed and administered a survey on pharmacogenomics to physicians working in a neurology unit. Three drug-gene pairs clinically relevant to neurology were presented as case scenarios to assess physicians’ perceived competency, attitudes, and risks towards pharmacogenomics. Thirty-two (27.4%) physicians responded to the survey. Respondents were most confident about activities involving the HLA-B*15:02 test, followed by the CYP2C19 test, and lastly the CYP2C9 test. However, perceived competency in drug-drug-gene-interactions was consistently low for all three tests. Referring to the product information leaflet ranked highest in usefulness, followed by local practice guidelines. In conclusion, local clinical guidelines and training on the clinical applications of pharmacogenomics are important to facilitate implementation of pharmacogenomics.

Biological sciences/Genetics/Genomics/Pharmacogenomics

Biological sciences/Genetics/Clinical genetics/Genetic testing

Health sciences/Health care/Health services

Precision medicine is an emerging approach to disease treatment and prevention that considers an individual’s gene variability, environment and lifestyle. Pharmacogenomics, the study of how DNA and RNA variation relate to drug response, is a core element of precision medicine. Pharmacogenomics helps to guide drug prescription and dosing decisions with the overall aim to improve clinical outcomes and reduce adverse effects(1). Concerns about the clinical validity and utility of pharmacogenomics, lack of clear clinical guidelines, perceived lack of self-competency, and contradictory results from pharmacogenomic studies are some of the macro level barriers to clinical implementation of pharmacogenomics(2, 3). Turnaround time to receive results, accuracy and cost of pharmacogenomic testing, and lack of clarity on how test results affect patient’s insurability are some of the micro level barriers to clinical implementation of pharmacogenomics(4).

The United States Food and Drug Administration recommendations on pharmacogenomics can be classified into three groups (5): 1. Associations for which the data support therapeutic management recommendations; 2. Associations for which the data indicate a potential impact on safety or response; 3. Associations for which the data indicate a potential impact on pharmacokinetic properties only. Neurologists have several drug-gene associations for which the data support therapeutic management recommendation, including HLA-B*15:02 genotyping before initiation of carbamazepine in new patients of Asian ancestry, CYP2C19 genotyping to assess treatment efficacy of clopidogrel, and recently CYP2C9 genotyping before initiation of siponimod. At present, little is known about neurologists’ concerns about the practical aspects of incorporating pharmacogenomics into clinical decision making.

As such, we developed a survey to determine the knowledge, attitudes, and practice towards the clinical application of pharmacogenomics within a tertiary neurology centre.

Study design and participants

This cross-sectional study was conducted using a voluntary, self-administered questionnaire using the online survey tool Qualtrics (UT, USA) between 28th March 2022 and 30th June 2022. The target population for this survey was all 117 physicians engaged in clinical practice of neurology from the National Neuroscience Institute campuses in Singapore General Hospital and Tan Tock Seng Hospital. This included non-specialists (general medicine medical officers who were doing a rotation in Neurology, Neurology trainees) and Neurologists registered with the Singapore Medical Council. Study invitation was disseminated through the principal investigator via department e-mails to all prospective participants. Reminder e-mails were sent on Day 4, Day 22, and 1 week prior to the end of the survey to increase response rates. All participants provided electronic consent before starting the survey. Institutional review board exemption was granted as this was an anonymous survey, and no identifiable information was collected.

Questionnaire

A 27-item questionnaire was developed for this study in consultation with the principal investigator and study collaborators who have experience with pharmacogenomics. The questionnaire was checked for face and content validity, with pre-testing done with a group of six individuals comprising two neurologists who were study collaborators, two internal medicine residents and two neurology trainees. Their feedback on the relevance, clarity, order, and time taken to complete the survey were collected.

The questionnaire (Supplementary Fig. 1) contained three sections, requiring approximately 15 minutes to complete. The first section comprised seven multiple-choice-questions on demographics. The second section comprised three case scenarios regarding the clinical application of pharmacogenomics on the use of carbamazepine, clopidogrel, and siponimod. Case scenarios were formulated based on guidelines recommendations and clinical information reflecting current standard of practice(5, 6).

Each case scenario contained four to five main questions evaluating perceived competency and perceived attitudes. Questions on perceived competency required participants to rank confidence levels when performing clinical tasks, using a 5-point Likert scale. It also included a multiple-choice-question on pharmacotherapeutic choice. Questions on perceived attitudes included two multiple-choice-questions to understand if the information provided along with the test result was adequate in providing guidance in the clinical decision-making and whether genetic testing is a useful tool in various scenarios.

The last section was on training, and perceived risks and barriers. The former involved two multiple-choice-questions on whether participants had received formal training in pharmacogenomics and their preferred learning modality. A question requiring participants to rank the usefulness levels of various resources in incorporating pharmacogenomics in their practice was included. The question on perceived barriers required participants to rank what they felt was the most important barrier in utilising pharmacogenomic testing in clinical practice.

Statistical analysis

Anonymised survey responses were collated electronically on Qualtrics, summarised using descriptive statistics. Analysis was performed using IBM SPSS Statistics for Windows, Version 21.0 (Armonk, NY). Categorical variables were presented as frequencies and percentages. Ordinal variables were presented as median (interquartile range) or number (percentage). For ranking questions on the usefulness level of information sources and importance levels of barriers, responses were assigned numerical scores, and means compared between subgroups. Correlation between confidence levels in performing clinical activities and demographics was evaluated using Spearman’s. Statistical significance was set at p < 0.05. Narrative comments were extracted and analysed thematically.

Thirty-two of 117 (27%) of invited doctors responded to the survey. Majority of the respondents were male (75.0%), had between 11 and 20 years of experience (43.8%), and were specialists (68.7%). Around half of respondents ordered more than 20 pharmacogenomic tests in the past year (Table 1).

Perceived attitudes and competency

Thirty-one respondents (96.9%) agreed that the application of pre-emptive pharmacogenomic testing for actionable drug-gene pairs in neurology likely improves clinical outcomes. Respondents were more confident about activities involving the HLA-B*15:02 test, CYP2C19 test, than the CYP2C9 test (Table 2). Confidence in knowing how the test result affects the patient’s other drugs was consistently ranked low.

Correlational analysis found perceived competency in interpreting various test results, choosing an appropriate drug, and knowing how test results affect other drugs have a statistically significant positive correlation with perceived competency providing pre- and post-test counselling, and answering patient’s queries for all three tests (Table 3). No statistically significant correlation was found between demographic characteristics and the confidence level of various activities.

Carbamazepine-HLA-B*15:02 scenario

An HLA-B*15:02 positive male patient with newly diagnosed focal onset seizure was presented and participants had to select an alternative anti-seizure medication. Participants were allowed to select more than one option. Levetiracetam was the most selected option (90.6%, n = 29), followed by sodium valproate (50.0%, n = 16), lamotrigine (31.3%, n = 10), and phenytoin (12.5%, n = 4). Both specialists (26%) and non-specialists (55%) chose lamotrigine or phenytoin as alternatives to carbamazepine in the HLA-B*15:02 positive patient. All respondents viewed pre-emptive HLA-B*15:02 genotyping as a useful tool to prevent severe adverse effects when choosing anti-seizure medications.

Clopidogrel-CYP2C19 scenario

A patient started on dual anti-platelet therapy (DAPT) consisting of clopidogrel and aspirin for secondary prevention of stroke was presented. Given her CYP2C19 *2/*2 genotype, 50% of respondents indicated they would switch to ticagrelor/aspirin for 3 weeks, then aspirin lifelong. A quarter however indicated they would continue clopidogrel/aspirin for 3 weeks, then aspirin lifelong, and 9.4% of respondents indicated switching to aspirin/dipyridamole. Overall, 96.9% of respondents viewed CYP2C19 genotyping useful to guide the choice of anti-platelet in secondary stroke prevention.

Siponimod-CYP2C9 scenario

The case involved a CYP2C9 *1/*1 individual recently started on siponimod. Respondents were least confident in navigating the use of siponimod (Table 2). Overall, 87.5% of respondents agreed that pharmacogenomic-guided therapy for specialty drugs such as siponimod can be safely implemented if clear guidance was available.

Information sources

Most agreed that pharmacogenomic test results were adequately visible in the electronic medical records (71.9%, n = 23). Almost all respondents agreed the information provided along with all three test results were adequate in providing guidance in the clinical decision-making. Product information leaflets and local practice guidelines were the two resources (Fig. 1A) that participants found most useful. International pharmacogenomic guidelines ranked fourth.

Perceived risks and barriers and training

Lack of clear clinical guidelines was the most important barrier to integrating pharmacogenomics into clinical practice, followed by lack of information technology (IT) infrastructure to integrate pharmacogenomic information systematically (Fig. 1B). Of respondents with experience ordering pharmacogenomic tests, 18.7% found it difficult to order them in their institutions.

Three-quarters of respondents did not receive formal training on pharmacogenomics, 12.5% were unsure and the rest felt training was inadequate. For desired training modalities, e-learning was top (71.9%, n = 23), followed by lecture (53.1%, n = 17), and workshop (15.6%, n = 5).

In this study, we surveyed a group of physicians in the neurology unit of two major hospitals on their opinions on pharmacogenomics. While physicians found pharmacogenomic testing was useful in the three clinical scenarios, they found that lack of clear clinical guidelines was the biggest barrier for implantation of pharmacogenomics. Physicians valued the product information leaflets and local practice guidelines as two most useful sources of information when making clinical decisions.

The case scenarios in our survey highlight the complexities amongst interpretation of pharmacogenomic guidelines. In the carbamazepine example, 55% of non-specialists and 26% of specialists chose phenytoin or lamotrigine despite a positive HLA-B*15:02 result(7). In the Asian context, the cost effectiveness of screening HLA-B*15:02 before prescribing carbamazepine has been established, while for lamotrigine and phenytoin, while positive patients are at an increased risk of Stevens-Johnson syndrome/toxic epidermal necrolysis, screening has not been shown to be cost effective. This has resulted in local guidelines requiring the testing of HLA-B*15:02 prior to the initiation of carbamazepine(6) while phenytoin and lamotrigine do not have similar guidelines. The large proportion of physicians who have chosen either phenytoin or lamotrigine reflects the complexities of interpreting guidelines. Similarly, a quarter of physicians continued aspirin and clopidogrel in a poor metaboliser of CYP2C19, reflecting the complexities of prescribing when there are no clear local guidelines despite the availability of international recommendations(5).

We found an increased perceived competency in interpreting test results, choosing an appropriate drug, and knowledge in drug-drug-gene-interactions were associated with increased perceived competency in providing pre- and post-test counselling and answering patient’s queries. We also found that despite all participants ordering pharmacogenomics tests, three quarters received no formal training to do so. A recent survey of American medical programs suggests that most do not have formal education programs in pharmacogenomics, a majority (85%) have at least 10 hours’ worth of didactic lectures on pharmacogenomics. However, despite this time, the majority of medical programs do not find that physicians have sufficient knowledge to address pharmacogenomics(8).

A limitation of this study was the non-response bias due to the low response rate. Some sub-groups (non-specialists and senior consultants) were not well-represented. This inherently limited the statistical tests performed. There were also limitations to using a self-administered online questionnaire like difficulty to elicit nuanced responses and tendency to leave open-ended questions blank. The strength is that this is the first study assessing the knowledge, attitudes, and concerns about pharmacogenomics within multiple subspecialties within neurology.

In this study, while pharmacogenomic testing was perceived as useful in physicians working in a neurology ward, lack of clear clinical guidelines was the biggest barrier for implantation of pharmacogenomics. Physicians valued the product information leaflets and local practice guidelines as two most useful sources of information when making clinical decisions.

Competing Interests:

The authors declare no competing financial interests.

Funding:

None

Author Contributions:

KN and PSW had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. JS analysed the data. KN drafted the manuscript. KN, PSW, JS, and JYS conceptualised the data and critically revised the manuscript.

Data availability statement:

The datasets generated during and/or analysed during the current study are not publicly available due to lack of consent during the waiver of consent application but are available from the corresponding author on reasonable request.

Cecchin E, Stocco G. Pharmacogenomics and personalized medicine. Genes (Basel). 2020;11(6).
Chan CYW, Chua BY, Subramaniam M, Suen ELK, Lee J. Clinicians’ perceptions of pharmacogenomics use in psychiatry. Pharmacogenomics. 2017;18(6).
Mrazek DA, Lerman C. Facilitating clinical implementation of pharmacogenomics. Vol. 306, JAMA. 2011.
Jameson A, Fylan B, Bristow GC, Sagoo GS, Dalton C, Cardno A, et al. What Are the Barriers and Enablers to the Implementation of Pharmacogenetic Testing in Mental Health Care Settings? Vol. 12, Frontiers in Genetics. 2021.
Administration F and D. Table of Pharmacogenetic Associations [Internet]. 2022 [cited 2023 Sep 3]. Available from: https://www.fda.gov/medical-devices/precision-medicine/table-pharmacogenetic-associations
Authority HS. HLA-B*1502 genotyping and carbamazepine-induced severe cutaneous adverse reactions [Internet]. Drug Safety Alerts. 2016 [cited 2023 Sep 3]. Available from: https://www.hsa.gov.sg/announcements/safety-alert/hla-b-1502-genotyping-and-carbamazepine-induced-severe-cutaneous-adverse-reactions
Zheng Z, Zhong H, Zhang Q, Huang Q, Wu H. HLA-B*1502 is associated with aromatic anticonvulsant drug-induced cutaneous adverse drug reactions among the Hakka population in China. J Int Med Res. 2020;48(3).
Basyouni D, Shatnawi A. Pharmacogenomics Instruction Depth, Extent, and Perception in US Medical Curricula. J Med Educ Curric Dev. 2020;7.

Table 1. Baseline demographic of respondents (n = 32).

Characteristics	Respondents, n (%)
Gender
Male	24 (75.0)
Female	8 (25.0)
Years of experience
1 – 5	5 (15.6)
6 – 10	10 (31.2)
11 – 20	14 (43.8)
> 20	3 (9.4)
Job Title
Non-specialists	10 (32.3)
Specialists¹	22 (68.7)
Experience ordering PGx tests
Yes	32 (100.0)
No	0 (0.0)
PGx tests ordered/requested in last one year
< 5	1 (3.1)
5 – 20	13 (40.6)
> 20	18 (56.3)

PGx Pharmacogenomics.

Table 2. Median [interquartile range] confidence levels performing various activities stratified by pharmacogenomic test.

Activities	*HLA-B15:02**	*CYP2C19*	*CYP2C9*
Deciding when to order the pharmacogenomic test	5 (4 – 5)	4 (4 – 5)	2 (1 – 3)
Interpreting test result	5 (4 – 5)	4 (4 – 5)	2 (2 – 3)
Providing pre- and post-test counselling	4 (3 – 5)	4 (3 – 5)	2 (1 – 3)
Choosing the appropriate drug^a in view of test result	5 (4 – 5)	4 (4 – 5)	NA^b
Knowing how test result affects patient’s other drugs	4 (3 – 5)	3 (2 – 4)	2 (1 – 2)
Answering patient’s queries	4 (3 – 5)	4 (3 – 4.75)	2 (1 – 3)

Measurement made using 5-point Likert scale with 1 being the least confident and 5 being the most confident. NA not applicable.

^aRefers to anti-seizure drug for HLA-B*15:02 test and anti-platelet drug for CYP2C19 test.

^bNot applicable as the siponimod-CYP2C9 involves a genotype-guided dosing change rather than a change in drug selection

Table 2. Spearman correlational analysis of factors including perceived competency interpreting genotype test result, knowing how test results affect patient’s other drugs, providing pre- and post-test counselling, and answering patient’s queries (n = 32).

Activities	Providing pre- and post-test counselling		Answering patient’s queries
	Correlation coefficient	P value	Correlation coefficient	P value
Deciding when to order test
HLA-B*15:02	0.255	0.158	0.406	0.021
CYP2C19	0.461	0.008	0.541	0.001
CYP2C9	0.746	<0.001	0.769	<0.001
Interpreting test result
HLA-B*15:02	0.466	0.007	0.590	<0.001
CYP2C19	0.589	< 0.001	0.557	<0.001
CYP2C9	0.707	<0.001	0.719	<0.001
Choosing appropriate drug in view of test result
HLA-B*15:02	0.601	<0.001	0.604	<0.001
CYP2C19	0.491	0.004	0.641	<0.001
CYP2C9^a	NA	NA	NA	NA
Knowing how test result affects patient’s other drugs
HLA-B*15:02	0.670	<0.001	0.705	<0.001
CYP2C19	0.504	0.003	0.708	<0.001
CYP2C9	0.880	<0.001	0.918	<0.001

NA not applicable.

^aNot applicable as the siponimod-CYP2C9 involves a genotype-guided dosing change rather than a change in drug selection

There is NO conflict of interest to disclose.

SupplementaryTablesandFigures20230518formatPGx.docx

Pharmacogenomics knowledge within Neurology: A brief survey amongst physicians

Status:

Version 1

Abstract

Figures

Introduction

Subjects and Methods

Study design and participants

Questionnaire

Statistical analysis

Results

Perceived attitudes and competency

Carbamazepine-HLA-B*15:02 scenario

Clopidogrel-CYP2C19 scenario

Siponimod-CYP2C9 scenario

Information sources

Perceived risks and barriers and training

Discussion

Conclusion

Declarations

Competing Interests:

Funding:

Author Contributions:

Data availability statement:

References

Tables

Additional Declarations

Supplementary Files

Status:

Version 1