Impact of First‐of‐Its‐Kind Patient‐Facing Pharmacogenetics Tool on Dosing Decisions and Treatment Outcomes

Abstract

Germline pharmacogenetics (PGx) is increasingly used to tailor medication selection/dosing. However, existing systems primarily communicate PGx results to providers, limiting direct patient engagement. To address this, we developed YourPGx Oncology, an innovative patient‐facing portal that delivers multi‐gene PGx results (CYP2D6, UGT1A1, DPYD) through 33 unique, patient‐friendly summaries. The utility of this tool was prospectively evaluated in an oncology population, where these pharmacogenes impact high‐stakes treatments. Patients enrolled in the PhOCus study (NCT04541381) participated in single‐session evaluations of the tool in‐person or via videoconference, with a pharmacist available for questions and administering pre‐ and post‐surveys that assessed educational impact. Each patient viewed their own previously obtained PGx results. Of 190 eligible patients, 70 responded to solicitations via email, phone, and in‐person, of whom 51 (73%) completed an observed session and completed surveys. Patients spent a median of 13.4 minutes (range 8.1–21.0) navigating YourPGx Oncology. After portal interaction, patients' ability to identify individual efficacy and safety estimates for chemotherapies and pain medications significantly improved, with the proportion accurately recognizing PGx‐informed drug efficacy likelihoods rising from 32% to 72% (Odds Ratio [OR] = 5.8 for the shift from discordant to concordant efficacy knowledge, P < 0.001), and PGx‐related toxicity recognition increasing from 31% to 57% (OR = 3.2, P = 0.01). Our findings show that a customized patient‐facing PGx results portal enhances patient understanding of individual medication efficacy and toxicity likelihoods, highlighting the potential key role of direct‐to‐patient PGx tools to facilitate optimized treatment‐informed care and promote genetically guided shared decision‐making.

Study Highlights.

• WHAT IS THE CURRENT KNOWLEDGE ON THE TOPIC?

While PGx is increasingly integrated into prescribing, most existing systems primarily communicate results to healthcare providers, with limited patient‐facing tools available to support patient understanding and engagement with PGx information.

• WHAT QUESTION DID THIS STUDY ADDRESS?

This study evaluated whether a patient‐facing PGx results portal could enhance patients' comprehension of their personal PGx results and improve perceptions of individualized medication risk and efficacy.

• WHAT DOES THIS STUDY ADD TO OUR KNOWLEDGE?

This study demonstrates that complex PGx information can be effectively communicated directly to patients through an interactive, customized portal. Portal interaction significantly improved patients' understanding of drug efficacy and toxicity risks, and increased interest in PGx‐informed prescribing.

• HOW MIGHT THIS CHANGE CLINICAL PHARMACOLOGY OR TRANSLATIONAL SCIENCE?

This study is the first to show that direct‐to‐patient PGx tools with multi‐gene/drug information can facilitate optimized treatment‐informed care and promote genetically guided shared decision‐making among patients.

PGx has emerged as a promising approach to personalize treatments by leveraging genetic information to optimize drug selection and dosing. ¹ , ² With advances in health information technology, clinical decision support (CDS) tools have been implemented in some settings to integrate PGx guidance into clinical workflows, ³ , ⁴ , ⁵ allowing providers to consider genotyping results when making medication decisions. While cost and timing remain key factors for the effective use of these provider‐facing electronic medical record tools, there remains a notable lack of tailored resources to help patients receive and understand their own PGx results, highlighting an important gap in patient‐centered care. ⁶ , ⁷

Recent efforts are increasingly underway to deliver PGx results directly to patients. Some academic centers have developed patient education materials to explain PGx both before and after results are provided, as well as offering consultation services to improve patient health literacy regarding PGx. ⁸ , ⁹ , ¹⁰ , ¹¹ As specific examples, Sloat et al. evaluated the impact of pre‐test education on patient PGx knowledge and perceptions, demonstrating that a pre‐test video significantly enhanced understanding of PGx testing. ¹² Haga et al. provided valuable insights by conducting pre‐ and post‐test assessments following PGx testing and provider discussions, prospectively demonstrating increased patient confidence in PGx‐guided treatment decisions and reduced concerns about medications. ¹³ Olson et al. assessed patient understanding of CYP2D6 PGx results in the Right Drug, Right Dose, Right Time (RIGHT) study that utilized the Mayo Clinic Patient Portal, revealing comprehension gaps and the need for better education and support tools to boost confidence in medication decisions. ¹⁴ The Mayo group also examined general patient attitudes toward PGx testing as a pilot of the RIGHT study by assessing multimedia educational materials, including brochures and videos, using “prototype” PGx reports, and emphasized the importance of tailoring these resources to address misconceptions and improve comprehension. ¹⁵ Lemke et al. explored patient experiences with PGx result return using a process by which patients were given (or sent) a hard‐copy report, finding that while most patients viewed PGx testing as beneficial, concerns about privacy and discrimination highlighted the need for clearer communication and improved patient education. ¹⁶ Our own group also extensively previously investigated patient preferences for PGx result delivery, specifically focusing on discerning effective elements of a hypothetical interactive patient “PGx portal.” Those studies found that patients value clear, concise, and trustworthy information presented with user‐friendly graphics using layperson language. ¹⁷ , ¹⁸ The current work is the initial prospective evaluation of the realization of those design efforts, ¹⁹ in which we now tested a fully functioning patient portal (termed “YourPGx”) containing personalized, patient‐specific PGx results during clinical care.

This study represents the first prospective evaluation of a first‐of‐its‐kind interactive, dedicated PGx patient portal containing broad PGx results informing multiple drug‐gene pairs during the clinical receipt of such medications. We focused this initial study on an oncology population, given the potential relevance of several pharmacogenes to the potentially high‐stakes treatment choices in this setting, and given that oncology is one of the fields where genomic information generally has seen the earliest advanced adoption. The present study was conducted within the framework of the PhOCus trial (NCT04541381) to evaluate patients' comprehension of their personal PGx results spanning multiple genes (CYP2D6, DPYD, and UGT1A1) and testing perceptions and knowledge of personalized medicine using YourPGx Oncology.

METHODS

Study participants

This study engaged patients participating in the ongoing PhOCus trial (NCT04541381), a randomized, prospective study investigating the effects of preemptive PGx testing on chemotherapy dosing decisions. ²⁰ Eligible participants for this sub‐study were adult oncology patients enrolled in the genotyping arm of this trial. Patients could participate if they were receiving care at any of the four academic and community‐based hospitals across Illinois where PhOCus was being conducted. Only individuals whose PGx results had already been released to their treating providers were approached for participation. Patients were identified from a list provided by the unblinded data scientist and were invited by a pharmacist using a combination of telephone calls, emails, and in‐person solicitation during scheduled clinic visits.

Study design

We employed a single‐session evaluation design to assess patient understanding and perceptions of personalized cancer care after interacting with YourPGx Oncology. Sessions were conducted either in‐person using a tablet or virtually via Zoom and were observed by a PGx‐trained pharmacist member of the research team. Before the start of the session, participants completed a pre‐survey assessing their baseline knowledge of PGx, perceptions of personalized care, and perceived health status. After completing the pre‐survey, participants were given access to YourPGx Oncology, which included general educational content on PGx and its role in personalizing prescriptions and, importantly, their own PGx results for medications potentially impacted by three key pharmacogenes (CYP2D6, DPYD, and UGT1A1). After a brief introduction of the portal by the session organizer, participants were left to independently browse the information (with the pharmacist available to answer questions). After each patient finished navigating the portal, a post‐survey was administered to evaluate changes in knowledge and perceptions.

It is important to note that for all patients, since their patient‐specific PGx results had already been made available to their treating provider team (in some cases for many weeks already), it is possible that some patients may have known about their PGx results prior to their patient‐only session. This was entirely permissible, as providers in the PhOCus study are free to share any and all PGx information with patients during routine care interactions, at the treating providers' discretion. The specific intervention of this sub‐study, thus, is that patients were now, for the first time, provided direct access to seek and examine their own PGx results using the YourPGx Oncology portal.

Design of YourPGx Oncology

The YourPGx Oncology portal builds on the conceptual model established by the YourPGx mock portal, wherein patient‐facing summaries map one‐to‐one with provider‐facing summaries that are available in the electronic medical record. ¹⁸ With the core technological infrastructure to link these summaries in place, the key focus was to develop a user interface that effectively communicated the relevance of PGx to the high‐stakes drug toxicity and efficacy outcomes that could be encountered in this study during oncology care. Images were developed to illustrate major toxicities, risk levels, and drug efficacy, supported by enhanced designs for the PGx educational materials. Images were paired with lay‐friendly language to provide intuitive explanations for how a patient's genetic phenotype may affect drug metabolism (Figure 1 ). Both patient‐friendly text and image asset creation were iteratively refined through several rounds of feedback from clinicians, hospital personnel, and laypersons. Individual summaries were delivered through a pop‐out drawer component that would appear upon clicking read more on the summary list item. The patient summary page also included a button to export summaries into a print‐friendly PDF file. YourPGx Oncology was developed as a Node.js‐driven server rendered multi‐page application. The existing database schema from the mock portal was maintained, with additions being a table to enumerate drug toxicities and another to link toxicities to patient‐facing summary records in a many‐to‐one fashion. Core backend functionality was architected using a model‐view‐controller‐like pattern, with a service layer that facilitates the retrieval of patient demographics, medications, and summaries. A hybrid controller layer sits atop the service layer, providing two groups of endpoints: ones that consume data from the services and return HTML templates and ones that offer a standard RESTful interface for create, read, update, and delete (CRUD) operations on patient‐facing summaries. Endpoints in the YourPGx Oncology backend communicate with an intermediate logging service to capture relevant user activity. To enhance data comprehensiveness, we implemented backend tracking of user interactions with the YourPGx Oncology midway through the study. This feature, applied to the final 22 participants (43% of the cohort), allowed us to measure time spent navigating the tool and identify the most frequently accessed sections. All data was stored in an encrypted SQL Server database housed at the HIPAA compliant Kenwood Datacenter, managed by the University of Chicago Center for Research Informatics.

Figure 1.

Example screenshot from the YourPGx Oncology portal showing representative PGx results and information for DPYD and CYP2D6 for patients receiving 5‐fluorouracil and tramadol.

Data collection

Data collection was facilitated through pre‐ and post‐surveys administered before and after participants used YourPGx Oncology. The survey instruments, which were previously tested and validated, utilized a combination of multiple‐choice questions and Likert scales to capture participants' responses. ¹⁴ , ¹⁸ , ²¹ All data, including survey responses and user interaction metrics, were securely collected and managed using REDCap (Research Electronic Data Capture), a secure, web‐based software platform designed for research data capture.

Statistical analyses

Patient‐reported knowledge, attitudes, outcome expectancy, and perceptions of care were assessed through survey responses. The response data was extracted by the study data scientist from the REDCap database, after which it was systematically organized and categorized for relevant question answers. The primary statistical method employed was McNemar's test, which was used to analyze paired nominal data from pre‐ and post‐intervention surveys. This test was chosen to assess shifts in paired categorical responses. Key variables analyzed included changes in congruent versus incongruent answers as informed by CYP2D6, UGT1A1, and DPYD, as well as shifts in levels of agreement or disagreement on Likert‐scale questions regarding patient attitudes toward personalized medicine and preferences for PGx‐informed decision‐making.

To facilitate the use of McNemar's test for Likert‐scale questions, a specific categorization approach was applied. Responses of “strongly agree” were compared against the combined responses of “agree somewhat,” “not sure,” “somewhat disagree,” and “strongly disagree.” This collapsing of response options created a binary outcome, enabling the detection of significant shifts toward stronger agreement following patient interaction with the YourPGx Oncology portal. For all statistical analyses, a two‐sided testing approach was used, with P < 0.05 considered statistically significant. Odds ratios (ORs) were calculated where applicable to estimate effect sizes. All analyses were performed using GraphPad Prism software.

RESULTS

Patient demographics

From a pool of 190 eligible cancer patients, 70 responded to various recruitment methods including email, phone calls, and face‐to‐face invitations. Of these, 51 individuals (72.9%) enrolled in the sub‐study evaluating the patient‐centered PGx portal. The participant demographics were: 30 males and 21 females; 32 White, 12 Black, and 7 from other racial backgrounds (Table 1 ). The age of participants ranged from 33 to 84 years, with a mean age of 59.3. Genotypic analysis revealed expected variability in PGx metabolizer status among participants. For UGT1A1, 16 (39.0%) were classified as normal metabolizers (NMs), 21 (51.2%) as IMs, and 4 (9.8%) as PMs, while 10 were not genotyped. For DPYD, 46 (95.8%) were NMs, 2 (4.2%) were IMs, and no PMs were identified, while 3 had an unknown status. Regarding CYP2D6, 31 (63.3%) were NMs, 14 (28.6%) were IMs, 1 (2%) was a PM, and 1 (2%) was an UM; 2 (4.1%) were indeterminate, with 2 having an unknown status. Prior to accessing the PGx portal, the majority of patients reported satisfaction with their healthcare provider, with 33 (64.7%) very satisfied and 18 (35.3%) somewhat satisfied. However, their self‐reported health status varied considerably, likely reflecting the challenges of living with cancer. When asked to rate their health, 1 (2%) described it as excellent, 9 (17.6%) as very good, 23 (45.1%) as good, 15 (29.4%) as fair, and 3 (5.9%) as poor.

Table 1.

Demographics

Participant characteristics (n = 51)
Sex, N (%)
Male	30 (58.8%)
Female	21 (41.2%)
Race, N (%)
White	32 (62.8%)
Black or African American	12 (23.5%)
More Than One Race	2 (3.9%)
Other	5 (9.8%)
Education, N (%)
High School or Less	12 (23.5%)
Some College	11 (21.6%)
College/Univ. Grad	15 (29.4%)
Advanced Degree	6 (11.8%)
Unknown	5 (9.8%)
Age
Average	59.3
Range	33–84
Cancer Types, N (%)
Gastrointestinal	41 (80.4%)
Head & Neck	4 (7.8%)
Others	6 (11.8%)
UGT1A1 phenotype, N (%)a
Normal metabolizer	16 (39.0%)
Intermediate metabolizer	21 (51.2)
Poor metabolizer	4 (9.8%)
DPYD phenotype, N (%)a
Normal metabolizer	46 (95.8%)
Intermediate metabolizer	2 (4.2%)
Poor metabolizer	0 (0%)
CYP2D6 phenotype, N (%)a
Normal metabolizer	31 (63.3%)
Intermediate metabolizer	14 (28.6%)
Poor metabolizer	1 (2%)
Ultra‐rapid metabolizer	1 (2%)
Indeterminate	2 (4.1%)

^a 10 patients for UGT1A1, 3 patients for DPYD, and 2 patients for CYP2D6 were not genotyped or had their status reported as unknown due to missing variant(s).

Improved patient understanding of PGx and perceptions of personalized medicine

Prior to exploring the portal, 62.8% of patients reported no knowledge of PGx. After using the portal, 78.4% of patients correctly defined the concept, with only 13.7% still reporting uncertainty. Concurrently, patients' perception of their healthcare providers incorporating “personalized medicine” into their treatment increased significantly, with those strongly agreeing rising from 37.3% to 68.6% (Figure 2 a ). Among 18 patients who changed their responses to this question before and after using the portal, 17 shifted toward stronger agreement, while only 1 moved away from agreement (χ ² = 12.5, two‐tailed P < 0.001).

Figure 2.

(a) Proportion plot illustrating the shift in survey responses before (Pre‐Survey) and after (Post‐Survey) using the PGx portal. Participants were asked to rate their agreement with the statement: “My healthcare providers incorporate ‘personalized medicine’ into my treatments.” The difference in bar lengths on the left and right sides of the chart represents changes in the distribution of respondents across the Likert‐scale options (Strongly Agree, Somewhat Agree, Somewhat Disagree, Not Sure). (b) Proportion plot illustrating the shift in survey responses before (Pre‐Survey) and after (Post‐Survey) using the PGx portal. Participants were asked to rate their agreement with the statement: “Based on my current knowledge, I wish my doctor had pharmacogenomic information before prescribing a new medication for me.” The difference in bar lengths on the left and right sides of the chart represents changes in the distribution of respondents across the Likert‐scale options before and after using the PGx portal (Strongly Agree, Somewhat Agree, Somewhat Disagree, Not Sure).

Increased desire for PGx‐informed decision‐making

Before exploring the portal, participants showed mixed responses regarding the perceived utility of the incorporation of PGx information for new medications. After portal use, there was a significant shift toward stronger agreement. The proportion of patients strongly agreeing with incorporating PGx information in medication decisions increased from 43.1% to 74.5% (Figure 2 b ), with 17 of 18 participants who changed their responses intensifying their agreement to the strongest level, contrasted by only one participant who lessened their agreement (χ ² = 12.5, two‐tailed P < 0.001).

Enhanced awareness of medication efficacy

We next tested specific awareness of personal likelihood of medication efficacy. For this analysis, we focused the analysis on patient responses about perceived efficacy of specific opioids to treat pain, given the high prevalence of this condition among cancer patients, ²² using specific information provided in the personalized portal about various opioids' chances of effectiveness as predicted by individual CYP2D6 results. The analysis included patients with available CYP2D6 data (n = 47). “Correct” (or concordant) responses were defined as those that aligned with the expected clinical outcome based on CYP2D6 metabolizer status. For NMs (n = 31), the concordant response was that tramadol, hydrocodone, and codeine would be likely to be “very effective.” For IMs (n = 14), the concordant response was “somewhat effective.” For PM (n = 1), “ineffective” was considered correct. For UM (n = 1), the concordant response was “effective”. Across all metabolizer types, patients also had the opportunity to answer “I can't remember.”

Prior to portal interaction (that is, without the availability of direct access to PGx results), 31.9% (15/47) of patients characterized their personal potential for receiving benefit from opioids as congruent with their PGx profile, while 46.8% (22/47) answered that they did not know (Figure 3 a ). Post‐portal interaction, congruent responses increased to 72.3% (34/47), while incorrect responses were 23.4% (11/47), and only 4.3% (2/47) remained unsure. A comparison of patients with divergent responses (defined as those who changed their responses from pre‐ to post‐accession of the portal) revealed a significant improvement in recognition (Odds Ratio [OR] = 5.8 for the shift from discordant to concordant efficacy knowledge, χ ² = 12, P < 0.001), with 85.2% of these patients (23/27) demonstrating improved understanding (Pre‐survey Incorrect → Post‐survey Correct) (Figure 3 b ).

Figure 3.

(a) Proportional shift in patient‐perceived opioid efficacy before and after using the PGx portal, based on responses to the question: “Based on my pharmacogenomic results, there are medications I may be prescribed that could potentially…” with the options: Very effective, Somewhat effective, Ineffective, or I can't remember. (b) Patient level metabolizer status and knowledge concordance for opioids efficacy before and after interacting with the PGx portal. Each icon represents an individual patient. Columns represent patient survey responses before (Pre) and after (Post) interacting with the PGx portal. Icon colors denote metabolizer phenotypes: Red = Poor Metabolizer (PM), Yellow = Intermediate Metabolizer (IM), Green = Normal Metabolizer (NM), Ultra Metabolizer (UM). A checkmark (✔) indicates a correct response, an “X” denotes an incorrect response, and a question mark (?) signifies “I can't remember”.

Enhanced awareness of medication side effects

The assessment of medication toxicity risk awareness was based on patients' understanding of potential adverse effects of their chemotherapy regimen (as informed by DPYD and UGT1A1 status), and/or from opioid‐related toxicities (informed by CYP2D6). Responses were deemed congruent based on specific criteria aligned with metabolizer status: NM for both DPYD and UGT1A1 genes were expected to identify their medications as “generally safe”; IM for UGT1A1 (n = 21) as having “somewhat increased risk of side effects”; IM for DPYD (n = 2) and PM for either gene (n = 4) as having “dangerously increased risk of side effects”; and the single patient with UM status for CYP2D6 as having either “somewhat” or “dangerously increased risk of side effects”. The response “I can't remember” was available for all scenarios.

Prior to portal interaction, 31.4% (16/51) of patients characterized their personal potential for medication toxicity from chemotherapy regimen and/or from opioids as congruent with their PGx profile, while 49.0% (25/51) answered that they did not know (Figure 4 a ). Post‐portal interaction, congruent responses increased to 56.8% (29/51), while incorrect responses were 31.4% (16/51), and only 11.8% (6/51) remained unsure. A comparison of patients with divergent responses revealed a significant improvement in recognition (Odds Ratio [OR] = 3.2 for the shift from discordant to concordant efficacy knowledge, χ ² = 5.76, P = 0.01), with 76.0% of these patients (19/25) demonstrating improved understanding (Pre‐survey Incorrect → Post‐survey Correct) (Figure 4 b ).

Figure 4.

(a) Proportional shift in patient‐perceived safety of their chemotherapy regimen and/or opioid‐related toxicities before and after using the PGx portal, based on responses to the question: “Based on my pharmacogenomic results, there are medications I may be prescribed that could potentially…” with the options: “Have dangerously increased risk of side effects,” “Have somewhat increased risk of side effects,” “Are generally safe for me,” or “I can't remember.” (b) Patient level metabolizer status and knowledge concordance for chemotherapy regimen and/or from opioid‐related toxicities before and after interacting with the PGx portal. Columns represent patient survey responses before (Pre) and after (Post) interacting with the PGx portal. Icon colors denote metabolizer phenotypes: Red = Poor Metabolizer (PM), Yellow = Intermediate Metabolizer (IM), Green = Normal Metabolizer (NM). A checkmark (✔) indicates a correct response, an “X” denotes an incorrect response, and a question mark (?) signifies “I can't remember.”

User experience and engagement

Despite the wide age range and varying education levels, most participants found the portal easy to use and straightforward. Users spent a median of 13.4 minutes (range: 8.1–21.0 minutes) navigating YourPGx Oncology, with the results section receiving the most attention. The color‐coded risk information, particularly in yellow and red, effectively captured users' focus. Patients found the depiction of risk and efficacy through color gradients informative, and many valued the option to print PDF results for future reference and provider discussions. Patients repeatedly said that the use of facial expression icons to summarize results on a three‐step scale, complemented by pharmacist explanations, enhanced comprehension.

DISCUSSION

This study is the first to prospectively demonstrate the utility of directly delivering multi‐gene PGx results to oncology patients who were actively receiving care with the medications being disclosed through an interactive, direct‐access patient‐facing portal. Although prior efforts to provide PGx results directly to patients have shown promise in improving health literacy and engagement, none have prospectively evaluated a structured tool specifically designed for patients. ¹² , ¹⁴ , ¹⁵ , ¹⁶ Our study addressed this gap by integrating multiple PGx results within an interactive, intuitive lay framework, addressing the complexities of treatment and its multifaceted therapeutic considerations. In this prospective testing within the context of clinical care, using an interface that was informed by direct patient input and feedback, we observed significant improvements in patient comprehension, perceptions of PGx, and understanding of medication responses. Collectively, these findings underscore a promising pathway toward optimizing treatment‐informed care and promoting genetically guided shared decision‐making.

The recognition of effective communication in guiding patients through their therapeutic journey has grown, underscoring its importance in delivering high‐quality care and the central role of shared decision‐making in clinical practice. ²³ , ²⁴ Meanwhile, technological advancements continue to reshape healthcare, enhancing patient engagement and transforming the way medicine is practiced. ²⁵ , ²⁶ Applications like MyChart have revolutionized patient engagement, improving access to health information and enhancing communication with providers. ²⁷ , ²⁸ As a result, more patient‐facing tools are emerging, not only to streamline administrative tasks like appointment scheduling but also to support more complex aspects of care, such as treatment management and personalized medicine. ²⁹ , ³⁰ In particular, the integration of genomics into digital health is expanding the role of these tools, making access to personalized genetic information more critical than ever. ³¹ Several biotechnology companies now offer patient‐oriented PGx reports that incorporate results based on CPIC guidelines, FDA labeling, and proprietary algorithms. ³² , ³³ , ³⁴ These reports often share similar design features such as visually intuitive formats and color‐coded indicators for metabolizer status. YourPGx Oncology aligns with these trends but was uniquely developed to emphasize interactivity within the portal environment, in contrast to the nearly universal “static reports” offered by most commercial vendors. YourPGx also features structured, user‐friendly layouts, with simplified language, enhanced by visual aids such as toxicity icons and a facial expression rating scale to further support patient comprehension.

The recent 2025 FDA‐AACR Workshop debating the potential role of widespread DPD deficiency testing prior to the use of fluoropyrimidines and the subsequent FDA communication emphasizing the need for discussion with patients about a shared decision around pursuing formal testing underscores the critical importance of patient communication regarding high‐stakes medication toxicities. ³⁵ Our study demonstrates that complex PGx information can be effectively communicated to patients in a digestible format, with nearly 60% of study participants correctly understanding the potential side effects of their potential chemotherapeutic regimen, and approximately three‐quarters able to recall efficacy likelihood for specific opioids to treat cancer pain. Notably, the proportion of patients unable to recall side effects and efficacy also decreased by nearly 40% overall. Taken together, these findings suggest that our PGx tool equipped patients with a better understanding of their genetic profile and implications for treatment, which may potentially foster more informed discussions between patients and providers toward more personalized and shared decision‐making. This approach aligns with the growing emphasis on shared decision‐making in healthcare and could contribute to improved patient engagement throughout their treatment journey. ³⁶

There were several limitations to this study. Although the large majority (73%) of individuals who responded to outreach efforts ultimately completed the sub‐study, the overall response rate from the total eligible population was modest, raising the possibility of selection bias. However, the demographic characteristics of those who completed the sub‐study were comparable to the total eligible PhOCus population in terms of sex, age, race, and educational level. Specifically, the educational attainment profile of those completing the sub‐study was consistent with data from the US Department of Education National Center for Education Statistics, supporting the representativeness of our sample. ³⁷ Additionally, at the time of the portal accession evaluation, some patients had already completed their chemotherapy regimen, whereas others were not actively taking any of the pain medications for which results were displayed in the portal, potentially affecting recall. While efforts were made for all patients to provide in‐person access to the portal for facilitated (observed) sessions, most evaluations (>85%) were conducted via videoconference (Zoom) to minimize the need for additional visits to the medical center. This difference in format may have influenced patient engagement, as in‐person participants interacted directly with the portal on a tablet in the physical presence of a pharmacist, whereas remote participants navigated it using Zoom's screen control feature. While it is difficult to make a direct comparison between in‐person and online groups given the relatively small number of the former sessions, the results were generally comparable. Also, although sessions followed a standardized format, where participants independently explored the portal after a brief introduction and a pharmacist addressed questions and clarified information, variability in information delivery may have occurred based on individual engagement levels, potentially resulting in more comprehensive discussions for those who sought additional clarification. The most frequent questions asked involved meaning of gene and phenotype annotations, their role in influencing medication effects, and whether the same information was available to providers—the answers to which could have impacted participants' understanding and potentially influenced the results. However, our study analysis focused on responses systematically collected through the patient survey, and the type and frequency of participant questions were not formally recorded or analyzed. Separately, while our study focused on the initial reporting of three genes especially relevant to oncology care, we are now expanding the portal to encompass the reporting of additional genes for broader impact on patients within other disease contexts. We see no reason why patients could not similarly be educated by, and benefit from, accessing their own PGx results in those contexts to better understand medication efficacy and safety. Lastly, as previously mentioned, backend tracking of user interactions was implemented later in the study and was only available for the final 22 participants, potentially limiting insights into earlier user behaviors.

In conclusion, our study demonstrates that interaction with the multi‐gene patient‐facing PGx portal significantly enhances patients' understanding of drug efficacy and toxicity risks while increasing interest and perception of PGx‐informed prescribing. The observed utility of patient‐facing PGx tools, along with heightened patient engagement, highlights their potential to enhance shared decision‐making in clinical care. Whether these improvements in patient understanding and engagement persist over time (for example, 3 to 6 months) remains an important question. The broader PhOCus trial, which includes follow‐up through 48 weeks and captures both quality of life measures and clinical endpoints, will provide an opportunity to assess the durability of these effects.

FUNDING

This work was supported by NIH 1R01HG012273–01 (P. H. O'Donnell as PI), the University of Chicago Committee on Clinical Pharmacology and Pharmacogenomics T32 training grant NIH 5T32GM007019 (Y. Cho as trainee), and the Benjamin McAllister Research Fellowship Award (Y. Cho as trainee).

CONFLICT OF INTEREST

Dr Mark Ratain is a coinventor on patents related to pharmacogenetic diagnostics and receives royalties related to UGT1A1 genotyping that were not connected to the genotyping performed in this research. Dr Peter O'Donnell reports personal fees from O'Brien and Ryan LLP for consultation regarding pharmacogenomics in litigation, has received honoraria from ISMIE for educational talks about pharmacogenomics, and is a compensated member of the data safety monitoring board of the NIH IGNITE II network, all outside the submitted work. All other authors declared no competing interests. As an Associate Editor for Clinical Pharmacology & Therapeutics, Peter O’Donnell was not involved in the review or decision process for this paper.

AUTHOR CONTRIBUTIONS

Y. Cho wrote the manuscript, designed the research, performed the research, analyzed the data, and contributed new reagents/analytical tools; S. Elahi wrote the manuscript, designed the research, and analyzed the data; M. Jack designed the research and analyzed the data; J. Campbell contributed new reagents/analytical tools; E. Smith contributed new reagents/analytical tools; T. Karrison designed the research; R.W. Knoebel designed the research; D. George contributed new reagents/analytical tools; L. House contributed new reagents/analytical tools; K.J. Yeo contributed new reagents/analytical tools; S.L. Volchenboum designed the research; R.Z. Szmulewitz designed the research; D.O. Meltzer designed the research; E.E. Vokes designed the research; M.J. Ratain designed the research and contributed new reagents/analytical tools; P.H. O'Donnell wrote the manuscript, designed the research, performed the research, analyzed the data, and contributed new reagents/analytical tools.