Challenges of Incorporating Pharmacogenomics Into Clinical Practice

In this issue of GCR, Yalcin reviews current knowledge of pharmacogenomic markers for personalized therapy selection in gastrointestinal cancer.¹ Pharmacogenomics holds the promise to improve patients’ response to drugs, reduce adverse events, and reduce health care costs.² In recent years, the US Food and Drug Administration (FDA) has recommended pharmacogenomic testing for a variety of medications,³ including UGT1A1*28 testing, a predictor of irinotecan toxicity.¹ Other pharmacogenomic markers for gastrointestinal cancer therapy are not yet being used routinely, as they still require validation and FDA approval.

The implementation of pharmacogenomic testing in the clinic still faces several challenges, including the following:

• The need for validated markers that have received administrative approval

• Availability of approved assays

• Access to approved assays through Good Laboratory Practice (GLP) or regulated laboratories

• Availability of dosing algorithms that take into account pharmacogenomic tests

• An effective delivery system to transfer test results and dosing recommendations to the clinic (and back again for the fine tuning of algorithms based on postmarket surveillance)

VALIDATED MARKERS

The literature contains a plethora of pharmacogenomics data and the majority of gastrointestinal cancer medications have been studied in a pharmacogenomics context. ^1,4,5 However, the majority of significant findings are often seen in small sample sets and lack statistical validation in large, well-defined studies. Consequently, markers that receive approval for clinical use from regulatory agencies such as the FDA are few in number. This is changing with a greater understanding of how to perform pharmacogenomics studies, along with a greater depth of knowledge of the human genome sequence, its associated genetic variation, and the range of fast, ever-improving, and accurate genotyping technologies available on the market today.

Paradoxically, growing knowledge in the field also generates more questions and uncertainty. Even with robust clinical pharmacogenomic markers such as UGT1A1*28 for irinotecan, the resulting clinical guidance is not yet clear-cut. For instance, not every patient homozygous for UGT1A1*28 will experience irinotecan toxicity, and patients without UGT1A1*28 alleles could still suffer adverse events. Other polymorphisms in UGT1A1, particularly in populations where UGT1A1*28 occurs at a low frequency, may also need to be screened prior to irinotecan therapy. In addition, polymorphisms in other UGT1A gene family members, as well as other genes involved in irinotecan transport, metabolism, and pharmacodynamics could also play a significant role in predicting toxicities.^6,7

Finally, some pharmacogenomics markers, though significantly associated with adverse events, may not occur frequently enough within the general population to justify the expense of screening at this time, as illustrated by DPYD*2A, a predictor of severe toxicity from 5-fluorouracil (5-FU) therapy. This variant occurs at a frequency of less than 1% in the general population and is thus unlikely to be routinely screened for in patients.⁵ DPYD expression or enzyme activity represent more likely testing scenarios for patients prior to receiving 5-FU or its analogs. Pharmacoeconomics will play a major role in these cases by subjecting a pharmacogenomics test to cost-utility analysis.

APPROVED ASSAYS

Once pharmacogenomic markers have received approval for incorporation into drug package labeling we need tests approved for in-vitro diagnostics. Many companies offer tests and kits for research use only; regulatory approval following strict guidelines is required before tests become available for clinical use.

ACCESS TO ASSAYS

A range of options are available to the clinician, depending on local resources. Some pharmacogenomic tests can be run locally with the purchase of the relevant kits and the availability of specialized genotyping equipment and trained personnel on site. The majority of approved pharmacogenomic assays can also be run at centralized laboratories and through the manufacturers, such as Labcorp (www.labcorp.com/wps/portal) and Third Wave Technologies (www.twt.com), for example. Knowledge of the availability of these tests and how to access them is essential for successful implementation of pharmacogenomic testing into clinical practice.

Of increasing concern is the widespread availability and advertising of Direct To Consumer (DTC) testing, where patients can send their own samples to be assessed for a range of genetic polymorphisms and submit these data to their physicians. Often, no clear clinical utility of the polymorphisms tested has been established, and the physicians have no specific criteria for how to handle the data.⁸ The pros and cons of DTC testing have generated considerable debate,^9,10 and the need for strict regulatory guidelines is essential.

DOSING ALGORITHMS

One of the major limitations for implementing pharmacogenomics into clinical practice is the lack of context for the data, specifically in the form of dosing recommendations. Despite recommendations by the FDA to screen patients for UGT1A1*28, there are no specific guidelines on irinotecan dose reduction required for patients homozygous for the polymorphism. To exacerbate the problem, in the case of UGT1A1*28, the starting dose of irinotecan appears to determine the relevance of the polymorphism. At lower doses, UGT1A1*28 is not associated with toxic events; at higher doses, the polymorphism is significantly predictive of irinotecan toxicity.^1,11

The lack of dosing algorithms incorporating pharmacogenomics data is a huge barrier to clinical adoption. Physicians may be reluctant to order tests if there is no clear context with which to interpret the results.

PHARMACOGENOMICS HEALTH INFORMATION MANAGEMENT SYSTEM

With the current impetus of Electronic Health Records (EHRs),¹² integrating pharmacogenomics data at this early stage of EHR adoption is essential. The often cited promise of pharmacogenomics can only be realized when clinicians, patients and researchers can access, retrieve and integrate genomic information to manage health care collaboratively. A Pharmacogenomics Health Information Management System (PHIMS) guidance engine¹³ would bridge the gaps between current pharmacogenomic research and point of care with the following benefits:

• Facilitate the automated transfer of relevant and current pharmacogenomics information between the lab and point of care

• Integrate other relevant clinical and demographic information, such as the EHR or validated dosage algorithms, to help the physician determine an appropriate course of therapy

• Provide validated clinical guidance on drug prescriptions to the physician based on pharmacogenomic lab results

• Track the therapy and dosage prescribed on the basis of the pharmacogenomics test

Of particular interest would be if test results trigger a change in a proposed treatment regimen, and how well the patient is performing on the new personalized drug regimen, and whether this information can be integrated into future algorithms. This process would allow pharmacogenomics to be seamlessly incorporated into clinical practice. However, it relies on the availability of all the requirements stated above, in particular, access to approved assays and dosing algorithms.

FUTURE DIRECTIONS

As a field of research, pharmacogenomics continues to become ever more intricate. For some medications, it is rapidly becoming apparent that individual genotypes have limited predictive value^14,15 and combinations of genotypes and genomic variation within genes or from a range of genes associated with each given drug are more likely to provide clinical relevance. For gastrointestinal cancers, where a range of therapy options are available, it could be envisioned that a panel of polymorphisms could be screened to provide a pharmacogenomics profile that would predict the appropriate drug and dose for each patient.

As more pharmacogenomics tests become approved for use by the appropriate regulatory agencies, the need for integration into EHRs to compile the data, provide dosing recommendations and track clinical outcomes will become more acute. This will also provide a useful feedback mechanism, whereby knowledge of both genotypes and the frequency of adverse events can be incorporated into future pharmacogenomics research. However, before clinical effectiveness can be derived, a considerable amount of research to identify and validate new pharmacogenomic markers is still required for the majority of medications.