Is It Time to Expand Our Definition of “Clinical Utility”?

Abstract

Currently, genetic tests that predict cancer risk or risk of recurrence in patients who have had their cancer treated with curative intent must have proven “clinical utility” to be recommended by the organizations responsible for publishing the standard-of-care guidelines for cancer care.

Based on the current definition of clinical utility, most patients are denied testing for cancer-predisposing genes or pathogenic germline variants even though germline testing has been proven as highly accurate in identifying pathogenic germline variant carriers, there are measures recommended to prevent and diagnose early cancers associated with particular PGVs, and disparities in patient access to genetic tests are well described.

Similarly, despite dozens of studies demonstrating that detected circulating tumor DNA (ctDNA) after curative intention therapy of different cancer types is a highly accurate biomarker that predicts recurrence, the major organizations that publish guidelines for cancer monitoring after curative intention therapy recommend against using ctDNA assays to detect minimal residual disease and thereby predict recurrence for all solid tumor malignancies.

Here, the primary reasons that these genetic tests are considered to lack proven clinical utility and the primary evidence suggesting that a broader definition of clinical utility should be considered are discussed. By expanding the definition of clinical utility, many patients will benefit from the information gained from having these genetic tests.

Commentary

Precision oncology includes the appropriate application of genetic tests to prevent, detect earlier, diagnose, monitor, and treat cancer. To be recommended, typically a test must be proven to have “clinical utility.” Without proof of clinical utility per a National Comprehensive Cancer Network (NCCN) expert panel, for example, coverage for the costs of that test are typically denied by payers. ¹ According to the NCCN, clinical utility refers to the ability of the assay to improve clinical decision making and patient outcomes and “even if a marker clearly distinguishes between “positive” and “negative” groups, if evidence for differential treatment is not available, then no reason exists to test for the marker”. ² By limiting genetic testing solely to tests that inform treatment decisions, patients are denied information that might be viewed as highly valuable for other reasons.

For example, for patients who have a low pretest probability of testing positive for a cancer-causing pathogenic germline variant (PGV) (eg, BRCA1/2), the NCCN does not recommend germline testing (GT) because testing of this population is felt to lack clinical utility, according to the NCCN definition. ³ However, the American College of Genetics and Genomics (ACMG) defines clinical utility far more broadly. Per the ACMG definition, tests that affect “diagnostic or therapeutic management, implications for prognosis, health and psychological benefits to patients and their relatives, and economic impacts on health-care systems” are all considered as having clinical utility. ⁴

Similarly, circulating tumor DNA (ctDNA) assays are not recommended to monitor patients during and after curative intention therapy for any solid tumor, because ctDNA monitoring in these patients is said to lack clinical utility per a joint review from the American Society of Clinical Oncology and the College of American Pathologists. ⁵ Last year, Thomsen et al underscored the lack of evidence of clinical utility as the reason that ctDNA assays should not yet be routinely used for monitoring patients with metastatic cancer. ⁶

Expanding the definition of clinical utility might be reasonable so that each patient can decide for themselves whether-according to their own values, goals, and beliefs-the benefits associated with undergoing one of these genetic tests outweighs the harms associated with that test and therefore has clinical utility for them. A more personalized definition might also include the notion that sometimes the information gleaned from the test result, even if that result has not been shown-if applied-to favorably affect outcome, can be clinically useful to patients.

In recent articles published in Cancer, the Journal of Clinical Oncology, JCO Precision Oncology and elsewhere, experts have considered the challenges in defining “clinical utility.”^6–8 But clinical utility is “rarely easily quantified” and is “frequently a matter of judgment.” ⁷

In 2020, Hayes wrote “In simple words, it is the demonstration that the patient’s outcome is likely to be better because the TBT [tumor biomarker test] result was used to select the clinical strategy compared to that of a patient whose care is managed without knowledge of that result.” ⁹ Certainly, for nearly all the tests ordered on behalf of cancer patients, this definition seems reasonable. For example, applying the results of immunohistochemical evaluation of breast cancers for estrogen receptors and HER2 expression has improved the outcome for millions of patients diagnosed with breast cancer.

In 2022, Pritchard et al ⁸ suggested an “improved definition” of clinical utility. The authors suggested that tests are clinically useful when the results can be used to enroll more patients on clinical trials and when testing mitigates health care disparities. For example, PARP inhibitor therapies have been proven from trials as effective in patients with ovarian, breast, pancreatic and prostate cancers for those patients who carry BRCA1/2 pathogenic germline variants (PGVs). ¹⁰ Enrollment in clinical trials to test the utility of PARP inhibitors in patients who carry BRCA1/2 PGVs who have other tumor types will be increased if testing for BRCA1/2 PGVs is considered to have clinical utility for those patients with, for example, lung or colon cancer.

Related to disparities, there are enormous gaps between socioeconomic, racial, and ethnic groups in whom GT for PGVs is offered. Recently, Dillon et al. ¹¹ concluded that “disparities in receipt of genetic testing have been consistently demonstrated and undoubtably have significant implications for the populations not receiving standard of care. If the Cancer Moonshot version 2.0 recommendation that all patients with a diagnosis of any solid tumor type be referred for evaluation for eligibility for genetic testing is adopted, that gap will be narrowed. ¹²

Understanding why testing patients for PGVs who have low pretest probabilities of carrying a PGV is not considered clinically useful is helpful in contemplating expanding the definition of clinical utility. Implicit in the definition is the assumption that to be clinically useful, the harms of a test must be outweighed by the benefits.

Related to those with a low pretest probability of carrying a PGV, the major concern is that the “harms” (eg, uncovering a variant of uncertain significance (VUS) leading to inappropriate conclusions regarding the significance of the VUS) outweigh the benefits (eg, uncovering a PGV with implications for prevention, early detection of cancers and cascade testing associated with that PGV) of GT. ¹³ Yet a particular patient with a low pretest probability-after hearing the harms and benefits of undergoing testing for a BRCA1/2 PGV might reasonably decide, for themselves, that the harms are outweighed by the benefits and wish to be tested. In other words, it is important to consider expanding the definition of clinical utility to recognize that-in this era that emphasizes personalized care-different patients might have different notions of whether the same test result is useful to them, even though it might not be considered to have clinical utility based on the current definition.

Cancer Moonshot recognizes the shortage in genetic counselors and the need for more training in genetics for primary care and non-genetic counselor specialists. ¹² Still, this goal is daunting. Currently, the NCCN Genetic/Familial Assessment: Breast, Ovarian and Pancreatic Guidelines are roughly 150 pages. The expertise of genetic counselors is vital to ensuring patient satisfaction and appropriately interpreting test results, including uncovering PGVs and VUS. For example, from a report involving 120 PCPs in two states, the authors concluded that “perhaps the missing element is that busy physicians need systems-level support to engage in meaningful discussions around genetic issues. ¹⁴

A more fundamental question is whether clinical utility is possible, even if the results of a test do not inform management decisions. In 2010 Lesko et al agreed with a definition of clinical utility used by Secretary’s Advisory Committee on Genetics, Health, and Society: “Clinical utility refers to…the value of information to the person being tested…Even if no interventions are available to treat or prevent disease, there may be benefits associated with knowledge of a result.”1 ¹⁵

Information that does not inform treatment decisions is particularly complicated when related to germline testing. For example, it might be far more challenging to quantify the cost-effectiveness and consider the ethics and legal issues of information regarding the risk of developing cancers for patients and their family members and designating who will be paying for genetic tests in family members when the indication is prompted by testing in a relative. When clinical utility is defined as only related to treatment decisions, ethical, legal, psychological, societal, and cost-effectiveness issues might understandably be more easily quantified. Yet these areas might all reasonably be considered aspects of clinical utility and are nicely discussed in recent publications.^16,17

Besides GT for PGVs, other genetic and genomic tests warrant consideration related to their clinical utility in a variety of settings, even though the tests might not inform treatment decisions. For example, the NCCN panels have yet to endorse ctDNA assays for diagnosing solid tumor monitoring during treatment, or for predicting recurrence after completion of curative-intention therapy. In fact, in one of the very few mentions in any of the NCCN guidelines of ctDNA utility in the evaluation or management of cancer patients with solid tumors, the NCCN colon cancer panel writes that although detectable post-adjuvant therapy ctDNA predicts a 17-fold increased risk of recurrence compared to having undetectable post-adjuvant therapy ctDNA, the panel does not recommend testing and “questions the value” of monitoring ctDNA, ostensibly because it is unproven that detectable ctDNA post-adjuvant therapy can inform treatment decisions or result in improved outcomes.^18,19 Although the above shortcomings of ctDNA testing are all true, a patient might understandably see value in knowing that their risk of recurrence is 17-fold higher if their post-adjuvant therapy ctDNA is detectable.

Also, although intervention studies have not yet proven that identifying detectable ctDNA post-definitive therapy can suggest a more effective therapy to prevent recurrence, management decisions-such as the timepoints for routine surveillance imaging-can be altered and discovery of recurrent disease and initiation of systemic therapy for recurrent disease started sooner if detectable ctDNA is identified post-definitive therapy.

Currently, there are no clear examples where genetic or genomic testing informs an escalation or de-escalation adjuvant therapy strategy for patients with solid tumors, patient satisfaction and adherence to recommended therapy is another benefit that may result from testing and therefore considered a clinically useful application. For example, non-adherence to adjuvant endocrine therapy is seen in patients with a good prognosis, and multiple endorsed genomic assays predict prognosis with or without chemotherapy more accurately than non-genetic tests. ²⁰ Also, although adjuvant endocrine therapy is recommended for most patients with early-stage hormone receptor positive breast cancer, Black women had a lower rate of genomic testing and were less likely to receive endocrine therapy. ²¹ Regarding extended endocrine therapy (EET), if the quantitative RNA NCCN-endorsed Breast Cancer Index (BCI) test result demonstrates that continuing adjuvant endocrine therapy beyond 5 years will be beneficial, Sanft reported that 82% of patients stated they would more likely adhere to that recommendation if it was based on the BCI test result.^10,22

Similarly, Fear of Cancer Recurrence (FCR) is among the most common concerns of patients who have completed definitive therapy with intent to cure. In one study, 59% of cancer survivors reported moderate FCR and 19% reported severe FCR. ²³ Many patients would view tests that only predict greater or lesser likelihoods of recurrence as offering useful information that could affect their degree of FCR.

Louis Armstrong was once asked “What is jazz?” He replied, “If you have to ask what jazz is, you’ll never know.” Despite the elasticity and challenges in defining clinical utility, for over 25 years, the NCCN panels have masterfully published guidelines that are based on evidence of clinical utility. Applying test results to affect outcome will remain the most important component of the definition of clinical utility. However, personalized oncology includes recognizing that one patient’s notion of clinical utility may be very different from that of another patient or that of an expert panel. Also, patients may benefit from knowing test results, even if applying that information has not been proven to affect outcome.