The Seduction of Biomarkers in the Practice of Medicine and the Tyranny of Power in the Drug Approval Process: Lessons From Niacin

The practice of clinical medicine continues to evolve. The proper use of technology has undoubtedly improved the longevity and quality of life of patients throughout the world. However, overreliance on and overinterpretation of selected surrogates (biomarkers) for disease improvement, subsequent to therapeutic interventions, threaten to supplant a focus on complete history taking and thorough physical examination of the patient. This more comprehensive approach, although ideal, may not be operational to an extent sufficient to relegate the interpretation of biomarkers to a supportive role as one of many factors guiding optimization of therapeutic interventions. Several recent developments have likely contributed to this trend. Time pressure on all physicians, but especially those in primary care, may abrogate the due diligence necessary to adequately gain a complete picture of illness. Although overall physician-patient communication seems not to be negatively impacted by the availability of computers in examination rooms, the instant availability of laboratory tests at the time of the patient visit may seduce both the patient and the physician to emphasize communications on questions such as “Doc, what do my numbers look like today?” to the exclusion of other pertinent issues. Lastly, the intensity of direct-to-consumer advertising, complemented by pharmaceutical detailing techniques, reinforces a natural human tendency to focus on numerical parameters that can be most directly impacted by therapeutic interventions. The clinical trial NCT00120289, a study to determine whether adding niacin to statin therapy would reduce the incidence of vascular events, is the most recent example that further informs the proper role regarding the use of biomarkers in clinical medicine.1 This study was stopped 18 months early because there was no possibility that adding extended-release niacin to either the simvastatin or atorvastatin treatment arms would achieve the desired clinical end-points of reducing cardiovascular disease and because it unexpectedly documented a small unexplained increase in ischemic strokes in the high-dose niacin group, despite achieving marked improvements in serum triglycerides and high-density lipoprotein (HDL).1 Niacin's lack of impact on clinically relevant cardiovascular outcomes in this clinical trial, despite improvement in both surrogate biomarkers, is consistent with and extends observations from previous reports of the lack of benefit on cardiovascular outcomes in patients with type 2 diabetes when fenofibrate is added to simvastatin, despite a significant reduction in triglycerides.2 Careful interpretation of these developments may clarify the singular importance of lowering low-density lipoprotein (LDL) cholesterol, according to our current understanding of the mechanism of the lipogenesis-cellular signaling continuum as it relates to the development of drug targets. Statin drugs are unquestionably effective in improving cardiovascular outcomes, largely due to their impact on LDL. Similarly, clinical trials with niacin, fibrates, and resins also have data supporting their benefit in reducing such outcomes by degrees roughly correlative to their efficacy in lowering LDL.3 In addition, niacin is unparalleled in the pharmacopoeia in its potency to simultaneously lower triglycerides and increase HDL, often markedly so.1 Hence, a reasonable conclusion to be derived from these observations is that LDL reduction currently should be the sole lipid biomarker targeted for pharmacological manipulation by clinicians. However, an important reminder is that methods still routinely used by many laboratories estimate LDL by the Friedewald equation, using measured levels of total cholesterol, HDL, and triglycerides. Thus, LDL becomes unable to be interpreted if the concurrent serum triglyceride concentration exceeds 400 mg/dL (4.52 mmol/L). This clinical situation, in addition to the need to reduce pancreatitis risk with extreme elevations in triglycerides, may mandate additional drug therapy. As such, the addition of niacin or a fibrate (other than gemfibrozil to avoid a significant interaction with many statins) may be necessary.4 Indeed, although there is a consensus that low HDL correlates with adverse cardiovascular outcomes with less clarity in regards to hypertriglyceridemia, at least 3 points need to be considered. First, the combination of these 2 aberrant lipid parameters frequently coincides, thereby making the relative contribution of either to cardiovascular risk difficult to separate.5 Second, aggravating this phenomenon is that they are also frequently associated with, but not necessarily causal of, endocrinopathies with complex pathophysiology, most notably type 2 diabetes.6 Third, although improvement in the serum levels of both lipid abnormalities is correlated with improved cardiovascular outcome, strategies to do so are multifactorial, with drug treatment being only one (and perhaps minor) within a broader treatment plan to improve cardiovascular health in such patients.7 Therefore, making the assumption that isolating a biomarker and designing a specific drug that targets it to improve actual clinical outcome becomes suspect, if there is not concurrent understanding and investigation of the broader pathophysiology of the disease process in question. The preceding considerations provide a backdrop for the tyranny of the central role that the application of statistical power holds in the drug development and approval processes, which has a direct impact on the strategies for drug marketing following US Food and Drug Administration (FDA) approval. Entirely obvious are the enormous financial and time-constraint burdens faced by pharmaceutical manufacturers via their commitments to shareholders and their relationships to competitors, respectively. These realities pressure the search for and identification of isolated biomarkers as surrogates of disease, perhaps without an equally thorough attention to other physiological (and potentially adverse) effects that may be predictable from the mechanisms of drug action.8 Furthermore, once a biomarker becomes the primary outcome measure for testing in the later phases of the drug development process, it further necessitates a highly restrictive exclusion process.9 The FDA further promotes such strategies, thereby applying additional pressure on pharmaceutical manufacturers. Specifically, through its critical path and similar initiatives, the FDA stresses the importance of biomarkers throughout the drug development process.10 The aggregate effect of these factors is to promote patient selection methods in phase 3 that maximize study power and minimize the sample size needed to show efficacy in order to obtain more expeditious and cost-effective FDA approval. During phase 4 of the drug development process, pharmaceutical detailing strategies directed toward clinicians in practice and direct-to-consumer advertising techniques may underplay the lack of the ability to extrapolate results of phase 3 studies to the more representative patient population cared for by practicing clinicians. Further exacerbating this phenomenon is the overemphasis on and overinterpretation of modest pharmacokinetic and pharmacodynamic differences by the pharmaceutical industry to both patients and practitioners between drugs marketed as metabolites or enantiomers derived from parent compounds. Three particularly poignant recent examples of these pitfalls in drug development and testing include torcetrapib, rosiglitazone, and rofecoxib. Torcetrapib, an inhibitor of the cholesterol ester transfer protein, showed particular promise to solely increase HDL by up to 61%.11 However, the drug failed to achieve FDA approval due to an association with an increase in sudden death in phase 3 testing.12 Rosiglitazone is a PPAR- agonist that profoundly decreased insulin resistance, thereby markedly reducing the glycosylated hemoglobin biomarker. However, the same mechanism also regulates a variety of intracellular proteins, some of which are associated with an increase in intracellular oxidized LDL.13 These and other effects likely contributed to the observed association between rosiglitazone and an increase in adverse cardiovascular events.14 This constellation of factors led the FDA to require the drug sponsor to submit to a Risk Evaluation and Mitigation Strategy in 2010. Similarly, rofecoxib and valdecoxib, 2 selective cyclooxygenase inhibitors (coxibs), were removed from the US market in 2005. The removal of these 2 drugs was the culmination of a trend in the pharmaceutical industry in Europe and the United States towards the development and eventual human testing of coxibs, using the primary criterion for selection of compounds as their potency on the ratio of COX-2 to COX-1 inhibition.9 That the identical criterion (COX-2 to COX-1 potency) is also directly proportional to cardiovascular toxicity risk from these same compounds clearly underscores the dangers of having a myopic view on a single biomarker in the drug development process.15 So what useful messages can practitioners in the front lines, particularly those in the practice of primary care, derive from these recent developments in lipid research and such ingrained practices in the pharmaceutical industry? First, a commitment to sift through the marketing noise to arrive at a biomarker signal that does significantly impact actual disease outcomes must be made. Such a commitment can be based on a greater appreciation of what underlies the drug development process through phase 4 and will ultimately lead to better patient care. Second, it is the responsibility of practitioners to further commit to efforts at reclaiming the central place of treating the whole patient and the entirety of disease by correctly prioritizing the panoply of factors, most of which are often vastly more important than biomarkers (eg, smoking and body mass index [BMI]). Financial disclosure: None declared.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.