Introduction

A systems therapeutics framework, “Systems Therapeutics: A Diagram and Four Categories”, was recently presented on this website, tri-institute.org (2015), illustrating pharmacologic processes and pathophysiologic processes separately. This framework incorporated four different levels of interactions between these two fundamental processes, ranging from the molecular level, through the cellular and tissue/organ levels, to the clinical level; these four different levels of interactions are referred to as systems therapeutics categories. This conceptual framework builds on previous work by others, principally Grahame-Smith & Aronson (1992) and Post et al. (2005), and significantly expands previous work by Bjornsson (1996). The purpose of this initial work on systems therapeutics was to present a theoretical framework, which could be useful when examining different types of therapeutic effects; definitions and a few examples were presented for each systems therapeutics category.

Fundamental to considerations of therapeutic effects is the issue of interpatient variabilities, and determinants or drivers of such variabilities where these are understood, as discussed by Rowland & Tozer (2011) and Eichler et al. (2011). Variability in pharmacologic processes for approved drugs, principally in pharmacokinetics and pharmacodynamics, have been well recognized and characterized over that past several decades, although the final step, the translation from pharmacologic response to therapeutic effect, still remains elusive in many instances, particularly in drug discovery and development. Interpatient variability in pathophysiologic processes, from disease initiation to disease manifestations or signs and symptoms, however, has received much less attention than that in pharmacologic processes. Thus, when addressing interpatient variability in therapeutic response to approved drugs, including response range, response magnitude, responder rate, and lack of response, it is not always obvious to what extent interpatient variability in pathophysiologic processes, or disease progression, might contribute to the observed therapeutic response, in addition to variability in pharmacologic processes.

Since issues of such variabilities were not addresses in the previous paper, and considering their importance, the purpose of the present paper is to highlight key areas in systems therapeutics variabilities. Below is a graph of the systems therapeutics framework, illustrating the pharmacologic processes and pathophysiologic processes separately (click for a larger graph).

Variability in Pharmacologic Processes (Pharmacokinetics and Pharmacodynamics)

For the past three to four decades it has been assumed that interpatient variability in therapeutic effect is determined by individual differences in pharmacokinetics, pharmacodynamics and compliance. Note that the net effect of pharmacokinetics and compliance, the drug and its concentration at the site of action, as illustrated in the top left hand corner of the graph, can be considered as the initial determinant or driver of pharmacologic processes.

Variability is pharmacokinetics (PK), which covers drug absorption, distribution, drug metabolism and drug elimination, has been amply demonstrated. Tables of pharmacokinetic parameters of approved drugs are widely available and accessible, e.g., in textbooks and reviews of individual drugs or pharmacologic classes. A comprehensive table of pharmacokinetic parameters of approved drugs is included as an appendix by Thummel et al. (2011) in Goodman and Gilman’s The Pharmacologic Basis of Therapeutics. This table lists for each drug its bioavailability, urinary excretion, plasma protein binding, clearance, volume of distribution, elimination half-life, peak time and peak concentration, typically expressed as mean ± standard deviation, but sometimes as mean and range in values. A coefficient of variation (CV) for clearance, one estimate of interpatient variation, of approx. 30% is common, but these can range considerably; obviously, the range in values can be significantly larger. A recent paper by Gao et al. (2011) on over 15 targeted anticancer drugs illustrates their wide pharmacokinetic variability, showing CV for exposure varying between approx. 25-82%, and trough levels varying between approx. 6-26 fold.

Variability in pharmacodynamics (PD), covering the relationship between drug concentration and pharmacological response intensity, has received less attention than that in pharmacokinetics. While there was an earlier belief that pharmacokinetic differences and compliance were more important as determinants of pharmacologic response than pharmacodynamic differences, it has become accepted that pharmacodynamic variability is at least as wide as pharmacokinetic variability, sometimes exceeding one order of magnitude, for early reviews see Levy et al. (1994) and Levy (1998). Commonly used examples of variation in pharmacologic response in the earlier literature involved various physiological, biochemical or related measures, e.g., pain relief and sedation, cardiovascular, anticoagulant, and metabolic measures. As a specific example, pharmacokinetic-pharmacodynamic modeling of the opioid analgesics fentanyl, alfentanil and trefentanil by Lemmens et al. (1994; cited in Levy 1998), using EEG as a pharmacodynamic measure, reported a range in CV for EC₅₀ (drug concentration producing 50% of E_max, i.e., the maximum effect) of 47 to 85%. Interpatient variability in pharmacodynamics can be manifested in the different parameters of the concentration vs. response relationship, i.e., EC₅₀, E_max, slope parameter of the sigmoid curve, and baseline; numerous factors may impact these.

Variability in Pathophysiologic Processes (Disease Initiation and Disease Progression)

While detailed descriptions of etiology, pathogenesis, pathophysiology, pathology, prognosis, and clinical manifestations and treatment of diseases are provided in textbooks of pathology and medicine, there is limited discussion of quantitative patient variability in disease progression over time. Ideally, such information would come from longitudinal studies on the natural history of diseases, but unfortunately, these are not available for most diseases, although clinically, such variability has been well recognized for centuries.

Disease pathophysiologic processes can be visualized as starting with disease initiation (refer to the bottom left hand corner of the graph). Recent work using advanced bioinformatics and network-based approaches has started to elucidate some of the components of disease initiation and their interactions. Although such work has not been directly aimed at variability, because of its promise in the future, two examples are included. First, Liu et al. (2009) introduced the term “etiome” to describe the combination of genetic and non-genetic etiologic factors associated with diseases; this publication identified 863 diseases as having both etiologic genetic and environmental factors (the latter referring to non-genetic factors in the broadest sense). Second, work by Barabasi et al. (2011) and others on so-called network medicine has sought to explore systematically the molecular connections and complexity of individual diseases and the molecular relationships among different (patho)phenotypes. The exciting promise of such approaches includes how these network constructs may connect with pathophysiologic processes or disease progression – systems pathobiology and clinical phenotypes – and thus contribute to determinants or drivers of individual patient’s disease progression; see opinion paper by Loscalzo & Barabasi (2011).

As was mentioned above, there are only limited examples of the natural history of diseases or disease progression. However, data on disease progression over relatively short periods of time, involving repeated measures of disease status, are available and have been collected as part of clinical trials of drug efficacy, as part of control or placebo groups, and as analyzed by disease progress models using pharmacostatistical techniques. Note some assumptions may need to be made regarding the use of placebo groups as representative of natural history of disease. Several different disease progress models have been described, depending on the nature of the disease in question, including symptomatic and disease-modifying effects, and continuous or discontinuous progression, see for example Holford et al. (2012). Such pharmacometrics models have been developed for several diseases, e.g., Alzheimer’s disease, Parkinson’s disease, type-2 diabetes, depression, schizophrenia, chronic obstructive pulmonary disease, and osteoporosis; see summaries by Holford (2013). One example of disease progression involves assessment of cognitive deterioration in Alzheimer’s disease by Ito et al. (2011), based on ADIS-cog measurements, which showed significant variability, with age, gender, APOEe4 genotype and baseline status determining the decline in cognition. Studies on Parkinson’s disease have shown marked variability among patients on the rate of progression, assessed with global functional score, see Holford & Nutt (2008), suggesting there are distinct disease subtypes with different rates of progression.

Comments

Considering the significant interpatient variability in therapeutic response to approved drugs, including responder rate and response range, it is important to understand what processes contribute to such variability and their approximate contributions. This paper looked at three key areas, i.e., pharmacokinetic and pharmacodynamic variabilities in pharmacologic processes and disease progression in pathophysiologic processes; in addition, there was a brief introduction to bioinformatics efforts on disease initiation. Key comments are as follows:

• There are well documented significant interpatient pharmacokinetic and pharmacodynamic differences; there are indications that the latter may be at least as wide as the former.
• As expected, it is noted that reporting of PK parameters is better standardized than that of PD parameters. It is desirable that more uniform approaches are used in the reporting of PK and PD data and parameters, including the reporting of range in values.
• It is recommended that the control or placebo data from disease progress models be routinely reported as an estimate of disease progression variability, including any caveats regarding placebo response.
• It was noted during the preparation of this report that there are relatively few good up-to-date summaries/reviews on variabilities within these three key processes, particularly pharmacodynamics and disease progression.
• Knowledge of variabilities in key underlying processes will provide better understanding of therapeutic response characteristics of approved drugs.
• One looks forward to the time when bioinformatics approaches to disease initiation can be linked to pharmacometrics approaches to disease progression.

References

• Barabasi AL, Gulbahce N, Loscalzo J (2011). Network medicine: a network-based approach to human disease. Nat. Rev. Genetics, 12:56-68.
• Bjornsson TD (1996). A classification of drug action based on therapeutic effects. J. Clin. Pharmacol., 36:669-673 (p. 670, Fig. !).
•  Eichler HG, Abadie E, Breckenridge A, Flamion B, Gustafsson LL, Leufkens H, Rowland M, Schneider CK, Bloechl-Daum B (2011). Bridging the efficacy-effectiveness gap: a regulator’s perspective on addressing variability of drug response. Nat. Rev. Drug Disc., 10:495-506.
• Gao B, Yeap S, Clementis A, Balakrishnar B, Wong M, Gurney H (2012). Evidence for therapeutic drug monitoring of targeted anticancer therapies. J. Clin. Oncol., 30:4017-4025.
• Grahame-Smith DG, Aronson JK (1992). Oxford Textbook of Clinical Pharmacology and Drug Therapy, Oxford University Press, Oxford (Chapter 5. The Therapeutic Process, pp. 55-66, Fig. 5.1).
• Holford N, Nutt JG (2008). Disease progression, drug action and Parkinson’s disease: Why time cannot be ignored. Eur. J. Clin. Pharmacol., 64:207-216.
• Holford NHG, Mould DR, Peck CC (2012). Disease progress models. In: Principles of Clinical Pharmacology, Adkinson AJ, Huang SM, Lertora JJL, Markey SP, editors, Third edition, Academic Press, New York, pp. 369-382.
• Holford N (2013). Clinical pharmacology = disease progression + drug action. Br. J. Clin. Pharmacol., 79:18-27.
• Ito K, Corrigan B, Zhao Q, French J, Miller R, Soares H, Katz E, Nicholas T, Billing B, Anziano R, Fullerton T (2011). Disease progression model for cognitive deterioration from Alzheimer’s Disease Neuroimaging Initiative database. Alzheimer’s & Dementia, 7:151-160.
• Lemmens HJM, Dyck JB, Shafer SL, Stanski DR (1994). Pharmacokinetic-pharmacodynamic modeling in drug development: application to the investigational opioid trefentanil. Clin. Pharmacol. Ther., 56:261-271 (p. 270, Table VI).
• Levy G, Ebling WF, Forrest A (1994). Concentration- or effect-controlled clinical trials with sparse data. Clin. Pharmacol. Ther., 56:1-8 (p.2, Table 1).
• Levy G (1998). Predicting effective drug concentrations for individual patients. Clin. Pharmacokinet., 34:323-333.
• Liu YI, Wise PH, Butte AJ (2009). The “etiome”: identification and clustering of human disease etiologic factors. BMC Bioinformatics, 10 (Suppl. 2):S14, 1-10.
• Loscalzo J, Barabasi AL (2011). Systems biology and the future of medicine. Wiley Interdiscip. Rev. Syst. Biol. Med., 3:619-627.
• Post TM, Freijer JI, DeJongh J, and Danhof M (2005). Disease system analysis: Basic disease progression models in degenerative disease. Pharm. Res., 22:1038-1049 (p. 1039, Fig. 1).
• Rowland M, Tozer TN (2011). Clinical Pharmacokinetics and Pharmacodynamics: Concepts and Applications, Fourth edition, Lippincott Williams & Wilkins, Philadelphia (Chapter 12, Variability, pp. 333-356).
• Thummel KE, Shen DD, Isoherranen N (2011). Design and optimization of dosage regimens: Pharmacokinetic data. In: Goodman and Gilman’s The Pharmacological Basis of Therapeutics,  Brunton, L, Chabner B, Knollman, B, editors, Twelfth edition, McGraw Hill, New York, p. 1891.
• Therapeutics Research Institute (2015). Systems Therapeutics: A Diagram and Four Categories, April, 2015. tri-institute.org/niDFW.