Full Review Drugs meeting the molecular basis of diabetic kidney ...

Nephrol Dial Transplant (2015) 30: iv105–iv112 doi: 10.1093/ndt/gfv210

Full Review Drugs meeting the molecular basis of diabetic kidney disease: bridging from molecular mechanism to personalized medicine Hiddo J. Lambers Heerspink1, Rainer Oberbauer2, Paul Perco3, Andreas Heinzel3, Georg Heinze4, Gert Mayer5 and Bernd Mayer3 1

Department of Clinical Pharmacy and Pharmacology, University of Groningen, University Medical Center Groningen, Groningen, The

Netherlands, 2Department of Internal Medicine III, Medical University of Vienna, Vienna, Austria, 3Emergentec biodevelopment GmbH, Vienna, Austria, 4Center for Medical Statistics, Informatics and Intelligent Systems, Medical University of Vienna, Vienna, Austria and 5Department of Internal Medicine IV, Medical University of Innsbruck, Innsbruck, Austria

Correspondence and offprint requests to: Bernd Mayer; E-mail: [email protected]

A B S T R AC T Diabetic kidney disease (DKD) is a complex, multifactorial disease and is associated with a high risk of renal and cardiovascular morbidity and mortality. Clinical practice guidelines for diabetes recommend essentially identical treatments for all patients without taking into account how the individual responds to the instituted therapy. Yet, individuals vary widely in how they respond to medications and therefore optimal therapy differs between individuals. Understanding the underlying molecular mechanisms of variability in drug response will help tailor optimal therapy. Polymorphisms in genes related to drug pharmacokinetics have been used to explore mechanisms of response variability in DKD, but with limited success. The complex interaction between genetic make-up and environmental factors on the abundance of proteins and metabolites renders pharmacogenomics alone insufficient to fully capture response variability. A complementary approach is to attribute drug response variability to individual variability in underlying molecular mechanisms involved in the progression of disease. The interplay of different processes (e.g. inflammation, fibrosis, angiogenesis, oxidative stress) appears to drive disease progression, but the individual contribution of each process varies. Drugs at the other hand address specific targets and thereby interfere in certain disease-associated processes. At this level, biomarkers may help to gain insight into which specific pathophysiological processes are involved in an individual followed © The Author 2015. Published by Oxford University Press on behalf of ERA-EDTA. All rights reserved.

by a rational assessment whether a specific drug’s mode of action indeed targets the relevant process at hand. This article describes the conceptual background and data-driven workflow developed by the SysKid consortium aimed at improving characterization of the molecular mechanisms underlying DKD at the interference of the molecular impact of individual drugs in order to tailor optimal therapy to individual patients. Keywords: drug, personalized medicine, prediction, systems biology, type 2 diabetes

INTRODUCTION About one of three patients with diabetes mellitus develops some degree of chronic kidney disease (CKD) [1]. Given that the prevalence of type 2 diabetes mellitus is approximately 8% in industrialized countries [2], diabetic kidney disease (DKD) has become a huge global health burden. For example the US government (via the National Center for Chronic Disease Prevention and Health Promotion) has recently issued an initiative to provide comprehensive public health strategies for promoting kidney health in diabetes [3]. These plans are designed to raise awareness and control renal risk factors in order to prevent end-stage renal disease (ESRD) and maintain quality of life of affected individuals. Haemodialysis as treatment modality of most ESRD patients with diabetes consumes about 1–3% of the national health care budgets although the iv105

FULL REVIEW

prevalence is only 0.1% [4, 5]. There is thus a strong economic imperative to improve the outcomes of type 2 diabetes as well as a strong personal and societal health rationale. Risk factors and intervention studies predominantly addressed established cardiovascular risk profiles such as arterial hypertension, hyperlipidaemia and poor glucose control. Although these risk factors predict renal function decline on a population level, the individual prediction of the course of glomerular filtration rate (GFR) decline remains poor. Of the novel promising predictive biomarkers, only very few have been validated in large prospective cohorts, and thus the explained variability of GFR loss on top of established clinical risk factors remains elusive in the everyday clinical setting. Past intervention studies have addressed the established cardiovascular risk factors and demonstrated the importance of HbA1c control and blood pressure control in delaying the progression of DKD. Blood pressure control with drugs intervening in the renin-angiotensin-aldosterone-system (RAAS) is particularly effective and remain so far the only established prevention strategy in already proteinuric patients with type 2 diabetes [6– 8]. Even in these successful studies, the number needed to treat for prevention of ESRD averages about 60 patients suggesting that the absolute risk reduction is only 1–2%. Thus the obvious question is why some patients experience a progressive loss of renal function whereas others do not. Accordingly, treatment is beneficial in some patients but fails in many others. There are several potential explanations for the lack of treatment effect. Besides low adherence to therapy due to polypharmacy and prescription of drugs for many years without an immediate positive effect, the complex and heterogenic pathophysiology of DKD is likely another key reason. Morphological features of DKD range from predominantly atherosclerosis to severe podocyte injury and subsequently partial obliteration of the glomerular tuft to full glomerulosclerosis. Many deregulated molecular pathways have been identified in different stages of DKD, but the integration of these data into a wider molecular network to better discern the key mediators of progression of DKD and how drugs intervene in these processes in individual patients only recently started [9, 10]. Given the large heterogeneity in pathophysiology of DKD and the substantial variability in response to renoprotective drugs, the SysKid consortium developed a data-centric workflow to improve molecular phenotyping and characterization of the underlying molecular mechanisms of DKD on the level of individual patients in order to develop novel treatment approaches with a special emphasis on personalized medicine.

T R E AT M E N T O F D I A B E T I C K I D N E Y D I S E A S E Optimal blood pressure and HbA1c control are the mainstay of treatment for diabetic nephropathy. HbA1c control is particularly important to prevent microvascular complications as shown in landmark clinical trials such as the United Kingdom Prospective Diabetes Study (UKPDS) and Action in Diabetes and Vascular disease: preterAx and diamicroN-MR Controlled Evaluation (ADVANCE) trials [11, 12]. Blood pressure control with intervention in the RAAS is vital at any stage of

iv106

nephropathy and slows the progression of renal disease [13]. The success of these strategies is illustrated by a recent study of Andrésdóttir et al., who showed that intensified RAAS blockade, metabolic control and improved cholesterol treatment results in an almost 50% reduction in mortality in both type 1 and type 2 diabetic subjects with CKD [14]. However, despite much progress has been made still many patients progress to ESRD, or are confronted with cardiovascular complications. There is thus an urgent need for tailoring medication or identifying novel drugs to mitigate the complications of diabetes. In the last decade, several novel drugs and strategies have been tested, without success however [15]. First, intensification of RAAS blockade by combinations of Angiotensin-Converting-Enzyme-inhibitors (ACEi), ARBs or direct renin inhibitors did not lead to the expected improvement in renal and cardiovascular outcome. The ALiskiren TrIal in Type 2 diabetes Using carDiorenal End points (ALTITUDE) and Veterans Affair Nephropathy in Diabetes trials (VA NEPHRON-D) were stopped early because of more cases of worsening of renal function and hyperkalemia in the dual RAAS blockade treatment arm [16, 17]. In addition, in the ALTITUDE trial dual RAAS blockade did not confer renal or cardiovascular protection, in fact it showed a trend towards more stroke events, despite additional blood pressure and albuminuria lowering. In the NEPHRON-D trial, blood pressure and albuminuria also decreased and dual RAAS blockade caused a 34% risk reduction of ESRD, but this was not statistically significant (P = 0.07) possibly due to the early discontinuation of the trial. Endothelin receptor blockade represents another promising therapeutic strategy because of the powerful blood pressure and albuminuria lowering effects [18]. However, the hard outcome study with one of the first endothelin receptor antagonists (ERA) in this class (avosentan) demonstrated that its use was associated with increased incidence of oedema and hospitalization for heart failure which led to the premature discontinuation of the trial [19]. The high risk of heart failure was attributed to the drug’s sodium retaining effects. Key aspects of DKD pathogenesis include oxidative stress, inflammation and fibrosis. A recent phase III trial with bardoxolone methyl addressed inflammation and oxidative stress in order to delay ESRD. However, this trial also had to be stopped early due to increased incidence of heart failure and mortality in the bardoxolone methyl treatment arm [20]. Thus, various trials with different drugs and targets failed to demonstrate the efficacy and safety of new interventions for DKD. In fact, some trials even showed a higher risk of complications, mainly stroke and heart failure. The recent failures in developing novel treatments for DKD reveal a fundamental problem in drug research and development productivity. Bringing a new molecular entity to the patient takes approximately 13 years with costs exceeding 1 billion US dollars [21]. Despite the enormous investments in human and financial recourses to develop new drugs still approximately 50% of all drugs fail in late stage drug development as clearly exemplified by past clinical trials in diabetic nephropathy [22]. Apparently the efficiency of the current drug development process needs conceptual improvements to successfully deliver new drugs to the patient.

H. J. Lambers Heerspink et al.

F I G U R E 1 : Cumulative distribution of changes in albuminuria after

6 months treatment with losartan or placebo in the RENAAL trial.

alternative strategies by for example using biomarkers to predict and monitor the individual drug response have to be considered and implemented to resolve the fundamental problems in drug development for DKD. New interventions In the context of DKD a broad spectrum of targets addressing diverse mechanisms are presently undergoing preclinical or early clinical evaluation, including further NOX inhibitors, interference with protein kinase C, chemokine receptor 2 antagonists and inhibitors of TGFB-linked SMAD2 signalling activation, or direct inhibition of TGFbeta. A PubMed scientific literature search provided 104 protein coding genes implicated as potential drug targets in DKD [25]. From this plethora of drug candidates the key question is how to successfully bring drugs addressing these targets to the right patients. A strategy of tailoring these drugs to individual patients using biomarkers to select the right drug for the right patient seems the most logical and fruitful approach. Obviously, such an approach is not only limited to new drugs but should also be considered for already registered drugs. Precision medicine as a concept As outlined by Klonoff in the context of managing diabetes precision medicine aims at tailoring medication with respect to a specific clinical presentation [26]. Precision medicine may be seen as successor of personalized medicine. Personalized medicine conceptually refers to the development of specific drugs which are specifically targeted to one individual. These drugs thus vary for each patient. An example is the development of an autologous vaccine utilizing tumour cells of an individual. A less extreme example of personalization is the stratification of patients based on clinical parameters or disease biomarkers and tailoring therapies to specific sub-group of patients [27]. Hence, precision medicine rests on phenotype profiling and allows selection of a sub-group of patients who are more likely to respond or tolerate a specific treatment. This concept is certainly not new, and has been applied in clinical trials focussing on delaying the progression of DKD. For example, past clinical trials selected patients with a high albuminuria level or low eGFR in order for sufficient ESRD end points to occur without the need for an extremely large clinical trial of long duration. However, with respect to treatment response, applying albuminuria and eGFR thresholds did not lead to the identification of responder patients (e.g. in the RENAAL trial) or identifying patients prone to side effects (e.g. in the BEACON trial) [6, 20]. It therefore appears that readily available clinical risk markers for disease development and progression (e.g. albuminuria/ eGFR) are insufficient to predict the individual’s response to treatment and are therefore unlikely to be used in the context of precision medicine for selecting responder and nonresponder patients. Optimally, precision medicine requires a match between the drug’s molecular mechanism of action and the molecular pathophysiology of a sub-group of patients. Therefore, the entire concept of precision medicine is based on the molecular mechanisms of drugs and disease. Risk factors for disease progression as well as indicators of organ (dys)function

Translation of drug candidates to personalized medicine

iv107

FULL REVIEW

Large variability between individuals in drug response The design of past clinical trials focussed on the drug effect on the overall population. The trial designs did not take into account the variable drug effects in individual patients. However, it is known for many years already that individual patients show a large variability in response to many drugs. For example, in the Reduction of End points in noninsulin-dependent diabetes mellitus with the Angiotensin II Antagonist Losartan (RENAAL) trial, a large variation in albuminuria was observed (Figure 1). This response variation is also observed in the abovementioned trials in diabetic nephropathy. Careful post-hoc analyses from these trials have suggested that a sub-group of patients may have benefitted from therapy or in contrast were prone to side effects such as heart failure and should not have been exposed to the investigational drug. A post-hoc analysis in the ALTITUDE trial showed that individuals with a larger albuminuria reduction during the first months of therapy had fewer renal events compared with patients with no reduction in albuminuria. These data suggest that enrichment of the population with albuminuria responders will result in a larger treatment effect. Since many trials were stopped early because of side effects, early identification of patients who do not tolerate the therapy is warranted. In the Bardoxolone Methyl Evaluation in patients with CKD and type 2 diabetes mellitus: the Occurrence of Renal Events (BEACON) trial patients with a brain natriuretic peptide (BNP) > 200 pg/ mL or previous heart failure were at highest risk of heart failure [23]. After exclusion of these patients, the risk of heart failure was similar in the bardoxolone methyl and placebo arm. In the ASCEND trial, the risk of heart failure during avosentan therapy was particularly pronounced in patients who had a rise in body weight more than 1.0 kg during the first weeks of therapy [24]. Thus, careful selection of patients (e.g. BNP criteria in BEACON) and patient monitoring (e.g. body weight in ASCEND) during the first weeks of treatment may help to identify patients who do not tolerate the study drug. These posthoc analyses (albeit post-hoc and only hypothesis generating) indicate that a large, individually determined response variability exists suggesting that the current ‘one-size fits all’ approach in drug development is no longer sustainable. Hence,

FULL REVIEW

may serve as proxy for such specific molecular match. However, the currently used markers for renal disease progression appear not to reflect a drug’s molecular mode of action and therefore do not predict the response to therapy. Following this line of argument a different strategy may be pursued in which the interference at the molecular level of a drug mode of action is established against the background of an individual patient’s or sub-group of patient’s pathophysiology. At such interference, biomarkers can be selected that can be used to monitor the drug effect in the background of a specific (individually determined) DKD mechanism. For implementing such an approach at first a detailed molecular process and mechanism map of DKD needs to be retrieved allowing for annotation of a molecular pathway representation of DKD pathophysiology. Second, biomarker candidates are to be determined serving as proxy for such individual molecular mechanisms. The biomarker thus serves as a proxy for a given status of a selected mechanism, and the biomarker panel covers all mechanisms considered relevant for characterizing disease presentation. Then such candidate biomarkers need to be individually tested as to whether they improve prediction of renal function decline. This leads to a (sub)set of prognostic renal biomarkers for progressive disease. As the biomarkers were derived on the basis of their mechanisms, these prognostic biomarkers seamlessly provide insight in the mechanisms associated with progressive disease. In turn, having established such biomarker panel provides enhanced molecular insight in the phenotype on a patient-specific level. As drugs need to address mechanisms associated with progressive disease the above-described phenotype profiling in turn allows categorization of patient strata for which specific drugs indicate beneficial interference. Implementing such approach is inherently data-driven.

A D ATA - D R I V E N P E R S P E C T I V E O N S T R AT I F I E D M E D I C I N E I N D I A B E T I C KIDNEY DISEASE In order to identify individuals who will respond and benefit from therapy, various types of data in a biological interaction context have to be considered such as the genetic background with direct impact on drug absorption, distribution, metabolism and excretion (ADME), and on top biomarkers reflecting disease presentation and progression including genetic background impacting specific disease mechanisms together with influence of environmental factors. Genetic background influencing drug effect and variability in drug response The genetic background of an individual may impact on drug ADME and thereby contribute to variability in drug response. Polymorphisms in genes afflicted in pharmacokinetic processes can result in drug therapy resistance for example in case of rapid drug metabolism, or can increase the susceptibility to side effects [28]. This is illustrated by a study demonstrating that polymorphisms in CYP2C9 impact on the pharmacokinetics of Angiotensin Receptor Blockers [29]. Further, sequence

iv108

variation resulting from gene polymorphism may be involved in the drug target itself. Some of these variations may be functional with respect to drug binding resulting in variability in pharmacodynamic response. A link between genetic polymorphisms and variability in drug response is illustrated by polymorphisms in the angiotensinogen 235 Met/Thr which seems to relate to RAAS activity and variability in drug response to Angiotensin Receptor Blockers [30]. However, polymorphisms in this gene and response variability were not consistently observed in all studies hampering development of genetic testing on ACE/ARB response in clinical practice [31]. Thus, pharmacogenomic studies have clearly demonstrated that variability in drug response can in part be attributed to variability of the genetic background of an individual. Yet, it is unlikely that drug response variability is entirely captured by gene polymorphisms which can alter drug disposition. Systems medicine to unravel individual differences in drug response A complementary view is that drug response variability can also be attributed to individual variability in underlying molecular mechanisms involved in the progression of nephropathy. Indeed, it is known that type 2 diabetes is a heterogeneous disease involving multiple pathophysiological processes. Accordingly, recent studies have shown that panels of biomarkers that capture at least several of the pathophysiological processes improve prediction of DKD progression [32–34]. Drugs may interfere in one or more of the processes that drive DKD progression in an individual. At this level, biomarkers may help to gain insight, which specific processes are involved in progression of disease in an individual and whether a drug indeed impacts on that process. In other words, a biomarker can serve (i) as a proxy of a key mechanistic factor characterizing the presentation and progression of disease in a certain individual and (ii) as an indicator for interference of a drug with such process. To identify biomarkers that meet these two requirements, two components are needed. First, a detailed model characterizing the molecular mechanisms of DKD is required, and second, details on how these mechanisms interfere with a drug’s molecular mode of action is to be factored in. Studies have shown that an Omics-based profiling approach combining several sources of data (genomics, transcriptomics, metabolomics, proteomics) at the same time taking molecular interactions into consideration is a successful strategy to generate these components. First, a DKD model characterizing the pathways and processes of DKD progression was developed within the SysKid programme through integrating Omics profiling data from the genetic up to the metabolite level combined with scientific literature mining for deriving a set of molecular features associated with DKD. In total, >2000 molecular features have provided ample information for reconstruction of molecular processes and pathways being involved in diabetic nephropathy. Interpreted on the level of molecular pathways, the diabetic nephropathy model includes signalling processes such as PI3-Akt, chemokines and cytokines, fibrosis (TGFbeta signalling), ECM receptor interaction and components of


metabolic pathways. In an alternative representation resulting from mapping the DKD-associated molecular features on a protein coding gene-centric, hybrid interaction network followed by network segmentation provided a DKD molecular process model consolidating a subset of the associated molecular features in more than 20 molecular processes resembling aspects of the molecular background of DKD [35]. By utilizing a molecular process representation (be it on the level of molecular pathways or network process segments) biomarkers ( protein coding gene-level) can be selected that

are embedded in each pathway or process in order to develop a biomarker panel aimed at covering relevant aspects of the molecular landscape of DKD (in contrast to using a single biomarker approach representing only a single process or pathway). Forwarding such panel to experimental testing regarding explaining variance of e.g. eGFR decline allows identifying the subset of markers—hence molecular mechanisms—indeed afflicted with progressive disease. In addition, the contribution of different process units at early and late stages of disease can be assessed thereby allowing for

FULL REVIEW F I G U R E 2 : Molecular process model representation of DKD (A), mode of action of ramipril (B), (C) interference of ramipril mode of action with

the model representation of DKD, where overlapping elements of phenotype and drug are indicated in red, and zoom into a specific process segment of the DKD model (D). The molecular models encode clusters of protein coding genes (nodes given in grey) and their interactions according to an underlying hybrid interaction network. Interactions between segments (molecular processes) are omitted in display.


iv109

FULL REVIEW

delineating a model for the progression of disease at the molecular mechanism level. Second, to determine a drug’s impact on these various mechanisms of progressive DKD in a patient-specific context, the drug mechanism of action must be determined. To characterize the effects of a drug on the processes of DKD, not only the drug target(s) itself should be considered but a broader description of the mode of action of drugs of interest should be obtained. In analogy to modelling disease mechanisms, molecular features associated with a drug’s effect can be obtained by combining Omics profiling data together with literature mining. Using such drug-associated molecular data allows the construction of a molecular pathway or process landscape of a drug effect. Interference of the drug pathways with DKD progression pathways provide hypotheses if and what specific pathways in DKD are indeed addressed by the effect of a drug. Biomarkers representing such pathways can subsequently be selected for testing the predictive performance whether the individual will respond to the drug. Such drug mechanism of action molecular model was recently developed for ACEi, consolidating a number of specific drugs including ramipril, captopril or enalapril in a general ACEi model. In order to assess which of the processes of a general ACEi mode of action molecular model impact on the progression of DKD, the ACEi molecular network model was mapped on the DKD molecular process model. This resulted in the identification of five processes on the DKD side significantly addressed by ACEi molecular effect. Biomarkers represented in these five processes included amongst others the cytokine tumour growth factor β (TGF-β), the nuclear transcription factor NFκB and the chemokine C-C motif ligand 5 (CCL5). This study showed that ACEi impact on different molecular processes, many of them also involved in the progression of DKD. As specific example, scientific literature mining regarding molecular features associated with ramipril effect resulted in the identification of about 2200 features which led to

the development of a ramipril molecular process model consisting of 30 molecular processes including 866 protein coding genes (Figure 2). Interference of this drug model with a DKD model identifies a number of DKD-associated processes apparently affected by ramipril going beyond the direct drug target (ACE) context. A systematic analysis of the molecular processes involved in DKD coupled with a systemic analysis of the impact of ACEi on these molecular processes allows selection of a set of biomarker candidates serving for evaluating drug response. After experimental validation, such biomarkers can be used in future studies to assess which molecular processes operate at an individual patient level and whether the individual patient is more or less likely to respond to ACEi.

FUTURE DIRECTIONS Substantial Omics profiling data have become available over the last decade which after consolidation on interaction networks and pathways enable to study the intricate molecular composition of DKD, and when combined with the large body of evidence from reductionist approaches finally provides the opportunity of establishing data-driven workflows aimed at matching clinical disease presentation and drug effect via utilizing predictive biomarkers (Figure 3). Although the above-described workflow appears appropriate to develop biomarkers for individual drug prediction and monitoring, the molecular models and their biomarkers have to be validated in prospective clinical trials. To this end, biomarkers derived from the molecular models have to be measured in prospective studies and their ability to predict drug response has to be ascertained. Such future studies will also help to gain more insight into the molecular processes involved in responder and non-responder patients. In addition, interference of the DKD molecular network model with a drug mode of

F I G U R E 3 : A data-driven workflow for matching disease pathophysiology and drug mode of action via linked biomarkers. Step 1 resembles

consolidation of molecular data found as associated with a phenotype, in an iterative procedure used for deriving molecular process/pathway models. On this basis biomarker candidates are selected for serving as proxy of the process/pathway landscape (2), subsequently forwarded to experimental evaluation regarding association with disease progression. In analogy, molecular data found associated with a drug’s effect need consolidation (3) for allowing delineation of a drug mechanism of action representation on a process/pathway level. Central step defines the match of drug molecular effect and molecular processes seen afflicted with progressive disease (4), at this stage already knowing biomarkers allowing monitoring such interference in the specific disease molecular background.

iv110


action model can help to reposition already registered drugs from other disciplines to diabetes. Most of the registered drugs already have a large safety record which increases the likelihood of successful new developments and evidence-based treatments for DKD. Future in-silico modelling and prospective randomized clinical trials will provide proof-of-concept of this approach.

CONCLUSIONS The SysKid consortium has developed an integration scheme for providing enhanced insight into the underlying molecular mechanisms of individual renal disease progression and disease pathophysiology. Coupling this mechanistic representation to a model describing the effects of a drug on a molecular level provides the possibility to develop novel drug response biomarkers which will help to tailor optimal therapy to the individual patient.

AC K N O W L E D G E M E N T S

C O N F L I C T O F I N T E R E S T S TAT E M E N T B. Mayer is co-founder and managing partner, A. Heinzel and P. Perco are employees of emergentec biodevelopment GmbH, a company developing methods and computational platforms for in silico biomarker, target and drug screening. H.J. Lambers Heerspink is consultant for Astellas, Abbvie, J&J, Reata Pharmaceuticals, Z-Pharma (honoraria paid to his institution).

REFERENCES 1. Plantinga LC, Crews DC, Coresh J et al. Prevalence of chronic kidney disease in US adults with undiagnosed diabetes or prediabetes. Clin J Am Soc Nephrol 2010; 5: 673–682 2. Tao Z, Shi A, Zhao J. Epidemiological perspectives of diabetes. Cell Biochem Biophys 2015; doi: 10.1007/s12013-015-0598-4 3. Promotion National Center for Chronic Disease Prevention and Health Promotion. http://www.cdc.gov/diabetes/projects/pdfs/ckd_summary.pdf (3 March 2015, date last accessed) 2015 4. De Vecchi AF, Dratwa M, Wiedemann ME. Healthcare systems and endstage renal disease (ESRD) therapies—an international review: costs and reimbursement/funding of ESRD therapies. Nephrol Dial Transplant 1999; 14 Suppl 6: 31–41 5. Caskey FJ, Kramer A, Elliot RF et al. Global variation in renal replacement therapy for end-stage renal disease. Nephrol Dial Transplant 2011; 26: 2604–2610 6. Brenner BM, Cooper ME, De Zeeuw D et al. Effects of losartan on renal and cardiovascular outcomes in patients with type 2 diabetes and nephropathy. N Engl J Med 2001; 345: 861–869


iv111

FULL REVIEW

H.J.L.H. is supported by a VENI grant from the Netherlands Organization for Scientific Research. The work leading to this paper received funding from the European Community’s Seventh Framework Programme under grant agreement no. HEALTH–F2–2009–241544 (SysKid).

7. Lewis EJ, Hunsicker LG, Clarke WR et al. Renoprotective effect of the angiotensin-receptor antagonist irbesartan in patients with nephropathy due to type 2 diabetes. N Engl J Med 2001; 345: 851–860 8. Parving HH, Lehnert H, Brochner-Mortenson J et al. The effect of irbesartan on the development of diabetic nephropathy in patients with type 2 diabetes. N Engl J Med 2001; 345: 870–878 9. Komorowsky CV, Brosius FC, III, Feng Y et al. Perspectives on systems biology applications in diabetic kidney disease. J Cardiovasc Transl Res 2012; 5: 491–508 10. Mischak H, Delles C, Klein J et al. Urinary proteomics based on capillary electrophoresis-coupled mass spectrometry in kidney disease: discovery and validation of biomarkers, and clinical application. Adv Chronic Kidney Dis 2010; 17: 493–506 11. Turner RC, Holman RR, Cull CA et al. Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). UK Prospective Diabetes Study (UKPDS) Group. Lancet 1998; 352: 837–853 12. Patel A, MacMahon S, Chalmers J et al. Intensive blood glucose control and vascular outcomes in patients with type 2 diabetes. N Engl J Med 2008; 358: 2560–2572 13. Roscioni SS, Lambers Heerspink HJ, de Zeeuw D. The effect of RAAS blockade on the progression of diabetic nephropathy. Nat Rev Nephrol 2014; 10: 243 14. Andresdottir G, Jensen ML, Carstensen B et al. Improved survival and renal prognosis of patients with type 2 diabetes and nephropathy with improved control of risk factors. Diabetes Care 2014; 37: 1660–1667 15. Lambers Heerspink HJ, de Zeeuw D. Novel drugs and intervention strategies for the treatment of chronic kidney disease. Br J Clin Pharmacol 2013; 76: 536–550 16. Parving HH, Brenner BM, McMurray JJ et al. Cardiorenal end points in a trial of aliskiren for type 2 diabetes. N Engl J Med 2012; 367: 2204–2213 17. Fried LF, Emanuele N, Zhang JH et al. Combined Angiotensin inhibition for the treatment of diabetic nephropathy. N Engl J Med 2013; 369: 1892–1903 18. Kohan DE, Pollock DM. Endothelin antagonists for diabetic and nondiabetic chronic kidney disease. Br J Clin Pharmacol 2013; 76: 573–579 19. Mann JF, Green D, Jamerson K et al. Avosentan for overt diabetic nephropathy. J Am Soc Nephrol 2011; 21: 527–535 20. de Zeeuw D, Akizawa T, Audhya P et al. Bardoxolone methyl in type 2 diabetes and stage 4 chronic kidney disease. N Engl J Med 2013; 369: 2492–2503 21. Paul SM, Mytelka DS, Durwiddle CT et al. How to improve R&D productivity: the pharmaceutical industry’s grand challenge. Nat Rev Drug Discov 2010; 9: 203–214 22. Arrowsmith J. Trial watch: phase III and submission failures: 2007–2010. Nat Rev Drug Discov 2011; 10: 87 23. Chin MP, Wrolstad D, Bakris GL et al. Risk factors for heart failure in patients with type 2 diabetes mellitus and stage 4 chronic kidney disease treated with bardoxolone methyl. J Card Fail 2014; 20: 953–958 24. Hoekman J, Lambers Heerspink HJ, Viberti G et al. Predictors of congestive heart failure after treatment with an endothelin receptor antagonist. Clin J Am Soc Nephrol 2013; 9: 490–498 25. Heinzel A, Perco P, Mayer G et al. From molecular signatures to predictive biomarkers: modeling disease pathophysiology and drug mechanism of action. Front Cell Dev Biol 2014; 2: 37 26. Klonoff DC. Precision medicine for managing diabetes. J Diabetes Sci Technol 2015; 9: 3–7 27. Trusheim MR, Berndt ER, Douglas FL. Stratified medicine: strategic and economic implications of combining drugs and clinical biomarkers. Nat Rev Drug Discov 2007; 6: 287–293 28. Ingelman-Sundberg M, Rodriguez-Antona C. Pharmacogenetics of drugmetabolizing enzymes: implications for a safer and more effective drug therapy. Philos Trans R Soc Lond B Biol Sci 2005; 360: 1563–1570 29. Samer CF, Lorenzini KI, Rollason V et al. Applications of CYP450 testing in the clinical setting. Mol Diagn Ther 2013; 17: 165–184 30. Winkelmann BR, Russ AP, Nauck M et al. Angiotensinogen M235T polymorphism is associated with plasma angiotensinogen and cardiovascular disease. Am Heart J 1999; 137(4 Pt 1): 698–705

34. Niewczas MA, Sirich TL, Mathew AV et al. Uremic solutes and risk of endstage renal disease in type 2 diabetes: metabolomic study. Kidney Int 2014; 85: 1214–1224 35. Fechete R, Heinzel A, Soellner P et al. Using information content for expanding human protein coding gene interaction networks. J Comput Sci Syst Biol 2013; 6: 2 Received for publication: 11.3.2015; Accepted in revised form: 15.4.2015

FULL REVIEW

31. Taverne K, de Groot M, De Boer A et al. Genetic polymorphisms related to the renin-angiotensin-aldosterone system and response to antihypertensive drugs. Expert Opin Drug Metab Toxicol 2010; 6: 439–460 32. Roscioni SS, de Zeeuw D, Hellemons ME et al. A urinary peptide biomarker set predicts worsening of albuminuria in type 2 diabetes mellitus. Diabetologia 2012; 56: 259–267 33. Pena M, Heinzel A, Heinze G et al. A novel panel of biomarkers to predict eGFR decline in type 2 diabetes. PLoS One 2015; 10: e0120995

iv112
