Review Improving the efficiency of the development of drugs for stroke David W. Howells1,2*, Emily S. Sena1,2, Victoria O’Collins1, and Malcolm R. Macleod3
The mortality and morbidity associated with stroke makes the development of new drugs a research priority. Recent unsuccessful clinical trials have reduced enthusiasm for the development of neuroprotective drugs. Here, we use empirical evidence derived from systematic reviews of stroke drug development to identify stages of drug development which might be improved. We then propose exemplar strategies which may be helpful, along with some basic economic modelling of what the impact of such strategies might be. This suggests that relatively straightforward measures might reduce the costs of drug development by $5·8 bn or 31%.
Key words: neuroprotection, pharmacoecomonics, translation
Introduction What does stroke cost our communities? Estimates of stroke incidence and type and approaches to stroke management vary considerably from community to community, and so establishing an overall cost of stroke is difficult. In the developed world (population 1·2 bn), Sudlow and Warlow suggested an incidence of 300–500 per 100 000 population, or 3·6 m to 6 m strokes per year (1). The cost of stroke is high both because of the immediate healthcare costs and the long-term disability which often results. US data from 1990 suggest a lifetime cost of around
Correspondence: David W. Howells*, Florey Neuroscience Institutes, Melbourne Brain Centre, 245 Burgundy Street, Heidelberg, Victoria 3084, Australia. E-mail:
[email protected] 1 Florey Neuroscience Institutes, Melbourne Brain Centre, Heidelberg, Victoria, Australia 2 University of Melbourne, Melbourne Brain Centre, Heidelberg, Victoria, Australia 3 Centre for Clinical Brain Sciences, Division of Clinical Neurosciences, University of Edinburgh, Edinburgh, United Kingdom Conflict of interest: The authors have no relevant financial interests and specifically have no commercial interests relating to any of the approaches described. M. M. is Clinical Lead for Nervous Systems Disorders Research for Scotland. DOI: 10.1111/j.1747-4949.2012.00805.x
© 2012 The Authors. International Journal of Stroke © 2012 World Stroke Organization
US$100 000 (2); For Australia, Cadilhac et al. estimated 2004 costs of AU$57 000 (3) for ischaemic stroke, and more recent figures from the United Kingdom suggest treatment costs and productivity loss arising from stroke of US$95 000 (4). After adjusting for inflation, this gives a range of possible costs from US$74 000 to US$173 000, and a range of annual costs in the developed world from US$266 bn to US$1038 bn. For our purposes here, we take conservative estimates of incidence (300 per 100 000) and lifetime cost (US$74 000).
Existing treatments Four acute interventions are of proven benefit (5). These are intravenous thrombolysis, oral aspirin, patient care in a dedicated stroke unit, and, where indicated, decompressive hemicraniectomy. Unfortunately, the most effective treatments (thrombolysis and hemicraniectomy) are only appropriate for a minority of patients. There remains substantial unmet need in the treatment of acute stroke, and even relatively modest gains provided by translation of new research could have large benefits.
Developing new drugs for stroke The recurrent failure to develop effective stroke drugs and bring them to market has led some to reflect on the reasons for this translational failure, with many reaching the rather nihilistic conclusion that further development of neuroprotective strategies for stroke is at best naïve and at worse foolhardy. Reasons given to support this view include a belief that animal models of focal cerebral ischaemia do not recapitulate human disease with sufficient fidelity to be useful in drug development, and that neuroprotection simply ‘doesn’t work’ in humans. We believe that there may be a more plausible but less palatable explanation. In short, we believe: that the designs of animal experiments used to support drug development programmes give risk of bias sufficient to render results of some individual studies unreliable; that publication bias, small sample size, and selective citation of available data have falsely inflated the perception of efficacy in animal studies; that too little attention has been given to the importance of demonstrating the validity of drug targets to human disease; and that
Vol 7, July 2012, 371–377
371
Review
423
229
D. W. Howells et al.
277
15 96
1
3
Fig. 1 Estimated fate of 1026 drugs with some testing in experimental stroke: 423 drugs were tested in vitro but not in vivo; 229 tested in vivo did not improve outcome; 277 which did improve outcome in vivo had not been taken forward to clinical trial; 96 drugs with in vitro and in vivo development history did not improve outcome in clinical trial and one did; of 18 drugs without in vitro or in vivo development history, three improved outcome in clinical trial (8).
the design of clinical trials has paid too little attention to the need for drugs to reach their putative site of action within a time frame where there is a reasonable prospect of efficacy. In an attempt to develop an evidence-based approach to the investigation of translational failure, we and others have presented empirical evidence to support this alternative hypothesis of the causes of translational failure (6), and these will not be recapitulated in further detail here. It is premature to abandon animal models currently available (7), although it is clearly important that they be used appropriately, and that proof-of-concept clinical trials should test efficacy under conditions as close as possible to those in which efficacy is observed in animals. Here, our aim is to investigate whether costs of drug development for acute stroke might be reduced if additional intermediate strategies were used to exclude less promising drugs prior to clinical trial. So how might the application of existing models be improved? We have previously reported a systematic overview of the development of new drugs for stroke (8) which allowed estimates of the attrition rates at each stage of drug development. Sixty per cent of agents tested in vitro went on to animal testing; about two-thirds of these showed efficacy in animal models; about 100 drugs had been tested in clinical trial; only one drug tissue plasminogen activator (tPA) improved outcome in clinical trial (Fig. 1). The process through which the 100 drugs tested in clinical trial were selected from the 400 showing efficacy in animal stroke models was not clear. Using these data for attrition rates and estimates of the costs of different stages of drug development, we present a model describing the costs of drug development under different scenarios.
Assumptions made regarding current costs We have assumed the most conservative values for the incidence and lifetime costs of stroke from the ranges discussed
372
Vol 7, July 2012, 371–377
above, and we have assumed present attrition rates to reflect those observed in our previous systematic review (8). While the costs of initial development (including drug discovery, medicinal chemistry, intellectual property protection costs, and in vitro work) will vary greatly from compound to compound, for the purpose of this simulation, we have assumed average costs of US$10 m. In-house costs of animal experiments (excluding overhead costs and indirect salary costs) taken from a recent grant application are around US$1000 per animal; based on findings from our systematic reviews, we have assumed that perhaps 1000 animals would be used for each drug. For clinical trials, we have assumed costs for a 2000-patient industry-sponsored trial of US$40 m. Because two such trials would generally be needed to show efficacy for a licence to be issued, we have assumed a single study for all ineffective treatments and two trials for effective treatment.
Current costs of stroke drug development Using this approach, we estimate the total costs of bringing one effective stroke drug to market using the current approach to be $14·64 bn (Table 1). This comprises the costs of in vitro studies for 1000 compounds ($10 bn), of animal studies for 600 compounds ($600 m), and of 101 clinical trials (one neutral trial for each of 99 drugs, two positive trials for one drug). On the basis of the societal costs of stroke in the developed world outlined earlier ($266 bn), a treatment received by 15% of stroke patients (9) which reduced these costs by 10% through improvements in outcome would repay the costs of development in under four-years.
Current implied value of successful stroke drug development It is difficult to estimate with precision the value to a pharmaceutical company of developing an effective stroke drug, as this depends on a number of factors including costs of production, market share, and patent life. Simplistically, a drug given to 15% of stroke patients at a price of $5000 might bring revenues of $2·7 bn per year or $27 bn over 10 years. We might also seek inferences from how the market values a successful stroke drug. In 2006, the Stroke-Acute Ischemic NXY Treatment (SAINT) II trial was widely expected to confirm an efficacy for NXY-059. The day before publication of the (neutral) trial results, Astra Zeneca shares were trading at $66·37, but within a few days had fallen to $58·68 (10). At that time, there were slightly over 1·5 bn shares in circulation, and the fall in share value, representing the market valuation of a positive trial result, was $12 bn. Applying inflationary adjustment for Pharmaceutical Preparations Producer Prices Index from the US Bureau of Labor Statistics (11) gives a present value, that is to say the impact on the valuation of a company of having rather than not having a successful stroke drug, of $15·7bn. There is therefore a relatively small difference between the cost to industry of developing a successful new drug
© 2012 The Authors. International Journal of Stroke © 2012 World Stroke Organization
Review
D. W. Howells et al.
Table 1 Costs ($million) of various strategies to improve development pathway for neuroprotective drugs for stroke. Assumptions are given in the text
Strategy Current Drugs remaining
Human cell culture
Systematic review and Multicentre meta-analysis confirmation AR 99.9% AR 99.8% 1000 1000 1000
Combined AR 99.9% 1000
AR 99.8% 1000
1000
1000
$10m $10bn 40%
$10m $10bn
$10m $10bn
$10m $10bn
$10m $10bn
$10m $10bn
$10m $10bn
600
600
600
600
600
600
$100k $60m
$100k $60m
$100k $60m
450
450
450
In vitro studies cost per compound cost per stage attrition rate Drugs remaining
600
Human cell culture cost per compound cost per stage attrition rate
25%
Drugs remaining
Animal studies cost per compound cost per stage attrition rate Drugs remaining
$1m $600m 83% 100
$1m $450m
$1m $600m
$1m $600m
$1m $600m
$1m $450m
$1m $450m
75
100
100
100
75
75
$100k $10m
$100k $10m
$100k $8m
$100k $8m
75
75
56
56
$1m $100m
$1m $56m
$1m $56m
50
28
28
Systematic review/ meta analysis cost per compound cost per stage attrition rate
25%
Drugs remaining Multicentre confirmatory animal studies cost per compound cost per stage attrition rate
50%
Drugs remaining Clinical trial cost per compound cost per stage attrition rate
$40m $4bn
$40m $3bn
$40m $3bn
$40m $3.1bn
$40m $2bn
$40m $1.2bn
$40m $1.2bn
1
1
1
2
1
1
2
Total cost
$14.6bn
$13.6bn
$13.6bn
$6.8bn
$12.7bn
$11.7bn
$5.9bn
Value
$15.7bn
$15.7bn
$15.7bn
$15.7bn
$15.7bn
$15.7bn
$15.7bn
$1.1bn
$1.1bn
$7.9bn
$2bn
$2bn
$8.8bn
$2.1bn
$2.1bn
$8.9bn
$3bn
$4bn
$9.8bn
Drugs remaining
Saving v status quo Profit/ loss
$1.1bn
($14·6 bn) and the market valuation of such a drug ($15·7 bn), and this may explain the current lack of enthusiasm for stroke drug development. It is in the interests of government that such treatments should become available, and these figures emphasize the important roles for government funding first in the initial in vitro and preclinical in vivo research effort, and second in the support of clinical trials (both directly through the creation of research infrastructures, and indirectly through the promotion of an efficient regulatory environment).
© 2012 The Authors. International Journal of Stroke © 2012 World Stroke Organization
What aspects of the drug development pathway might be altered to reduce aggregate costs? The most expensive aspects of stroke drug development are in vitro development ($10 m, $10 bn per successful compound) and clinical trials ($40 m, $4 bn per successful compound). Where potentially effective drugs are excluded from further development because of falsely negative laboratory studies (i.e. the process is not sensitive to efficacy), then the in vitro development costs cannot be recouped. Equally, where drugs
Vol 7, July 2012, 371–377
373
Review which have little prospect of efficacy are taken to clinical trial (i.e. the process is not specific), then again industry is incurring costs which cannot be recouped. These losses might therefore be reduced if the sensitivity and specificity of drug development could be improved.
Sensitivity At present, most focal cerebral ischaemia experiments are powered at around 30% to detect a 30% improvement in outcome (CAMARADES, data on file). That is, out of every three experiments testing an effective drug, we can expect two experiments to be neutral simply because the experiment was not large enough. Effect sizes in well-conducted studies are likely to be lower, but that in itself should not preclude their further development; even a 10% improvement in outcome would be important if it translated to similar human benefit, and in animal studies, tPA is one of the least effective interventions tested (12). However, where effect sizes are smaller, much larger numbers of animals are required; to detect a 10% improvement in outcome with 80% power would require 250 animals per group, and it is likely that no individual lab would have the capacity to conduct such an experiment within a reasonable timescale. Specificity It is highly likely that study quality (a failure to randomize and to blind treatment) and design (inappropriately short delays to treatment, failure to use animals with comorbidity) have led to the overstatement of efficacy in individual studies (13,14), and that publication bias has resulted in overstatements of efficacy in research summaries (15). Conducting animal studies to standards now recommended (6,16,17) is likely to lead to higher rates of attrition of stroke therapies prior to clinical trial (18). Improving the sensitivity and specificity of drug development The translational difficulties outlined earlier have discouraged the government and the pharmaceutical industry from investing in stroke drug development. However, the costs of drug development (and therefore the risks to government and the pharmaceutical industry) might be reduced if we were to able to develop intermediate steps to exclude more drugs prior to clinical trial and if we could reduce the likelihood that truly effective drugs were inappropriately excluded during the development process. Here, we focus on three specific strategies to achieve this.
The use of experiments with human tissue to confirm efficacy (19) One concern has been that some drugs have been tested in animal models of stroke without demonstration of efficacy in
374
Vol 7, July 2012, 371–377
D. W. Howells et al.
human tissues. While most pathophysiological pathways identified in, for instance, rodents have some homologue in humans, this may not always be the case. Until recently, the availability of human tissue has been limited either to human tumour cell lines or to tissue removed, for instance, during epilepsy or tumour surgery. However, the development of human neuronal stem cells which can be grown in large number, which are capable of undergoing terminal differentiation and which have some characteristics of mature human neurons now affords the opportunity to test drugs in vitro in human cells before embarking on an in vivo research program. We estimate that the costs of testing one drug in this way would be at most $100k, and that perhaps one in four drugs tested in this way might be excluded from further development.
The use of systematic review and meta-analysis to ascertain the internal and external validity of animal work presented (20) A series of systematic reviews and meta-analyses conducted by ourselves and others have shown limitations in the internal and external validity of the preclinical evidence supporting a number of drugs which had nonetheless been tested in clinical trial. With the benefit of hindsight it seems likely that, had we known then what we know now, trials of these compounds might not have gone ahead, at least in the form that they did. Equally, this approach has also been used to target further animal work required and, in the case of hypothermia, to aid clinical trial design (6). We estimate that the costs of conducting a systematic review and meta-analysis of a candidate drug would be at most $100k, and again that one in four drugs tested in this way might be excluded from further development. The use of confirmatory preclinical multicentre animal studies (21) Post hoc analyses of the animal data supporting the efficacy of a number of drugs which had failed in clinical trial suggest that there may be difficulties with the conduct of such experiments, with the numbers of animals used, with the range of circumstances under which efficacy was observed in animals, and with the extent to which this was reflected in clinical trial design. The animal datasets for hypothermia (22) and tPA (12) each include thousands of animals, but for others (NXY-059 (13), melatonin (23), tirilazad (14)), the datasets are smaller. Importantly, a cumulative meta-analysis of the efficacy of tPA in animal models (12) showed a stable effect size after the inclusion of around 1500 animals. One approach to increasing the size of animal experiments is to conduct the same experiment across a number of sites. Central randomization, outcome adjudication, and quality control can be used to improve validity (by ensuring randomization, concealment of allocation sequence, and the blinded assessment of outcome) and to ensure compliance with the
© 2012 The Authors. International Journal of Stroke © 2012 World Stroke Organization
D. W. Howells et al.
Review
Fig. 2 Effect of different strategies on drugs remaining at each development stage: Black bars represent number of drugs entering that phase and grey bars represent number of drugs taken to the next stage. (a) conventional approach, with selection at in vitro efficacy (1), in vivo efficacy (3) and clinical trial (6); (b) additional selection stage of human neuronal cell culture testing (2) following other in vitro testing; (c) additional selection stage of systematic review and meta-analysis (4) following conventional in vivo testing; (d) additional selection stage (5) of multicentre animal studies prior to clinical trial; (e) effect of measures (b) to (d) in combination. In each case, the final number of drugs remaining after clinical trial is one.
© 2012 The Authors. International Journal of Stroke © 2012 World Stroke Organization
Vol 7, July 2012, 371–377
375
Review study protocol (by external monitoring), and therefore will help avoid bias. This approach would also allow a robust exploration of the limits to efficacy (time to treatment, age of animal, efficacy in hypertensive or diabetic animals) which would allow interventions with a narrow window of therapeutic efficacy to be excluded at this stage. In these circumstances, efficacy is likely to be smaller, and only larger studies – such as would be possible using a multicentre approach – would have appropriate power to detect small but translationally significant effects. This multicentre approach has a number of additional advantages, not least the capacity to deliver large datasets quickly and, by showing efficacy in a number of different settings, to give some confidence in the generalizability of the findings. We estimate that, with infrastructure costs, this would require perhaps $2000 per animal for an additional 500 animals, or $1m per drug. We have previously shown that even efficacy suggested in systematic review may not be confirmed in high quality testing (18). Here we take what we believe to be a conservative estimate, that one in two interventions with efficacy in prior in vivo studies would not have this confirmed in robust high quality studies under circumstances (e.g. time to treatment, efficacy in context of hypertension) that would support progression to clinical trial.
Potential impact of these strategies on the costs of stroke drug development Here, we have made what we believe to be reasonable estimates of the costs of these strategies and their contribution to attrition rates, and used this to estimate their potential impact (Fig. 2). On the basis of O’Collins et al. (8), we have assumed a fixed attrition rate of 99·9%, that is to say that for every 1000 compounds entering drug development for stroke, one will be found to have efficacy in well-conducted clinical trials. However, if systematic review and meta-analysis were to lead to better clinical trial design (by allowing interventions to be tested in humans under similar circumstances to which efficacy was seen in animals), then this attrition rate might improve, and so we have also modelled a 99·8% attrition rate in those processes which included a systematic review and meta-analysis. Using this model, we estimate net cost savings from human cell culture studies of $1·1 bn, from systematic review and meta-analysis of $1·0 bn (with no improvement in attrition rate) or $7·8 bn (with attrition rate of 99·8%), and from multicentre animal studies of $1·9 bn. Taken together, the approaches outlined earlier would give estimated cost savings of $2·9 bn (with no change in attrition rate) or of $8·8 bn (with an improvement in attrition rate). Even under the conservative estimate of no improvement in attrition rates, the cost of development is almost $4 bn less than the market valuation of an effective stroke drug.
376
Vol 7, July 2012, 371–377
D. W. Howells et al.
Limitations of this approach Our approach has necessarily required estimation of factors which are not precisely known. Because we have taken conservative estimates of the current number and costs of stroke, and of the potential impact of the measures we describe, it is likely that our findings understate the potential impact of such measures. Our financial model is simplistic in that we take no account in the discounting of future value either of profits or of health gains. Our assumption that a company’s share price can be used to infer the valuation of a company, or that the valuation of a successful stroke drug was entirely captured by the immediate fall in the share price of a single company, is also likely to have led to some imprecision. Furthermore, we have assumed that stroke drug development is a linear process which proceeds from in vitro through in vivo to clinical trial, and this might not always be the case. These weakness are however unlikely to impact on our core predictions of the impact of these additional strategies on the economics of drug development. Importantly, these findings are cumulative, and would hold for any intermediate steps which reduced the number of ultimately ineffective interventions moving forward to the next step in development; or which reduced the number of effective interventions wrongly excluded at earlier steps in development.
Conclusion The nihilism surrounding stroke drug development is justified because of a high attrition rate of drugs in clinical trial. Developing intermediate strategies to exclude drugs prior to clinical trial (in our modelling based on human stem cell culture, systematic review and meta-analysis, and multicentre animal studies) has the potential to reduce the costs of developing an effective drug by around $5·8 bn or 31%. However, one of the most telling predictions from our model, even with the most conservative estimates of stroke incidence and lifetime costs, is that the societal costs of stroke would be recouped within a few years of such treatment becoming available. Given the economic benefits which are also likely to accrue from the development of a successful drug for stroke, we believe that this makes a powerful case for increased government support at all stages of the drug development pathway. As well as supporting basic science research and the in vitro work on which such progress will be built, and investing in the clinical research infrastructure which will facilitate the clinical trials required, the government might also have a crucial role in providing an infrastructure to develop capacity in the systematic review and meta-analysis of data from animal studies and in the infrastructure and regulatory framework required for multicentre animal studies.
References 1 Sudlow CL, Warlow CP. Comparable studies of the incidence of stroke and its pathological types: results from an international collaboration. International Stroke Incidence Collaboration. Stroke 1997; 28:491–9.
© 2012 The Authors. International Journal of Stroke © 2012 World Stroke Organization
D. W. Howells et al.
2 Taylor TN, Davis PH, Torner JC, Holmes J, Meyer JW, Jacobson MF. Lifetime cost of stroke in the United States. Stroke 1996; 27:1459– 66. 3 Cadilhac DA, Carter R, Thrift AG, Dewey HM. Estimating the longterm costs of ischemic and hemorrhagic stroke for Australia: new evidence derived from the North East Melbourne Stroke Incidence Study (NEMESIS). Stroke 2009; 40:915–21. 4 Saka O, McGuire A, Wolfe C. Cost of stroke in the United Kingdom. Age Ageing 2009; 38:27–32. 5 Donnan GA, Fisher M, Macleod M, Davis SM. Stroke. Lancet 2008; 371:1612–23. 6 van der Worp HB, Howells DW, Sena ES et al. Can animal models of disease reliably inform human studies? PLoS Med 2010; 7:e1000245. 7 Howells DW, Porritt MJ, Rewell SS et al. Different strokes for different folks: the rich diversity of animal models of focal cerebral ischemia. J Cereb Blood Flow Metab 2010; 30:1412–31. 8 O’Collins VE, Macleod MR, Donnan GA, Horky LL, van der Worp BH, Howells DW. 1026 experimental treatments in acute stroke. Ann Neurol 2006; 59:467–77. 9 Gilligan AK, Thrift AG, Sturm JW, Dewey HM, Macdonell RA, Donnan GA. Stroke units, tissue plasminogen activator, aspirin and neuroprotection: which stroke intervention could provide the greatest community benefit? Cerebrovasc Dis 2005; 20:239–44. 10 AstraZeneca. 2012. Available at http://www.astrazeneca.com/ Investors/Investor-tools/Share-price-download (accessed 3 January 2012). 11 U.S. Bureau of Labor Statistics. 2012. Available at http://data.bls.gov/ pdq/SurveyOutputServlet;jsessionid=A5DAD240C2CD43F9DB 400224D6E54E9C.tc_instance4 (accessed 3 January 2012). 12 Sena ES, Briscoe CL, Howells DW, Donnan GA, Sandercock PA, Macleod MR. Factors affecting the apparent efficacy and safety of tissue plasminogen activator in thrombotic occlusion models of stroke: systematic review and meta-analysis. J Cereb Blood Flow Metab 2010; 30:1905–13.
© 2012 The Authors. International Journal of Stroke © 2012 World Stroke Organization
Review 13 Macleod MR, van der Worp HB, Sena ES, Howells DW, Dirnagl U, Donnan GA. Evidence for the efficacy of NXY-059 in experimental focal cerebral ischaemia is confounded by study quality. Stroke 2008; 39:2824–9. 14 Sena E, Wheble P, Sandercock P, Macleod M. Systematic review and meta-analysis of the efficacy of tirilazad in experimental stroke. Stroke 2007; 38:388–94. 15 Sena ES, van der Worp HB, Bath PM, Howells DW, Macleod MR. Publication bias in reports of animal stroke studies leads to major overstatement of efficacy. PLoS Biol 2010; 8:e1000344. 16 Macleod MR, Fisher M, O’Collins V et al. Good laboratory practice: preventing introduction of bias at the bench. Stroke 2009; 40:e50–2. 17 Kilkenny C, Browne WJ, Cuthill IC, Emerson M, Altman DG. Improving bios cience research reporting: the ARRIVE guidelines for reporting animal research. PLoS Biol 2010; 8:e1000412. 18 O’Collins VE, Macleod MR, Cox SF et al. Preclinical drug evaluation for combination therapy in acute stroke using systematic review, metaanalysis, and subsequent experimental testing. J Cereb Blood Flow Metab 2011; 31:962–75. 19 Donnan GA. The 2007 Feinberg lecture: a new road map for neuroprotection. Stroke 2008; 39:242–8. 20 Sena E, van der Worp HB, Howells D, Macleod M. How can we improve the pre-clinical development of drugs for stroke? Trends Neurosci 2007; 30:433–9. 21 Bath PM, Macleod MR, Green AR. Emulating multicentre clinical stroke trials: a new paradigm for studying novel interventions in experimental models of stroke. Int J Stroke 2009; 4:471–9. 22 van der Worp HB, Sena ES, Donnan GA, Howells DW, Macleod MR. Hypothermia in animal models of acute ischaemic stroke: a systematic review and meta-analysis. Brain 2007; 130(Pt 12):3063–74. 23 Macleod MR, O’Collins T, Horky LL, Howells DW, Donnan GA. Systematic review and meta-analysis of the efficacy of melatonin in experimental stroke. J Pineal Res 2005; 38:35–41.
Vol 7, July 2012, 371–377
377
