- Email: [email protected]
Difficulties in replication of results
CORRESPONDENCE
Academia and industry Sir—Peter Wilmshurst (July 22, p 338)1 comments that payments seem to influence some cardiology consultants in their opinion of the effectiveness of certain drugs. This statement is brave and honest about what is a common issue and a marketing strategy on the part of the pharmaceutical industry. The receipt of payments is a major concern but of equal concern are the other forms of influence that industry has on consultants. For example, a public relations company, Key Communications, describes the brief given to them by Novo Nordisk when a group of patients started a class action against them because of adverse reactions to genetically engineered human insulin (see www.keycommunications.co.uk, accessed on Sept 18, 2000). The brief was to defend the safety profile of genetically engineered human insulin and they describe an issue and crisis management programme spanning 3 years, during which time “UK medical spokespeople were media trained”. In the results of this strategy, Key Communications say that Novo’s reputation remained intact among patients, and that medical professionals accepted that human insulin has an excellent safety profile. This insulin was given its licence quickly in 1982. The premarketing trials were small and the drug was subjected to no formal postmarketing surveillance. Research has still not clearly proved its total safety, but most of the studies can be strongly criticised because they did not test normal-use conditions and were not sufficiently large or long-term. Thousands of patients in many different countries have reported sideeffects with this form of insulin. It is worrying, therefore, that, in addition to the type of payments to which Wilmshurst refers, consultants are used in this way by industry. The media campaign commissioned by Novo Nordick and the payments described by Wilmhurst may greatly influence doctors’ prescribing of drugs, and these activities are questionable. We wonder why opinion-leaders in the field of diabetes do not give due weight to the frequent adverse reactions to human insulin; maybe the answer is commercial. Commercial, academic narrow-sightedness or not, it clearly confirms that policy based on academic research should not be accepted as sacrosanct or as complete evidence. At a time when the public’s confidence in the medical profession is
THE LANCET • Vol 356 • October 14, 2000
at an all time low, we suggest that the MCA assesses the whole issue of academia and industry and of payments made to consultants and its experts. However in doing so, we also suggest that the MCA adopts the Secretary of State for Health’s approach of directness, openness, and clarity rather than its normal methods of secrecy. *M R Kiln, J M D Hirst Insulin Dependent Diabetes Trust, PO Box 294, Northampton NN1 4XS, UK 1
Wilmshurst P. Academia and industry. Lancet 2000; 356: 338–39.
social consequences.3 The health benefit for society of a change from tobacco to cannabis is, therefore likely to be small, if not non-existent. The implication of the comparison of the behaviour of the addicted individual with that of society is that the debate will be more effective in terms of health gain if all psychoactive drugs are considered together rather than dealing with one drug at a time. Robert Cohen Drug Dependency Unit, Homerton Hospital, London E9 6SR, UK 1
2
Pitfall in the drugs debate 3
Sir—Two Aug 12 news items1,2 on the same page illustrate a worrying trend in societal attitudes to drugs. The first item describes how the WHO is trying to lower tobacco consumption by the use of tax increases and awareness of tobacco industry tactics1 and the next reports the use of a human-rights argument in a Canadian court to justify a demand to legalise the use of cannabis.2 This exemplifies how current attitudes throughout the world towards tobacco are hardening, whereas attitudes towards cannabis are softening. Clinicians who treat drug addicts will note how patients who are in the depths of their addiction will praise one drug while castigating another (“I’m reducing my heroin use, but my cannabis use is not a problem”; “I use heroin, but I would never touch alcohol, that really harms you”). Such patients, who show the psychological process of splitting in this way rarely do well and even when they seem to progress in the treatment of their drugs of choice, are especially prone to relapse. Only when patients recognise that the principle of using any psychoactive agent to deal with more complex problems is ineffective and stop the use of all psychoactive drugs do they recover and improve their lifestyle (in the patient’s terms). The approach of society, if it continues to behave in this ambivalent way towards different drugs, will achieve no more than to replace tobacco companies with cannabis companies. In pharmacological terms, there is no such thing as “soft” or “hard” drugs—ie, the binding of a cannabinoid to a cannabinoid receptor is not “softer” than the binding of an opiate to an opioid receptor) and chronic cannabis use has been associated with adverse health and
Kapp C. WHO calls for tobacco tax increases and warns against industry tactics. Lancet 2000; 356: 575. Kondro W. Canadian court tells government to relax marijuana laws for patients. Lancet 2000; 356: 575. Hall W, Solowij N. Adverse effects of cannabis. Lancet 1998; 352: 1611–16.
Difficulties in replication of results Sir—Johannes Schumacher and colleagues (Aug 5, p 506),1 with the genetic markers reported in our study (May 6, p 1615)2 cannot predict clinical response to the antipsychotic drug clozapine. They conclude that their inability to repoduce our results is due to false-positives or false-negative results or to unnoticed phenotypic differences. Replication of findings of association studies is frequently difficult because of a false-positive finding in the initial study. In some cases, however, such a result can be attributed to phenotypic differences in the replicating samples. We have reported an association between a 5-HT2A 102-T/C genetic variant and clozapine response. Workers in several studies, including one from Schumacher’s study, could not replicate this finding. However, a meta-analysis of published studies showed a strong association between the 102-T/C polymorphism and clozapine response.3 Important sample differences (ie, method of response assessment, duration of treatment, ethnic origin) can lead to discrepancies between studies. Schumacher and colleagues’ sample of patients include those who had been taking clozapine for 28 days. Only 35–50% of patients who finally respond to clozapine improve in the first 4 weeks.4,5 Later assessment of patients might lead to the reclassification of some nonresponders as responders. By contrast, we assessed patients on long-term
1359
For personal use only. Not to be reproduced without permission of The Lancet.
CORRESPONDENCE
clozapine treatment (mean duration >1 year). Schumacher and colleagues’ conclusion that our prediction model should be replicated and validated is correct. Replication should however, be done in a clinically similar sample. A strategy to clarify disagreement between studies could be the comparison of extreme phenotypes (ie, very good versus very bad responders), which clarifies apparent discrepancies between association studies.3 *M J Arranz, J Munro, S Osborne, D Collier, R W Kerwin *Clinical Neuropharmacology, Department of Psychological Medicine, Institute of Psychiatry, London SE5 8AF, UK (e-mail: [email protected]) 1
2
3
4
5
Schumacher J, Schulze TG, Wienker TF. Rietschel M, Nöthen MM. Pharmacogenetics of clozapine response. Lancet 2000; 356: 506–07. Arranz MJ, Munro K, Birkett J, et al. Pharmacogenetic prediction of clozapine response. Lancet 2000; 355: 1615–16. Arranz MJ, Munro J, Sham P, et al. Metaanalysis of studies on genetic variation in 5-HTA receptors and clozapine response. Schiz Res 1998; 32: 93–99. Miller DD, Fleming F, Holman TL, Perry PJ. Plasma clozapine concentrations as a predictor of clinical response: a followup study. J Clin Psychiatry 1994; 55: 117–21. Leiberman JA, Safferman AZ, Pollack S, et al. Clinical effects of clozapine in chronic schizophrenia: response to treatment and predictors of outcome. Am J Psychiatry 1994; 151: 1744–52.
Aristotle and the human genome project Sir—The completion of the sequencing of a human genome is deemed a major breakthrough in science. Scientists hope that knowledge of the genetic code will improve understanding of disease causes, provide options for genetic medical intervention, and cure disease. This enthusiastic line of thinking has been criticised as overoptimistic about the quantitative causal contribution of hereditary factors of the susceptibility to disease1 and as reductionist about social levels of disease cause.2 A more subtle conceptual point deserves attention. I purport that the human genome project is associated with a qualitative change in causality and, therefore, demands a shift in causal thinking. Modern notions, dating back to Isaac Newton, reduce causality to a functional relation and are characterised by the ideas of causal chains, causal complexes, and deterministic or probabalistic
1360
associations of events.3,4 Aristotelian metaphysics offers a different understanding of causes, and could contribute to the understanding of the causal role of genes. Functional causal relations relate to Aristotle’s system of effective cause.5 This type of cause is regarded as an external cause, as is the so-called final cause, or the intention behind an action. An example for external cause is genitourinary infection with human papillomavirus, which may cause cervical cancer. The virus’ strive for procreation could be interpreted as final cause. Aristotle supplemented these external causes with two internal causes—the formal cause, which is the blueprint that becomes manifest in a given substance, and the material cause. For human papillomavirus infection, therefore, the formal cause is the malignant transformation of cervical tissue, and the material cause is the affected cervical tissue itself. The two external and two internal causes are important for disease; the virus (effective cause), its drive to procreate (final cause), the cervix uteri (material cause), and its potential for malignantcell transformation (formal cause). Current understanding of disease cause is linked closely with external causation, especially the effective cause. The human genome project will shift causal research to internal causes: to the structure of tissue and cells (the material cause) and to the programmes of reaction in response to stimuli (the formal cause of events). Medical interventions derived from gene sequencing might, therefore, be directed towards structural vulnerability (eg, membrane proteins), or messengers of pathophysiological reactions. These powerful options have ethical consequences since the manipulation of internal causes affects personal rights more than an intervention directed against external causes.
Physics and chemistry Nobel Laureates and medicine Sir—Tonse Raju (July 29, p 436)1 completed a superb series recounting and discussing Nobel prize winners in physiology or medicine.1 However, as has been pointed out,2 the award committee seems to have been giving the prize more frequently for basic science discoveries, rather than those with more direct clinical implications, despite many potentially worthy recipients.3 Conversely, the impact of the work of many laureates in physics and chemistry has perhaps been even greater on the diagnosis and treatment of disease than that of some recipients’ work for the prize in physiology or medicine. For example, in physics: W C Roentgen, X-rays, 1901; A H Becquerel, M Cuire, and P Curie, radioactivity, 1903; F Bloch and E Purcell, nuclear magnetic resonance, 1952; W Shockley, J Bardeen, and W Brattain, the transistor effect (computers), 1956; C Townes, N Basov, and A Prokhorov, lasers, 1964; and E Ruska, electron microscopy, 1986. In chemistry: L Pauling, ␣-helix, -sheet, 1954; W Gilbert and F Sanger, DNA sequences, 1980; K Mullis, PCR, 1993. *Eric Lewin Altschuler, Robert Charlton Brain and Perception Laboratory, University of California, San Diego, La Jolla, CA 92093-0109, USA (e-mail: [email protected]) 1
2
3
Raju TNK. The Nobel chronicles— the first century. Lancet 2000; 356: 436. Narrowness of Nobel awards for physiology or medicine. Lancet 1999; 354: 1399. Marshall BJ, Warren JR. Unidentified curved bacilli in the stomach of patients with gastritis and peptic ulceration. Lancet 1984; 1: 1311–15.
Manfred Wildner Bavarian Public Health Research Center and Institute for Medical Informatics, Biometry and Epidemiology, Ludwig-Maximilians-University Munich, Germany (e-mail: [email protected]) 1
2
3
4
5
Lichtenstein P, Holm N, Verkasalo P, et al. Environmental and heritable factors in the causation of cancer. N Engl J Med 2000; 343: 78–85. Diez-Roux A. On genes, individuals, society, and epidemiology. Am J Epidemiol 1998; 148: 1027–32. Wulff H, Pedersen A, Rosenberg R. Philosphy of medicine. Oxford: Blackwell Scientific, 1990. Suppes P. A probalistic theory of causation. Amsterdam: North-Holland Publishing, 1970. Ulfig A. Lexikon der philosophischen Begriffe. Wiesbaden: Fourier, 1999.
DEPARTMENT OF ERROR Control of legionella in drinking water—In this Correspondence letter by John Hayes (Aug 26, p 773), the second sentence of the first paragraph should be, “The ionisation equipment of University Hospital, Bergmannsheil, Bochum, Germany, was correctly installed by our own engineers, and initially set to give the correct concentrations of silver and copper ions within the hot water circulating loops”. In the next sentence the concentrations of silver and copper should be 40 g/L and 400 g/L, respectively. In the last sentence of the letter the concentrations of silver should be between 20 g/L and 40 g/L.
THE LANCET • Vol 356 • October 14, 2000
For personal use only. Not to be reproduced without permission of The Lancet.
Academia and industry Sir—Peter Wilmshurst (July 22, p 338)1 comments that payments seem to influence some cardiology consultants in their opinion of the effectiveness of certain drugs. This statement is brave and honest about what is a common issue and a marketing strategy on the part of the pharmaceutical industry. The receipt of payments is a major concern but of equal concern are the other forms of influence that industry has on consultants. For example, a public relations company, Key Communications, describes the brief given to them by Novo Nordisk when a group of patients started a class action against them because of adverse reactions to genetically engineered human insulin (see www.keycommunications.co.uk, accessed on Sept 18, 2000). The brief was to defend the safety profile of genetically engineered human insulin and they describe an issue and crisis management programme spanning 3 years, during which time “UK medical spokespeople were media trained”. In the results of this strategy, Key Communications say that Novo’s reputation remained intact among patients, and that medical professionals accepted that human insulin has an excellent safety profile. This insulin was given its licence quickly in 1982. The premarketing trials were small and the drug was subjected to no formal postmarketing surveillance. Research has still not clearly proved its total safety, but most of the studies can be strongly criticised because they did not test normal-use conditions and were not sufficiently large or long-term. Thousands of patients in many different countries have reported sideeffects with this form of insulin. It is worrying, therefore, that, in addition to the type of payments to which Wilmshurst refers, consultants are used in this way by industry. The media campaign commissioned by Novo Nordick and the payments described by Wilmhurst may greatly influence doctors’ prescribing of drugs, and these activities are questionable. We wonder why opinion-leaders in the field of diabetes do not give due weight to the frequent adverse reactions to human insulin; maybe the answer is commercial. Commercial, academic narrow-sightedness or not, it clearly confirms that policy based on academic research should not be accepted as sacrosanct or as complete evidence. At a time when the public’s confidence in the medical profession is
THE LANCET • Vol 356 • October 14, 2000
at an all time low, we suggest that the MCA assesses the whole issue of academia and industry and of payments made to consultants and its experts. However in doing so, we also suggest that the MCA adopts the Secretary of State for Health’s approach of directness, openness, and clarity rather than its normal methods of secrecy. *M R Kiln, J M D Hirst Insulin Dependent Diabetes Trust, PO Box 294, Northampton NN1 4XS, UK 1
Wilmshurst P. Academia and industry. Lancet 2000; 356: 338–39.
social consequences.3 The health benefit for society of a change from tobacco to cannabis is, therefore likely to be small, if not non-existent. The implication of the comparison of the behaviour of the addicted individual with that of society is that the debate will be more effective in terms of health gain if all psychoactive drugs are considered together rather than dealing with one drug at a time. Robert Cohen Drug Dependency Unit, Homerton Hospital, London E9 6SR, UK 1
2
Pitfall in the drugs debate 3
Sir—Two Aug 12 news items1,2 on the same page illustrate a worrying trend in societal attitudes to drugs. The first item describes how the WHO is trying to lower tobacco consumption by the use of tax increases and awareness of tobacco industry tactics1 and the next reports the use of a human-rights argument in a Canadian court to justify a demand to legalise the use of cannabis.2 This exemplifies how current attitudes throughout the world towards tobacco are hardening, whereas attitudes towards cannabis are softening. Clinicians who treat drug addicts will note how patients who are in the depths of their addiction will praise one drug while castigating another (“I’m reducing my heroin use, but my cannabis use is not a problem”; “I use heroin, but I would never touch alcohol, that really harms you”). Such patients, who show the psychological process of splitting in this way rarely do well and even when they seem to progress in the treatment of their drugs of choice, are especially prone to relapse. Only when patients recognise that the principle of using any psychoactive agent to deal with more complex problems is ineffective and stop the use of all psychoactive drugs do they recover and improve their lifestyle (in the patient’s terms). The approach of society, if it continues to behave in this ambivalent way towards different drugs, will achieve no more than to replace tobacco companies with cannabis companies. In pharmacological terms, there is no such thing as “soft” or “hard” drugs—ie, the binding of a cannabinoid to a cannabinoid receptor is not “softer” than the binding of an opiate to an opioid receptor) and chronic cannabis use has been associated with adverse health and
Kapp C. WHO calls for tobacco tax increases and warns against industry tactics. Lancet 2000; 356: 575. Kondro W. Canadian court tells government to relax marijuana laws for patients. Lancet 2000; 356: 575. Hall W, Solowij N. Adverse effects of cannabis. Lancet 1998; 352: 1611–16.
Difficulties in replication of results Sir—Johannes Schumacher and colleagues (Aug 5, p 506),1 with the genetic markers reported in our study (May 6, p 1615)2 cannot predict clinical response to the antipsychotic drug clozapine. They conclude that their inability to repoduce our results is due to false-positives or false-negative results or to unnoticed phenotypic differences. Replication of findings of association studies is frequently difficult because of a false-positive finding in the initial study. In some cases, however, such a result can be attributed to phenotypic differences in the replicating samples. We have reported an association between a 5-HT2A 102-T/C genetic variant and clozapine response. Workers in several studies, including one from Schumacher’s study, could not replicate this finding. However, a meta-analysis of published studies showed a strong association between the 102-T/C polymorphism and clozapine response.3 Important sample differences (ie, method of response assessment, duration of treatment, ethnic origin) can lead to discrepancies between studies. Schumacher and colleagues’ sample of patients include those who had been taking clozapine for 28 days. Only 35–50% of patients who finally respond to clozapine improve in the first 4 weeks.4,5 Later assessment of patients might lead to the reclassification of some nonresponders as responders. By contrast, we assessed patients on long-term
1359
For personal use only. Not to be reproduced without permission of The Lancet.
CORRESPONDENCE
clozapine treatment (mean duration >1 year). Schumacher and colleagues’ conclusion that our prediction model should be replicated and validated is correct. Replication should however, be done in a clinically similar sample. A strategy to clarify disagreement between studies could be the comparison of extreme phenotypes (ie, very good versus very bad responders), which clarifies apparent discrepancies between association studies.3 *M J Arranz, J Munro, S Osborne, D Collier, R W Kerwin *Clinical Neuropharmacology, Department of Psychological Medicine, Institute of Psychiatry, London SE5 8AF, UK (e-mail: [email protected]) 1
2
3
4
5
Schumacher J, Schulze TG, Wienker TF. Rietschel M, Nöthen MM. Pharmacogenetics of clozapine response. Lancet 2000; 356: 506–07. Arranz MJ, Munro K, Birkett J, et al. Pharmacogenetic prediction of clozapine response. Lancet 2000; 355: 1615–16. Arranz MJ, Munro J, Sham P, et al. Metaanalysis of studies on genetic variation in 5-HTA receptors and clozapine response. Schiz Res 1998; 32: 93–99. Miller DD, Fleming F, Holman TL, Perry PJ. Plasma clozapine concentrations as a predictor of clinical response: a followup study. J Clin Psychiatry 1994; 55: 117–21. Leiberman JA, Safferman AZ, Pollack S, et al. Clinical effects of clozapine in chronic schizophrenia: response to treatment and predictors of outcome. Am J Psychiatry 1994; 151: 1744–52.
Aristotle and the human genome project Sir—The completion of the sequencing of a human genome is deemed a major breakthrough in science. Scientists hope that knowledge of the genetic code will improve understanding of disease causes, provide options for genetic medical intervention, and cure disease. This enthusiastic line of thinking has been criticised as overoptimistic about the quantitative causal contribution of hereditary factors of the susceptibility to disease1 and as reductionist about social levels of disease cause.2 A more subtle conceptual point deserves attention. I purport that the human genome project is associated with a qualitative change in causality and, therefore, demands a shift in causal thinking. Modern notions, dating back to Isaac Newton, reduce causality to a functional relation and are characterised by the ideas of causal chains, causal complexes, and deterministic or probabalistic
1360
associations of events.3,4 Aristotelian metaphysics offers a different understanding of causes, and could contribute to the understanding of the causal role of genes. Functional causal relations relate to Aristotle’s system of effective cause.5 This type of cause is regarded as an external cause, as is the so-called final cause, or the intention behind an action. An example for external cause is genitourinary infection with human papillomavirus, which may cause cervical cancer. The virus’ strive for procreation could be interpreted as final cause. Aristotle supplemented these external causes with two internal causes—the formal cause, which is the blueprint that becomes manifest in a given substance, and the material cause. For human papillomavirus infection, therefore, the formal cause is the malignant transformation of cervical tissue, and the material cause is the affected cervical tissue itself. The two external and two internal causes are important for disease; the virus (effective cause), its drive to procreate (final cause), the cervix uteri (material cause), and its potential for malignantcell transformation (formal cause). Current understanding of disease cause is linked closely with external causation, especially the effective cause. The human genome project will shift causal research to internal causes: to the structure of tissue and cells (the material cause) and to the programmes of reaction in response to stimuli (the formal cause of events). Medical interventions derived from gene sequencing might, therefore, be directed towards structural vulnerability (eg, membrane proteins), or messengers of pathophysiological reactions. These powerful options have ethical consequences since the manipulation of internal causes affects personal rights more than an intervention directed against external causes.
Physics and chemistry Nobel Laureates and medicine Sir—Tonse Raju (July 29, p 436)1 completed a superb series recounting and discussing Nobel prize winners in physiology or medicine.1 However, as has been pointed out,2 the award committee seems to have been giving the prize more frequently for basic science discoveries, rather than those with more direct clinical implications, despite many potentially worthy recipients.3 Conversely, the impact of the work of many laureates in physics and chemistry has perhaps been even greater on the diagnosis and treatment of disease than that of some recipients’ work for the prize in physiology or medicine. For example, in physics: W C Roentgen, X-rays, 1901; A H Becquerel, M Cuire, and P Curie, radioactivity, 1903; F Bloch and E Purcell, nuclear magnetic resonance, 1952; W Shockley, J Bardeen, and W Brattain, the transistor effect (computers), 1956; C Townes, N Basov, and A Prokhorov, lasers, 1964; and E Ruska, electron microscopy, 1986. In chemistry: L Pauling, ␣-helix, -sheet, 1954; W Gilbert and F Sanger, DNA sequences, 1980; K Mullis, PCR, 1993. *Eric Lewin Altschuler, Robert Charlton Brain and Perception Laboratory, University of California, San Diego, La Jolla, CA 92093-0109, USA (e-mail: [email protected]) 1
2
3
Raju TNK. The Nobel chronicles— the first century. Lancet 2000; 356: 436. Narrowness of Nobel awards for physiology or medicine. Lancet 1999; 354: 1399. Marshall BJ, Warren JR. Unidentified curved bacilli in the stomach of patients with gastritis and peptic ulceration. Lancet 1984; 1: 1311–15.
Manfred Wildner Bavarian Public Health Research Center and Institute for Medical Informatics, Biometry and Epidemiology, Ludwig-Maximilians-University Munich, Germany (e-mail: [email protected]) 1
2
3
4
5
Lichtenstein P, Holm N, Verkasalo P, et al. Environmental and heritable factors in the causation of cancer. N Engl J Med 2000; 343: 78–85. Diez-Roux A. On genes, individuals, society, and epidemiology. Am J Epidemiol 1998; 148: 1027–32. Wulff H, Pedersen A, Rosenberg R. Philosphy of medicine. Oxford: Blackwell Scientific, 1990. Suppes P. A probalistic theory of causation. Amsterdam: North-Holland Publishing, 1970. Ulfig A. Lexikon der philosophischen Begriffe. Wiesbaden: Fourier, 1999.
DEPARTMENT OF ERROR Control of legionella in drinking water—In this Correspondence letter by John Hayes (Aug 26, p 773), the second sentence of the first paragraph should be, “The ionisation equipment of University Hospital, Bergmannsheil, Bochum, Germany, was correctly installed by our own engineers, and initially set to give the correct concentrations of silver and copper ions within the hot water circulating loops”. In the next sentence the concentrations of silver and copper should be 40 g/L and 400 g/L, respectively. In the last sentence of the letter the concentrations of silver should be between 20 g/L and 40 g/L.
THE LANCET • Vol 356 • October 14, 2000
For personal use only. Not to be reproduced without permission of The Lancet.