- Email: [email protected]
Birthweights, maternal cardiovascular events, and Barker hypothesis
COMMENTARY
COMMENTARY
Birthweights, maternal cardiovascular events, and Barker hypothesis See pages 1997, 2002 It is well established that the intrauterine environment has an important impact on the subsequent health of various animals.1 However, whether intrauterine programming determines certain aspects of adult health in human beings remains controversial.2,3 Several anthropometric variables present at birth or in early infancy have been associated with the development in adulthood of risk factors for cardiovascular disease—ie, hypertension or diabetes—or of coronary heart disease itself.4 The many reports on this topic have polarised clinicians and scientists into camps of believers and non-believers. This segregation has even extended into the publication of the reports in The Lancet or the BMJ.5 The most extensive, consistent, and persuasive data relate to the association between low birthweight and high blood pressure in later life.6 As for any association, this relation may be the result of chance, bias, or confounding factors, or it may be a genuine causative effect. Criticisms of the intrauterine-programming hypothesis have been plentiful and varied.2,3 The favoured criticism is that the association is spuriously created by one or more confounding factors among parental characteristics, such as low socioeconomic status or a specific genotype that causes low birthweight and coronary heart disease in the offspring.7 Two reports in this week’s Lancet have attempted to evaluate the independent effect of intrauterine programming on risk of cardiovascular disease. These studies may be (mis)interpreted as findings against the hypothesis. Gordon Smith and colleagues report that mothers whose first child was in the lowest birthweight quintile or born prematurely, or who had pre-eclampsia during that delivery, have an increased risk of death from any cause, death due to ischaemic heart disease, and admission or death due to ischaemic heart disease. These results are compatible with those of earlier studies.8 The researchers therefore conclude that the relation between low birthweight and ischaemic heart disease later in life is in part, or wholly, a consequence of genotypes that predispose to both endpoints. Critics of this study might reasonably ask why only data on firstborns were included, how birthweight was measured, and how maternal riskfactor status after delivery of the firstborn might have affected cardiovascular events in the mother. Perhaps most important is the absence of any data on the confounding impact of smoking. Certainly the causes of excess death observed among the mothers are all associated with smoking or alcohol excess. The importance attached to such criticisms tends to depend 1990
on whether the results confirm or refute the reader’s prejudices about the intrauterine-programming hypothesis. Confounding by parental characteristics, genetic or environmental, will only be eliminated in studies of monozygotic twins that investigate associations between anthropometric features within the twin-pairs and the development of a cardiovascular risk factor (ie, hypertension or diabetes) or a cardiovascular event in later life. This approach was attempted in the second study reported today by A Hübinette and colleagues. The investigators looked at differences in anthropometric measurements within newborn twin-pairs who in later life were discordant for acute myocardial infarction. An additional control group of unrelated sets of twins unaffected by acute myocardial infarction was also used for comparison. A significant association between low birthweight and increased risk of acute myocardial infarction was seen when unrelated twin controls were used, but not when co-twins were the comparator group. The researchers conclude that the reported association between low birthweight and increased risk of coronary heart disease is a function either of confounding by genetics or by unmeasured maternal factors that operate independently of birthweight. Discordance between twins meant that the control twin had not had an acute myocardial infarction at the time the case was diagnosed. However, true concordance as defined in this study would require simultaneous acute myocardial infarctions. This study exemplifies the difficulty of classifying such endpoints at a single timepoint as “disease” vs “no disease”. A secondary analysis that excluded pairs of twins among whom the co-twin control later developed an acute myocardial infarction also suggested a lack of association between birth characteristics and acute myocardial infarction. However, risk factors found in adulthood, such as smoking, alcohol intake, and exercise level were not considered in the analyses, and the impact of adjustment for these factors would clearly be of interest. Of the 132 twin-pairs discordant for acute myocardial infarction, only 40 were monozygotic (assessed by questionnaire). Since several hundred twin-pairs are needed to show a significant correlation of adult bloodpressure values with birthweight in the within-twin model,9 it is perhaps not surprising that in a study of 132 twin-pairs, mean birthweights of cases and their co-twins did not differ. More valid evidence would arise from an analysis of rates of genuine discordance for coronary heart disease (perhaps evaluated by a good indirect measure of coronary atherosclerosis) across strata of birthweight
THE LANCET • Vol 357 • June 23, 2001
For personal use. Only reproduce with permission from The Lancet Publishing Group.
COMMENTARY
differences in a much larger sample of DNA-confirmed monozygotic twins. Although a significant association between birthweight and acute myocardial infarction among external twin but not co-twin controls is compatible with confounding by some parental characteristic—genetic or environmental, in view of the size of the study alternative explanations should be considered. For those who do not believe in the hypothesis, larger carefully controlled studies within monozygotic twin pairs are required. For those who do believe the hypothesis, it has long been time to move on to identifying relevant mechanisms. Neil R Poulter Cardiovascular Studies Unit, Imperial College School of Medicine, St Mary’s Campus, London W2 1PG, UK (e-mail: [email protected]) 1
2 3 4 5 6
7
8
9
Langley SC, Jackson AA. Increased systolic blood pressure in adult rats induced by fetal exposure to maternal low protein diets. Clin Sci (Colch) 1994; 86: 217–22. Paneth N, Susser M. Early origin of coronary heart disease (the “Barker hypothesis”). BMJ 1995; 310: 411–12. Susser M, Levin B. Ordeals for the fetal programming hypothesis. BMJ 1999; 318: 885–86. Barker DJP. Mothers, babies and disease in later life. London: BMJ Publishing, 1994. Y Ben-Schlomo, Davey Smith G. “Place of publication” bias? BMJ 1994; 309: 274. Huxley RR, Shiell AW, Law CM. The role of size at birth and postnatal catch-up growth in determining systolic blood pressure: a systematic review of the literature. J Hypertens 2000; 18: 815–31. Hattersley AT, Tooke JE. The fetal insulin hypothesis: an alternative explanation of the association of low birthweight with diabetes and vascular disease. Lancet 1999; 353: 1789–92. Smith GD, Whitley E, Gissler M, Hemminki E. Birth dimensions of offspring, premature birth, and the mortality of mothers. Lancet 2000; 356: 2066–67. Poulter NR, Chang CL, MacGregor AJ, Snieder H, Spector TD. Association between birthweight and adult blood pressures of twins: a historical cohort study. BMJ 1999; 319: 1330–33.
Leukotriene antagonists and symptom control in chronic persistent asthma See page 2007 Agents that interfere with the synthesis or action of cysteinyl leukotrienes belong to the first new class of antiasthma drugs to appear in the past 3 decades. Montelukast and zafirlukast are antagonists to the cysteinyl leukotriene receptor and both drugs inhibit exercise-induced bronchoconstriction1,2 and early and late responses to inhaled allergen.3,4 In patients with mild to moderate asthma taking inhaled 2 agonists as required, these drugs have improved lung function,5,6 reduced the need for rescue bronchodilators,5,7 and decreased the sputum eosinophil count.8 However, most clinical interest is in the use of leukotriene antagonists in patients who have symptoms despite treatment with inhaled corticosteroids. There is a reasonable expectation that agents from this class might have additive effects with inhaled steroids for this group since the overproduction of cysteinyl leukotrienes in airways is not inhibited by glucocorticoids.9 Clinical trials have indeed shown evidence of efficacy of leukotriene antagonists among patients taking inhaled steroids,10 and one study has shown that the introduction of montelukast allows a reduction in the dose of inhaled corticosteroid without loss of asthma control.11 In today’s issue of The Lancet, Douglas Robinson and colleagues report results from a double-blind placebocontrolled crossover trial of 2 weeks of treatment with
THE LANCET • Vol 357 • June 23, 2001
montelukast in patients with continued symptoms of asthma. This cohort, in a tertiary referral unit, differs from those in other studies in that almost all patients were already receiving high-dose inhaled corticosteroids, many in combination with long-acting 2 agonists, theophylline, and in some cases, oral prednisolone. Montelukast had no effect on peak expiratory flow, symptom scores, or spirometric measurements, and in the placebo and montelukast groups similar proportions of patients were classified as responders. Airway hyperresponsiveness and airway inflammation were not measured, so it is possible that montelukast had an effect on these important surrogate markers of asthma control. Also, a longer and larger study may have identified a clinically relevant reduction in exacerbation frequency. The researchers’ explanation for the negative findings is that montelukast adds little to the more well-established agents that the participants were already taking. One concern with the study is the selection of patients. The response to drugs that interfere with the leukotriene pathway is influenced by several genetic polymorphisms: patients with the promoter sequence of the 5-lipoxygenase gene seem to be resistant to treatment,12 whereas those with C/C and C/A genotypes of LTC4 synthase respond particularly well.13 These latter patients may be under-represented in Robinson and co-workers’ study because they have already been treated successfully with leukotriene antagonists in primary or secondary care. Patients referred to hospital with asthma are a highly atypical group who have not responded to various treatments started in primary care. In a recent community-based study of 983 patients, only 13 were seen in hospital because of poorly controlled symptoms during a 12-month period.14 Some of these patients may have difficulties with compliance (of note, 28% of Robinson’s patients did not complete diary cards properly), others have genuinely severe asthma that is refractory to treatment, and in some patients the symptoms are due to another disorder.15 It may be that any therapy would have a modest effect in this group and that Robinson and colleagues’ study says as much about the difficulty in treating this population of patients as about the efficacy of montelukast. The study does, however, raise an important question about the ability to generalise the results of clinical trials in asthma to everyday practice. Almost all of the published clinical studies of leukotriene receptor antagonists involved patients who demonstrated acute improvements in forced expiratory volume in 1 s (FEV1) of 15% or more after challenge with inhaled bronchodilators,5,6,8,10 and in one study the average bronchodilator response was 36%.6 Such values of reversibility are distinctly unusual in clinical practice. For example, in one study only 10% of the patients referred to a hospital asthma clinic had a 15% or greater improvement in FEV1 after 200 g inhaled salbutamol16 and only 28% of patients in general practice had this response after 2·5 mg nebulised salbutamol.17 Generalisability of trial data is a particular issue with recent highly influential trials of long-acting 2 agonists in which the entry criteria of a marked acute bronchodilator response is likely to have biased results in favour of the study drug.18,19 Patients included in many clinical trials may therefore be no more representative of the general asthma population than those included by Robinson and colleagues. The investigators have suggested that clinical trials include a more diverse range of patients. Another necessary step is to carry out extensive characterisation of 1991
For personal use. Only reproduce with permission from The Lancet Publishing Group.
COMMENTARY
Birthweights, maternal cardiovascular events, and Barker hypothesis See pages 1997, 2002 It is well established that the intrauterine environment has an important impact on the subsequent health of various animals.1 However, whether intrauterine programming determines certain aspects of adult health in human beings remains controversial.2,3 Several anthropometric variables present at birth or in early infancy have been associated with the development in adulthood of risk factors for cardiovascular disease—ie, hypertension or diabetes—or of coronary heart disease itself.4 The many reports on this topic have polarised clinicians and scientists into camps of believers and non-believers. This segregation has even extended into the publication of the reports in The Lancet or the BMJ.5 The most extensive, consistent, and persuasive data relate to the association between low birthweight and high blood pressure in later life.6 As for any association, this relation may be the result of chance, bias, or confounding factors, or it may be a genuine causative effect. Criticisms of the intrauterine-programming hypothesis have been plentiful and varied.2,3 The favoured criticism is that the association is spuriously created by one or more confounding factors among parental characteristics, such as low socioeconomic status or a specific genotype that causes low birthweight and coronary heart disease in the offspring.7 Two reports in this week’s Lancet have attempted to evaluate the independent effect of intrauterine programming on risk of cardiovascular disease. These studies may be (mis)interpreted as findings against the hypothesis. Gordon Smith and colleagues report that mothers whose first child was in the lowest birthweight quintile or born prematurely, or who had pre-eclampsia during that delivery, have an increased risk of death from any cause, death due to ischaemic heart disease, and admission or death due to ischaemic heart disease. These results are compatible with those of earlier studies.8 The researchers therefore conclude that the relation between low birthweight and ischaemic heart disease later in life is in part, or wholly, a consequence of genotypes that predispose to both endpoints. Critics of this study might reasonably ask why only data on firstborns were included, how birthweight was measured, and how maternal riskfactor status after delivery of the firstborn might have affected cardiovascular events in the mother. Perhaps most important is the absence of any data on the confounding impact of smoking. Certainly the causes of excess death observed among the mothers are all associated with smoking or alcohol excess. The importance attached to such criticisms tends to depend 1990
on whether the results confirm or refute the reader’s prejudices about the intrauterine-programming hypothesis. Confounding by parental characteristics, genetic or environmental, will only be eliminated in studies of monozygotic twins that investigate associations between anthropometric features within the twin-pairs and the development of a cardiovascular risk factor (ie, hypertension or diabetes) or a cardiovascular event in later life. This approach was attempted in the second study reported today by A Hübinette and colleagues. The investigators looked at differences in anthropometric measurements within newborn twin-pairs who in later life were discordant for acute myocardial infarction. An additional control group of unrelated sets of twins unaffected by acute myocardial infarction was also used for comparison. A significant association between low birthweight and increased risk of acute myocardial infarction was seen when unrelated twin controls were used, but not when co-twins were the comparator group. The researchers conclude that the reported association between low birthweight and increased risk of coronary heart disease is a function either of confounding by genetics or by unmeasured maternal factors that operate independently of birthweight. Discordance between twins meant that the control twin had not had an acute myocardial infarction at the time the case was diagnosed. However, true concordance as defined in this study would require simultaneous acute myocardial infarctions. This study exemplifies the difficulty of classifying such endpoints at a single timepoint as “disease” vs “no disease”. A secondary analysis that excluded pairs of twins among whom the co-twin control later developed an acute myocardial infarction also suggested a lack of association between birth characteristics and acute myocardial infarction. However, risk factors found in adulthood, such as smoking, alcohol intake, and exercise level were not considered in the analyses, and the impact of adjustment for these factors would clearly be of interest. Of the 132 twin-pairs discordant for acute myocardial infarction, only 40 were monozygotic (assessed by questionnaire). Since several hundred twin-pairs are needed to show a significant correlation of adult bloodpressure values with birthweight in the within-twin model,9 it is perhaps not surprising that in a study of 132 twin-pairs, mean birthweights of cases and their co-twins did not differ. More valid evidence would arise from an analysis of rates of genuine discordance for coronary heart disease (perhaps evaluated by a good indirect measure of coronary atherosclerosis) across strata of birthweight
THE LANCET • Vol 357 • June 23, 2001
For personal use. Only reproduce with permission from The Lancet Publishing Group.
COMMENTARY
differences in a much larger sample of DNA-confirmed monozygotic twins. Although a significant association between birthweight and acute myocardial infarction among external twin but not co-twin controls is compatible with confounding by some parental characteristic—genetic or environmental, in view of the size of the study alternative explanations should be considered. For those who do not believe in the hypothesis, larger carefully controlled studies within monozygotic twin pairs are required. For those who do believe the hypothesis, it has long been time to move on to identifying relevant mechanisms. Neil R Poulter Cardiovascular Studies Unit, Imperial College School of Medicine, St Mary’s Campus, London W2 1PG, UK (e-mail: [email protected]) 1
2 3 4 5 6
7
8
9
Langley SC, Jackson AA. Increased systolic blood pressure in adult rats induced by fetal exposure to maternal low protein diets. Clin Sci (Colch) 1994; 86: 217–22. Paneth N, Susser M. Early origin of coronary heart disease (the “Barker hypothesis”). BMJ 1995; 310: 411–12. Susser M, Levin B. Ordeals for the fetal programming hypothesis. BMJ 1999; 318: 885–86. Barker DJP. Mothers, babies and disease in later life. London: BMJ Publishing, 1994. Y Ben-Schlomo, Davey Smith G. “Place of publication” bias? BMJ 1994; 309: 274. Huxley RR, Shiell AW, Law CM. The role of size at birth and postnatal catch-up growth in determining systolic blood pressure: a systematic review of the literature. J Hypertens 2000; 18: 815–31. Hattersley AT, Tooke JE. The fetal insulin hypothesis: an alternative explanation of the association of low birthweight with diabetes and vascular disease. Lancet 1999; 353: 1789–92. Smith GD, Whitley E, Gissler M, Hemminki E. Birth dimensions of offspring, premature birth, and the mortality of mothers. Lancet 2000; 356: 2066–67. Poulter NR, Chang CL, MacGregor AJ, Snieder H, Spector TD. Association between birthweight and adult blood pressures of twins: a historical cohort study. BMJ 1999; 319: 1330–33.
Leukotriene antagonists and symptom control in chronic persistent asthma See page 2007 Agents that interfere with the synthesis or action of cysteinyl leukotrienes belong to the first new class of antiasthma drugs to appear in the past 3 decades. Montelukast and zafirlukast are antagonists to the cysteinyl leukotriene receptor and both drugs inhibit exercise-induced bronchoconstriction1,2 and early and late responses to inhaled allergen.3,4 In patients with mild to moderate asthma taking inhaled 2 agonists as required, these drugs have improved lung function,5,6 reduced the need for rescue bronchodilators,5,7 and decreased the sputum eosinophil count.8 However, most clinical interest is in the use of leukotriene antagonists in patients who have symptoms despite treatment with inhaled corticosteroids. There is a reasonable expectation that agents from this class might have additive effects with inhaled steroids for this group since the overproduction of cysteinyl leukotrienes in airways is not inhibited by glucocorticoids.9 Clinical trials have indeed shown evidence of efficacy of leukotriene antagonists among patients taking inhaled steroids,10 and one study has shown that the introduction of montelukast allows a reduction in the dose of inhaled corticosteroid without loss of asthma control.11 In today’s issue of The Lancet, Douglas Robinson and colleagues report results from a double-blind placebocontrolled crossover trial of 2 weeks of treatment with
THE LANCET • Vol 357 • June 23, 2001
montelukast in patients with continued symptoms of asthma. This cohort, in a tertiary referral unit, differs from those in other studies in that almost all patients were already receiving high-dose inhaled corticosteroids, many in combination with long-acting 2 agonists, theophylline, and in some cases, oral prednisolone. Montelukast had no effect on peak expiratory flow, symptom scores, or spirometric measurements, and in the placebo and montelukast groups similar proportions of patients were classified as responders. Airway hyperresponsiveness and airway inflammation were not measured, so it is possible that montelukast had an effect on these important surrogate markers of asthma control. Also, a longer and larger study may have identified a clinically relevant reduction in exacerbation frequency. The researchers’ explanation for the negative findings is that montelukast adds little to the more well-established agents that the participants were already taking. One concern with the study is the selection of patients. The response to drugs that interfere with the leukotriene pathway is influenced by several genetic polymorphisms: patients with the promoter sequence of the 5-lipoxygenase gene seem to be resistant to treatment,12 whereas those with C/C and C/A genotypes of LTC4 synthase respond particularly well.13 These latter patients may be under-represented in Robinson and co-workers’ study because they have already been treated successfully with leukotriene antagonists in primary or secondary care. Patients referred to hospital with asthma are a highly atypical group who have not responded to various treatments started in primary care. In a recent community-based study of 983 patients, only 13 were seen in hospital because of poorly controlled symptoms during a 12-month period.14 Some of these patients may have difficulties with compliance (of note, 28% of Robinson’s patients did not complete diary cards properly), others have genuinely severe asthma that is refractory to treatment, and in some patients the symptoms are due to another disorder.15 It may be that any therapy would have a modest effect in this group and that Robinson and colleagues’ study says as much about the difficulty in treating this population of patients as about the efficacy of montelukast. The study does, however, raise an important question about the ability to generalise the results of clinical trials in asthma to everyday practice. Almost all of the published clinical studies of leukotriene receptor antagonists involved patients who demonstrated acute improvements in forced expiratory volume in 1 s (FEV1) of 15% or more after challenge with inhaled bronchodilators,5,6,8,10 and in one study the average bronchodilator response was 36%.6 Such values of reversibility are distinctly unusual in clinical practice. For example, in one study only 10% of the patients referred to a hospital asthma clinic had a 15% or greater improvement in FEV1 after 200 g inhaled salbutamol16 and only 28% of patients in general practice had this response after 2·5 mg nebulised salbutamol.17 Generalisability of trial data is a particular issue with recent highly influential trials of long-acting 2 agonists in which the entry criteria of a marked acute bronchodilator response is likely to have biased results in favour of the study drug.18,19 Patients included in many clinical trials may therefore be no more representative of the general asthma population than those included by Robinson and colleagues. The investigators have suggested that clinical trials include a more diverse range of patients. Another necessary step is to carry out extensive characterisation of 1991
For personal use. Only reproduce with permission from The Lancet Publishing Group.