Developmental Origins of Hypertension: Biology Meets Statistics

Editorial

Susan P. Bagby Professor of Medicine and Physiology/Pharmacology, Oregon Health & Science University and Portland VA Medical Center, Portland, Oregon J Am Soc Nephrol 17: 2356–2358, 2006. doi: 10.1681/ASN.2006070744

he concept that environmental factors during develop- adults (15,22), the variability in children may indicate that ment—both in utero and in early childhood—can per- developmentally programmed hypertension requires cofactors T manently alter the adult phenotype derives from the that emerge only later in life. In the existing literature, there seminal epidemiologic observations of Professor David Barker have been no large, prospective, cohort studies that include (1). He and his colleagues showed in the early 1990s that low standardized BP measurements in children. birthweight predicted increased risk of cardiovascular diseases, In this issue of JASN, Hemachandra et al. report on the including insulin resistance/diabetes (2–4), obesity (5), coro- relationship between BirthWt and SysBP at age 7 yr in a large, nary disease (6), and, importantly, systolic hypertension (3,7,8). multicenter, US cohort of children born full-term between 1959 Birthweight was used as a readily available, albeit crude, sur- and 1965 (Collaborative Perinatal Project) (23). This database rogate for adequacy of fetal growth. While these early obser- (now in the public domain) is exceptionally large (n ϭ 29,710) vations have now been confirmed in studies around the globe and racially balanced (approximately 50% black). The major (9,10), the link between birthweight (BirthWt) and systolic BP finding reported is that black children exhibited an unexpected (SysBP)—both its presence and its significance—has continued positive BirthWt:SysBP relationship, whereas white children to generate controversy (11,12). The typical finding in adults showed no significant relationship. The racial difference is ro- has been an inverse relationship between BirthWt and SysBP bust and is not diminished by adjustment for current weight (9) that is independent of—but enhanced by—accelerated (CurrentWt) or by adjustment for childhood weight gain over childhood growth (7). The existence of this relationship in age 4 to 7 yr (see adjustment issues below). The pattern ob- white adults, not always apparent in studies with limited ma- served in blacks mirrors findings of Rostand et al. (24) in a small ternal information (12), was recently confirmed in 31-yr-old and selected group of 5-yr-old subjects in Alabama, in which subjects from a large prospective and broadly annotated North black children also exhibited a positive relationship between Finland birth cohort (13). While the differences in average BirthWt and SysBP. In contrast, the white children in the latter SysBP across the birthweight range are typically small, their study showed an inverse correlation. cardiovascular impact is predicted to be large based on modern Despite the cohort size and convincing racial disparities, the hypertension treatment trials (e.g., [14]) and based on a similar report by Hemachandra et al. may not fully resolve for white inverse relationship of BirthWt and prevalence of clinical hy- children the question of whether a true inverse relationship is pertension (15). The mechanisms linking BirthWt and adult present at age 7 yr. This relates to perhaps the stickiest contro- SysBP are not known, but reduced nephron number (16), fetally versy surrounding the BirthWt:SysBP relationship, a statistical programmed activation of the postnatal tissue renin/angioten- phenomenon known as the “reversal paradox” or “statistical sin II (AngII) system (17), and enhanced vascular reactivity to suppression” (25,26). This phenomenon argues that statistical pressors (18) and stress (19) have each been implicated. adjustment in multivariate regression (e.g., the BirthWt:SysBP In experimental models, it is abundantly clear that impaired relationship) for a third factor that lies in the etiologic pathway fetal growth can lead to altered BP regulation, often in prepu- (e.g., current body mass index [BMI] or weight) has the poten- bertal offspring (20). However, in humans, it is not yet clear at tial to create a spurious inverse relationship that has no biologic what age the SysBP effects of fetal programming become man- relevance. Importantly, this statistical adjustment for current ifest, especially since the inverse BirthWt:SysBP relationship body weight in multivariate regression emphasizes the effect of may be weak or absent in children, especially adolescents (21). change in body weight (BirthWt versus CurrentWt), and does Because the inverse relationship is strengthened with aging in not precisely assess separate effects of the two parameters (26). Whether this adjustment is appropriate or confounding may well depend on the specifics of the cohort (12,25). A consider- Published online ahead of print. Publication date available at www.jasn.org. ation of the biologic complexities now known to underlie fetal Address correspondence to: Dr. Susan P. Bagby, Professor of Medicine and growth restriction is instructive in pondering that question as Physiology/Pharmacology, Division of Nephrology & Hypertension, 3314 SW US Veterans Hospital Road, Suite PP 262, Portland, OR 97239-2940. Phone: 503-494- well as the decision of Hemachandra et al. to avoid adjustment 8490; Fax: 503-494-5330; E-mail: [email protected] for current weight.

In a unique longitudinal Helsinki cohort providing annual considerably fewer calories than would be the case today in a growth data for all children in the cohort, both a lower BirthWt financially comparable US population. If limited prenatal nutri- and an accelerated increase in BMI in childhood independently tion induced programming, and continued postnatal calorie def- increased the risk of adult hypertension (7). While this database icit with persisting poverty prevented programmed acceleration could not address whether impaired fetal growth was causal in of childhood growth, then the cohort overall might have a con- the later growth acceleration, both human and animal studies siderably weakened inverse BirthWt:SysBP relationship as com- strongly support an etiologic link (27). That is to say, global caloric pared with 7 yr olds in a nutrient-rich postnatal setting. As a restriction in rat dams leads to hyperphagia and excess food result, the findings in this low-income population, while represen- intake (“thrifty phenotype”) in offspring, together with acceler- tative of an unfortunately growing segment of the US population, ated weight gain and excess fat mass (28) under ad libitum condi- might thus not be generally applicable to more affluent US chil- tions. In microswine offspring after maternal protein restriction, dren. Moreover, with respect to racial differences, black mothers similar hyperphagia and accelerated growth on ad libitum intake exhibited twice the poverty rate of whites and, perhaps not sur- precede sustained hypertension (Du Priest et al., unpublished prisingly, were more anemic. Accordingly, more severe calorie observations, 2006). In South African children who were born deficits pre- and/or postnatally, or a higher prevalence of specific small and subsequently exhibited rapid growth, weight gain was nutrient deficits known to exert programming effects indepen- preferentially deposited as central fat rather than lean tissue (29). dently of birthweight (e.g. iron [30]), may have played a role in Thus, altered fetal nutrition/growth impacts not only fetal but black offspring independent of genetic factors. also postnatal growth trajectory when postnatal food is available. From a broader perspective, these issues highlight the in- There is, then, after intrauterine growth restriction, a fetally pro- creasingly recognized insensitivity of BirthWt as a surrogate for grammed distortion of the normal BirthWt:CurrentWt relation- fetal growth/nutrient adequacy (12). Fetal growth restriction, ship, and this varies depending on the nutrient availability in as manifested by the thin infant with its asymmetric growth postnatal nutritional environment. The argument to adjust or not phenotype, occurs within the normal BirthWt range. Moreover, to adjust for CurrentWt overlooks this complexity, since neither as suggested above, specific nutrient deficits can program per- option statistically accounts for these biologic interactions. Keizjer- manent effects (e.g., nephron number via prenatal glucocorti- Veen et al. (26) have suggested an alternative statistical approach coid [31] or BP via maternal iron [30] or calcium [32]) without that accounts independently for the BirthWt effect and for the modifying overall body/organ growth patterns. As Gillman effect of postnatal growth exceeding that expected based on has emphasized, understanding and detecting the specific un- BirthWt alone (as BirthWt and CurrentWt normally have a posi- derlying mechanisms—independent of BirthWt—are crucial to tive relationship). Thus, “current weight residuals,” the difference our future ability to identify mothers and infants at risk for between BirthWt and predicted versus actual CurrentWt, are used fetally programmed hypertension (12). to factor for CurrentWt and are independent of BirthWt. This In summary, Hemachandra et al. have provided an important approach specifically accounts for the presence (or absence) of the contribution by defining, in a large prospective US cohort, racial programmed distortion of the BirthWt:CurrentWt relationship discrepancies in the BirthWt:SysBP relationship that are already without the potential confounding inherent in simply adjusting apparent at age 7 yr. Whether these robust differences are genetic, for current body weight. While beyond the scope of this commen- socioeconomic/environmental, or an interaction between the two tary, the approach (26) provides a compelling argument that our is not yet clear, since other differences between the groups could statistical methods must evolve along with our biologic knowl- potentially explain the findings. Taking heed of these results, edge. future studies should be designed to differentiate environmental Hemachandra et al. include, for purposes of comparison, a from racial/genetic factors given their disparate implications for CurrentWt-adjusted analysis and find that white children in therapeutic intervention. As future reports from the Perinatal Col- fact convert from a neutral to a significantly inverse BirthWt: laborative Project are forthcoming, it is hoped that new statistical SysBP relationship, while black children convert from positive approaches that account for the biologic complexity of program- to neutral (23). While choosing to accept the BirthWt:SysBP ming effects will add to our understanding of growth patterns and relationship without CurrentWt adjustment, thereby avoiding BP profiles in children destined to be come hypertensive. the “reversal paradox,” Hemachandra et al. do not employ alternative approaches that would more appropriately factor for CurrentWt effects on SysBP. So, while it is possible that References white 7-yr-olds are in fact simply too young to exhibit an 1. Barker DJ: Introduction. In: Fetal Origins of Cardiovascular inverse BirthWt:SysBP relationship, it is also not possible to and Lung Disease, edited by Barker DJ, Lenfant C, New definitely exclude this idea as currently analyzed. York, Marcel Dekker, Inc., 2001, pp 1–22 2. Eriksson JG, Forsen T, Tuomilehto J, Jaddoe VW, Osmond This raises a final feature, unique to the Collaborative Perinatal C, Barker DJ: Effects of size at birth and childhood growth Project, which should inform interpretation of BirthWt:SysBP pat- on the insulin resistance syndrome in elderly individuals. terns and racial differences (23). The most striking feature of this Diabetologia 45: 342–348, 2002 cohort of pregnant mothers was the prevalence of poverty as 3. Barker DJ, Hales CN, Fall CH, Osmond C, Phipps K, Clark determined using 1960 to 1970 criteria: 39% in whites, 72% in PM: Type 2 (non-insulin-dependent) diabetes mellitus, hy- blacks, 54% overall. These mothers were remarkably thin, and, in pertension and hyperlipidaemia (syndrome X): Relation to a culture that predated cheap fast food, were likely consuming reduced fetal growth. Diabetologia 36: 62–67, 1993 2358 Journal of the American Society of Nephrology J Am Soc Nephrol 17: 2356–2358, 2006

4. Forsen T, Eriksson J, Tuomilehto J, Reunanen A, Osmond C, 19. Nichols SD, Boyne MS, Thame M, Osmond C, Wilks RJ, Barker D: The fetal and childhood growth of persons who Bennett FI, Farlane-Anderson N, Young RE, Forrester TE: develop type 2 diabetes. Ann Intern Med 133: 176–182, 2000 Cold-induced elevation of forearm vascular resistance is 5. Eriksson J, Forsen T, Tuomilehto J, Osmond C, Barker D: inversely related to birth weight. J Hum Hypertens 19: 309– Size at birth, childhood growth and obesity in adult life. Int 314, 2005 J Obes Relat Metab Disord 25: 735–740, 2001 20. Armitage JA, Khan IY, Taylor PD, Nathanielsz PW, Poston 6. Barker DJ, Winter PD, Osmond C, Margetts B, Simmonds L: Developmental programming of the metabolic syn- SJ: Weight in infancy and death from ischaemic heart dis- drome by maternal nutritional imbalance: How strong is ease. The Lancet 2: 577–580, 1989 the evidence from experimental models in mammals? 7. Barker DJ, Forsen T, Eriksson JG, Osmond C: Growth and J Physiol 561: 355–377, 2004 living conditions in childhood and hypertension in adult 21. Falkner B, Hulman S, Kushner H: Effect of birth weight on life: A longitudinal study. J Hypertens 20: 1951–1956, 2002 blood pressure and body size in early adolescence. Hyper- 8. Eriksson J, Forsen T, Tuomilehto J, Osmond C, Barker D: tension 43: 203–207, 2004 Fetal and childhood growth and hypertension in adult life. 22. Koupilova I, Leon DA, Lithell HO, Berglund L: Size at birth Hypertension 36: 790–794, 2000 and hypertension in longitudinally followed 50–70-year- 9. Huxley RR, Shiell AW, Law CM: The role of size at birth old men. Blood Press 6: 223–228, 1997 and postnatal catch-up growth in determining systolic 23. Hemachandra AH, Klebanoff MA, Furth SL: Racial dispar- blood pressure: A systematic review of the literature. J Hy- ities in the association between birth weight in the term pertens 18: 815–831, 2000 infant and blood pressure at age 7 years: Results from the 10. Barker DJ, Bagby SP: Developmental antecedents of car- Collaborative Perinatal Project. J Am Soc Nephrol 17: 2576– diovascular disease: A historical perspective. JAmSoc 2581, 2006 Nephrol 16: 2537–2544, 2005 24. Rostand SG, Cliver SP, Goldenberg RL: Racial disparities 11. Huxley R, Neil A, Collins R: Unravelling the fetal origins in the association of foetal growth retardation to childhood hypothesis: Is there really an inverse association between blood pressure. Nephrol Dial Transplant 20: 1592–1597, 2005 birthweight and subsequent blood pressure? Lancet 360: 25. Tu YK, West R, Ellison G, Gilthorpe MS: Why evidence for 659–665, 2002 the fetal origins of adult disease might be a statistical 12. Gillman MW: Epidemiological challenges in studying the artifact: The “reversal paradox” for the relation between fetal origins of adult disease. Paediatr Perinat Epidemiol birth weight and blood pressure in later life. Am J Epidemiol 19[Suppl 1]: 1, 2005 161, 27–32, 2005 13. Jarvelin MR, Sovio U, King V, Lauren L, Xu B, McCarthy 26. Keijzer-Veen MG, Euser AM, van Montfoort N, Dekker MI, Hartikainen AL, Laitinen J, Zitting P, Rantakallio P, FW, Vandenbroucke JP, Van Houwelingen HC: A regres- Elliott P: Early life factors and blood pressure at age 31 sion model with unexplained residuals was preferred in years in the 1966 northern Finland birth cohort. Hyperten- the analysis of the fetal origins of adult diseases hypothe- sion 44: 838–846, 2004 14. Hansson L, Zanchetti A, Carruthers SG, Dahlof B, Elmfeldt sis. J Clin Epidemiol 58: 1320–1324, 2005 D, Julius S, Menard J, Rahn KH, Wedel H, Westerling S: 27. Breier BH, Vickers MH, Ikenasio BA, Chan KY, Wong WP: Effects of intensive blood-pressure lowering and low-dose Fetal programming of appetite and obesity. Mol Cell Endo- aspirin in patients with hypertension: Principal results of crinol 185: 73–79, 2001 the Hypertension Optimal Treatment (HOT) randomised 28. Vickers MH, Breier BH, Cutfield WS, Hofman PL, Gluck- trial. HOT Study Group [see comments]. Lancet 351: 1755– man PD: Fetal origins of hyperphagia, obesity, and hyper- 1762, 1998 tension and postnatal amplification by hypercaloric nutri- 15. Curhan GC, Chertow GM, Willett WC, Spiegelman D, tion. Am J Physiol Endocrinol Metab 279: E83–E87, 2000 Colditz GA, Manson JE, Speizer FE, Stampfer MJ: Birth 29. Levitt NS, Lambert EV, Woods D, Seckl JR, Hales CN: weight and adult hypertension and obesity in women. Circ Adult BMI and fat distribution but not height amplify the 94: 1310–1315, 1996 effect of low birthweight on insulin resistance and in- 16. Bagby SP: Maternal nutrition, reduced nephron number, creased blood pressure in 20-year-old South Africans. Dia- and hypertension in later life. Pathways of nutritional pro- betologia 48: 1118–1125, 2005 gramming. J Nutr 2006; in press 30. Lisle SJ, Lewis RM, Petry CJ, Ozanne SE, Hales CN, For- 17. Vehaskari VM, Stewart T, Lafont D, Soyez C, Seth D, head AJ: Effect of maternal iron restriction during preg- Manning J: Kidney angiotensin and angiotensin receptor nancy on renal morphology in the adult rat offspring. Br J expression in prenatally programmed hypertension. Am J Nutr 90: 33–39, 2003 Physiol Renal Physiol 287: F262–F267, 2004 31. Ortiz LA, Quan A, Weinberg A, Baum M: Effect of prenatal 18. Franco MC, Akamine EH, Di Marco GS, Casarini DE, dexamethasone on rat renal development. Kidney Int 59: Fortes ZB, Tostes RC, Carvalho MH, Nigro D: NADPH 1663–1669, 2001 oxidase and enhanced superoxide generation in intrauter- 32. Gillman MW, Rifas-Shiman SL, Kleinman KP, Rich- ine undernourished rats: Involvement of the renin-angio- Edwards JW, Lipshultz SE: Maternal calcium intake and tensin system. Cardiovasc Res 59: 767–775, 2003 offspring blood pressure. Circulation 110: 1990–1995, 2004

See the related article, “Racial Disparities in the Association between Birth Weight in the Term Infant and Blood Pressure at Age 7 Years: Results from the Collaborative Perinatal Project,” pages 2576–2581.