Birth weight of twins: 2. Fetal genetic effect on birth weight

Birth weight of twins:

2. Fetal genetic effect on birth weight

Bernardo Beiguelman^1,2, Glória M.D.D. Colletto ², Carla Franchi-Pinto ³ and Henrique Krieger ²

¹ Curso de Pós-Graduação em Genética, UNICAMP, Campinas, SP, Brasil.

² Laboratório de Epidemiologia Genética, Departamento de Parasitologia, Instituto de Ciências Biomédicas, USP. Send correspondence to B.B. Rua Angelina Maffei Vita, 408, Apto. 41, 01455-070 São Paulo, SP, Brasil.

³ Serviço de Genética Médica da Santa Casa de Misericórdia de São Paulo, SP, Brasil.

ABSTRACT

Intraclass correlation coefficients were calculated for the birth weights of twins born at three southeastern Brazilian hospitals, after adjustment of the natural logarithms of these weights for gestational age, its quadratic and cubic terms, sex, and their interactions. The data indicate that fetal genetic effect on birth weight might have the opportunity to be demonstrated by children born to undernourished women. Undernourishment, acting as a selective force, might enhance the existence of genotypes that determine less need of food for normal development.

INTRODUCTION

For a long time, all authors have agreed, except Magnus (1984), that the fetus genotype is much less important than the maternal constitution or maternal environment in determining human birth weight (Karn et al., 1951; Karn, 1952; Penrose, 1952; Robson, 1955; Morton, 1955; Fraccaro, 1957; Nance et al., 1983; Carr-Hill et al., 1987). In spite of such general agreement, we felt that fetal genetic effect on birth weight might need more insight, since an indication exists that fetal growth of twins is delayed earlier than usual when they are generated by undernourished pregnant women (Beiguelman et al., 1998). Therefore, we decided to compare the birth weight adjusted for sex and gestational age of like-sex and unlike-sex twins born in three Brazilian hospitals, which are strikingly different with respect to the socioeconomic background of the patients. Obviously, the impossibility to separate twins according to their zygosity introduces a limitation to this study. However, we consider that, as a first approach, this investigation could give some indication of the fetal genetic effect on birth weight, since like-sex twins include all monozygotic pairs while unlike-sex twins are surely dizygotic, except the very rare heterokaryotic monozygotic pairs (Turpin et al., 1961).

SUBJECTS AND METHODS

The birth weights of 1,158 twin pairs, whose registered records did not fail to indicate gestational age in weeks, were obtained from the records of three maternity hospitals in the State of São Paulo, Brazil: 631 pairs from the Maternidade de Campinas (MC) in the city of Campinas, SP; 339 pairs from the Hospital Santa Catarina (HSC) in the city of São Paulo, SP, and 188 pairs from the Hospital e Maternidade Escola Vila Nova Cachoeirinha (VNC) also in the city of São Paulo, SP. MC is the largest obstetric center in Campinas; all social segments of the population are found among the women assisted there (40% private patients, 59% welfare program and 1% indigents). In contrast, HSC is mostly dedicated to private patients, while VNC is a public hospital where neither hospital nor medical care is charged, since it provides assistance to the poorest layers of the population. VNC is also a reference center that gives obstetrical care for high-risk poor pregnant women.

For the purposes of the present study, stillbirths were not omitted, but newborns weighing 500 g or less were excluded, as they are classified as abortions (Belitzki et al., 1978). Since birth weight has a large variance, the three hospitals were compared for this variable by the Kruskal-Wallis test. Intraclass correlation coefficients were calculated for the natural logarithms of the observed birth weights of the twins adjusted for gestational age in weeks, its quadratic and cubic terms, sex, and their interactions, by multiple stepwise regression analysis utilizing the SPSS® package.

RESULTS AND DISCUSSION

The results of the Kruskal-Wallis test, presented in Table I, show clearly that the twins born at the three hospitals cannot be pooled together, since they come from different birth weight populations. Mean birth weight was heavier at HSC, followed by MC, and VNC. Therefore, a multiple regression analysis was performed separately for each hospital. The natural logarithm of birth weight was considered as the dependent variable, while the independent variables included gestational age, (gestational age)², (gestational age)³, sex, gestational age sex, (gestational age)² ´ sex, and (gestational age)³ ´ sex.

Hospital Males Females No. Mean ± SD Mean rank No. Mean ± SD Mean rank HSC 361 2487 ± 530 659 317 2399 ± 546 635 MC 626 2335 ± 600 574 636 2268 ± 531 540 VNC 202 2272 ± 687 544 174 2231 ± 561 533 K-W Test 2 d.f. H = 19.351; P < 0.001 2 d.f. H = 19.856; P<0.001

Hospital	Males			Females
	No.	Mean ± SD	Mean rank	No.	Mean ± SD	Mean rank
HSC	361	2487 ± 530	659	317	2399 ± 546	635
MC	626	2335 ± 600	574	636	2268 ± 531	540
VNC	202	2272 ± 687	544	174	2231 ± 561	533
K-W Test	2 d.f.	H = 19.351; P < 0.001		2 d.f.	H = 19.856; P<0.001

Table I - Mean birth weights and standard deviations in grams of the twins born at three Brazilian hospitals, and Kruskal-Wallis test results.

HSC, Hospital Santa Catarina; MC, Maternidade de Campinas; VNC, Hospital e Maternidade Escola Vila Nova Cachoeirinha.

According to Table II, the independent variables selected for adjusting the birth weight of the newborns at HSC were gestational age, (gestational age)², and sex. For the newborns at MC the significant regression coefficients of the independent variables were gestational age, (gestational age)³, and sex, while for those at VNC were gestational age and (gestational age)³.

HSC MC VNC __ Y ± SD 7.76 ± 0.31 7.68 ± 0.32 7.66 ± 0.34 a -2.448 ± 0.617 1.696 ± 0.244 2.916 ± 0.490 gestational age 0.5099 ± 0.0360 0.2219 ± 0.0104 0.1666 ± 0.0214 (gestational age)2 -0.0063 ± 0.0005 - - (gestational age)3 - -4.3 x 10-5 ± 2.8 x 10-6 -2.6 x 10-5 ± 6.0 x 10-6 sex 0.0276 ± 0.0124 0.0331 ± 0.0109 - SQ regression 34.77 69.73 24.62 d.f. 3 3 2 SQ residual 17.96 47.70 18.85 d.f. 683 1266 373 F 440.66 617.15 243.60 r2 0.659 0.594 0.566

	HSC	MC	VNC
__ Y ± SD	7.76 ± 0.31	7.68 ± 0.32	7.66 ± 0.34
a	-2.448 ± 0.617	1.696 ± 0.244	2.916 ± 0.490
gestational age	0.5099 ± 0.0360	0.2219 ± 0.0104	0.1666 ± 0.0214
(gestational age)2	-0.0063 ± 0.0005	-	-
(gestational age)3	-	-4.3 x 10-5 ± 2.8 x 10-6	-2.6 x 10-5 ± 6.0 x 10-6
sex	0.0276 ± 0.0124	0.0331 ± 0.0109	-
SQ regression	34.77	69.73	24.62
d.f.	3	3	2
SQ residual	17.96	47.70	18.85
d.f.	683	1266	373
F	440.66	617.15	243.60
r2	0.659	0.594	0.566

Table II - Significant regression coefficients of data from three Brazilian hospitals, obtained by multiple regression analysis of the natural logarithms of twin birth weights in grams (Y) on gestational age, its quadratic and cubic terms, and sex, as well as their interactions.

For abbreviations see Table I.

Table III shows the intraclass correlation coefficients of like-sex and unlike-sex dizygotic twins born in the three hospitals, as well as the estimates necessary to calculate these coefficients. As expected, total variances of birth weight adjusted for sex and gestational age of like-sex and unlike-sex twins did not differ significantly at each hospital. Concerning within pair variances, in two hospitals (MC and VNC) unlike-sex twins showed a significantly larger within pair variance than like-sex pairs (F_(168;461) = 1.29 for MC, and F_(53;133) = 1.84 for VNC). Contrarily, in these two hospitals either the covariances or intraclass correlation coefficients of the adjusted birth weight for like-sex twins were larger than the corresponding values for unlike-sex pairs. However, significant differences between covariances (F_(133;53) = 2.39) and between intraclass correlation coefficients (t_¥ = 2.947) of like-sex and unlike-sex pairs were only assigned for VNC twins.

Estimates Like-sex (MM + FF) Unlike-sex Hospital Santa Catarina (HSC) s2 (B) 0.0366 ± 0.0033 0.0358 ± 0.0052 s2 (W) 0.0163 ± 0.0015 0.0134 ± 0.0019 Cov. 0.0102 ± 0.0018 0.0112 ± 0.0028 s2 (T) 0.0266 ± 0.0018 0.0246 ± 0.0028 ri 0.384 ± 0.054 0.455 ± 0.082 N 246 93 Maternidade de Campinas (MC) s2 (B) 0.0615 ± 0.0040 0.0575 ± 0.0062 s2 (W) 0.0126 ± 0.0008 0.0162 ± 0.0018 Cov. 0.0245 ± 0.0021 0.0207 ± 0.0032 s2 (T) 0.0370 ± 0.0021 0.0368 ± 0.0032 ri 0.660 ± 0.026 0.560 ± 0.053 N 462 169 Hospital e Maternidade Vila Nova Cachoeirinha (VNC) s2 (B) 0.0897 ± 0.0109 0.0644 ± 0.0123 s2 (W) 0.0181 ± 0.0022 0.0334 ± 0.0065 Cov. 0.0358 ± 0.0056 0.0150 ± 0.0069 s2 (T) 0.0537 ± 0.0056 0.0488 ± 0.0069 ri 0.664 ± 0.048 0.304 ± 0.122 N 134 54

Estimates	Like-sex (MM + FF)	Unlike-sex
Hospital Santa Catarina (HSC)
s2 (B)	0.0366 ± 0.0033	0.0358 ± 0.0052
s2 (W)	0.0163 ± 0.0015	0.0134 ± 0.0019
Cov.	0.0102 ± 0.0018	0.0112 ± 0.0028
s2 (T)	0.0266 ± 0.0018	0.0246 ± 0.0028
ri	0.384 ± 0.054	0.455 ± 0.082
N	246	93
Maternidade de Campinas (MC)
s2 (B)	0.0615 ± 0.0040	0.0575 ± 0.0062
s2 (W)	0.0126 ± 0.0008	0.0162 ± 0.0018
Cov.	0.0245 ± 0.0021	0.0207 ± 0.0032
s2 (T)	0.0370 ± 0.0021	0.0368 ± 0.0032
ri	0.660 ± 0.026	0.560 ± 0.053
N	462	169
Hospital e Maternidade Vila Nova Cachoeirinha (VNC)
s2 (B)	0.0897 ± 0.0109	0.0644 ± 0.0123
s2 (W)	0.0181 ± 0.0022	0.0334 ± 0.0065
Cov.	0.0358 ± 0.0056	0.0150 ± 0.0069
s2 (T)	0.0537 ± 0.0056	0.0488 ± 0.0069
ri	0.664 ± 0.048	0.304 ± 0.122
N	134	54

Table III - Intraclass correlation coefficients (r_i) of the natural logarithms of the twin birth weights of three Brazilian hospitals. Data adjusted for gestational age and sex according to Table II. s² (B) = between pairs variance; s² (W) = within pairs variance; Cov. = covariance; s² (T) = total variance; N = number of twin pairs.

Taking into account that: 1) among the twins born at HSC, the within pair variance of the adjusted birth weight for the surely dizygotic unlike-sex pairs was not larger, but even slightly smaller than that presented by like-sex pairs; 2) neither covariances nor intraclass correlation coefficients of like-sex and unlike-sex pairs born at MC differed significantly, it can be concluded that the twins born at HSC and MC did not fulfill the sine qua non conditions for investigating fetal genetic effect on birth weight. Therefore, a Holzinger’s-like formula (Holzinger, 1929), that is to say h² = (r_{il -} r_iu)/(1 - r_iu), and a Clark’s-like formula (Clark, 1955), i.e., h² = (s²_wu - s²_wl)/s²_wu, were applied only to the data in Table III for twins born at VNC. In these formulae, h² stands for "heritability", r_il and r_iu for the intraclass correlation coefficients of the respective like-sex and unlike-sex pairs, and s²_wuand s²_wl for the within pair variances of like-sex and unlike-sex twins, respectively.

The Holzinger’s-like formula applied to VNC data in Table III resulted in h² = 0.52, while Clark’s-like formula applied to the same data resulted in h² = 0.46. Otherwise stated, the data on twins born at hospitals attending basically high income patients (HSC) or women belonging to all social segments of the population (MC) gave no indication of fetal genetic effect on birth weight. Contrarily, the data on twins born at a public hospital that attends only poor and, usually, undernourished pregnant women (VNC) have indicated a strong fetal genetic effect on birth weight. How can this be explained?

It might be that the genetic effect on birth weight would be obscured in children born to well-nourished mothers, but would have the opportunity to be demonstrated among children born to undernourished women. Undernourishment, acting as a selective force, might have enhanced the existence of genotypes that determine less need of food for normal development.

ACKNOWLEDGMENTS

This work was supported by CNPq. Publication supported by FAPESP.

RESUMO

Os coeficientes de correlação intraclasse foram calculados para o peso de gêmeos nascidos em três maternidades do sudeste brasileiro, depois de ajustar os logaritmos naturais desses pesos para a idade gestacional, seus termos quadrático e cúbico, sexo e interações dessas variáveis. Os dados obtidos indicaram que o efeito genético fetal sobre o peso ao nascer teria a oportunidade de ser demonstrado por recém-nascidos de mães subnutridas. A subnutrição, atuando como força seletiva, realçaria a existência de genótipos que determinam menor necessidade nutricional para o desenvolvimento normal.

REFERENCES

Beiguelman, B., Colletto, G.M.D.D., Franchi-Pinto, C. and Krieger, H. (1998). Birth weight of twins: 1. The fetal growth patterns of twins and singletons. Braz. J. Genet. 21: 151-154.

Belitzki, R., Fescina, R.E. and Ucieda, F. (1978). Definiciones y terminologias aplicables al periodo perinatal. Recomendaciones de la Organización Mundial de la Salud y modificaciones de la F.I.G.O. Publicación Científica del CLAP, No. 757: 136-147.

Carr-Hill, R., Campbell, D.M., Hall, M.H. and Meredith, A. (1987). Is birth weigth determined genetically? Brit. Med. J. 295: 687-689.

Clark, P.J. (1955). The heritability of certain anthropometric characters as ascertained from measurements of twins. Am. J. Hum. Genet. 8: 49-54.

Fraccaro, M. (1957). A contribution to the study of birth weight based on an Italian sample twin data. Ann. Hum. Genet. 21: 224-236.

Holzinger, K.J. (1929). The relative effect of nature and nurture influences on twin differences. J. Educ. Psychol. 20: 241-248.

Karn, M.N. (1952). Birth weight and length of gestation of twins, together with maternal age, parity and survival rate. Ann. Eugen. 16: 365-377.

Karn, M.N., Lang-Brown, H., MacKenzie, H. and Penrose, L.S. (1951). Birth weight, gestation time and survival in sibs. Ann. Eugen. 15: 306-322.

Magnus, P. (1984). Distinguishing fetal and maternal genetic effects on variation in birth weight. Acta Genet. Med. Gemellol. 33: 481-486.

Morton, N.E. (1955). The inheritance of human birth weight. Ann. Hum. Genet. 20: 125-134.

Nance, W.E., Kramer, A.A., Corey, L.A., Winter, P.M. and Eaves, L.J. (1983). A causal analysis of birth weight in the offspring of monozygotic twins. Am. J. Hum. Genet. 35: 1211-1223.

Penrose, L.S. (1952). Data on the genetics of birth weight. Ann. Eugen. 16: 368-381.

Robson, E.B. (1955). Birth weight in cousins. Ann. Hum. Genet. 19: 262-268.

Turpin, R., Lejeune, J., Lafourcade, J., Chigot, P.L. and Salmon, C. (1961). Présomption de monozygotisme en dépit d’un dimorphisme sexuel: sujet masculin XY et sujet neutre haplo X.C.R. Acad. Sci. Paris 252: 2945-2946.

(Received October 8, 1997)

Publication Dates

• Publication in this collection
  06 Jan 1999
• Date of issue
  Mar 1998

History

• Received
  08 Oct 1997