British Journal of Obstetrics and Gynaecology July 1975. Vol. 82. pp. 562-567.
CORD PLASMA GLUCOSE AND INSULIN CONCENTRATIONS AND MATERNAL-FETAL RELATIONS BY
T. LIND ELIZABETH A. GILMORE AND
MAUREEN MCCLARENCE M.R.C. Reproduction and Growth Unit Princess Mary Maternity Hospital Newcastle upon Tyne Summary The concentration of glucose and insulin was estimated in 854 samples of umbilical cord blood plasma and 503 concurrently collected maternal blood samples. The mean cord insulin concentration, excluding all infants born to known diabetic mothers, was 7 pU./ml., but the distribution was skewed with 10 per cent of infants having a value of 12 pU./ml. or more. The giving of intravenous sugar-coating fluids to the mother during labour tended to exaggerate the skewness rather than move the whole distribution to the right. Babies born by the vaginal route, whatever the presentation or method of delivery, had slightly lower mean insulin values than those born by Caesarean section. It thus seems unlikely that “stress” is a factor causing high cord insulin values at birth. Other data concerning maternalfetal glucose and insulin relations are discussed.
THEplasma insulin concentration in the cord blood of babies born to insulin-treated diabetic mothers is higher than that found in babies of metabolically normal mothers. It is not immediately obvious why this should be so; it seems unlikely to be due to persistent maternal hyperglycaemia as, with good antenatal supervision, most diabetic patients can be maintained in a relatively normoglycaemic state. Again, while induced maternal anti-insulin gamma globulins can cross the placenta and interferewith immunoassay techniques for insulin determination, the original observations of this phenomenon were based on bioassay methods not so affected (Baird and Farquhar, 1962). In the course of a previous study (Lind et al., 1973) oral glucose tolerance tests (GTTs) were performed at intervals during pregnancy in a group of apparently healthy women selected as
being free of any stigmata of diabetes. One such volunteer patient was found to have an abnormal curve and shortly after delivery became an insulin-treated diabetic. The cord insulin concentration of her infant was 47 pU./ml. by a sensitive immunoassay technique (Lind et al., 1972). No insulin-binding activity could be demonstrated in the plasma. A further patient was found to be a gestational diabetic in that a markedly abnormal GTT at 32 weeks gestation returned to normal ten weeks after delivery; the cord insulin concentration of her infant was 54 pU./ml. The question we considered was whether a raised cord plasma insulin concentration might precede overt clinical diabetes in the mother and thus offer a cheap and easy method of screening large numbers of women for the early stage of diabetes mellitus. 562
MATERNAL AND FETAL GLUCOSE AND INSULIN
To begin to answer this question it was first necessary to define the range of cord plasma insulin concentrations in a large number of infants and to determine the various factors which might influence this concentration. In this paper we describe the plasma glucose and insulin concentrations from 854 cord blood samples and 503 concurrently collected maternal blood samples.
METHODS No selection of cases was attempted other than the exclusion of known diabetic patients, and the 854 patients represented virtually consecutive deliveries. The nursing staff collected the cord blood samples but the taking of maternal samples was a medical staff responsibility; as it was not practicable for a doctor to be present at every delivery there were fewer maternal blood samples available for analysis. A variety of infusion fluids were in current use: 5 per cent dextrose, mainly as a carrier for oxytocin and 5 per cent laevelose and Ringer lactate as part of the management of epidural analgesia. Although most patients receiving an infusion were having some intravenous sugar,
m w
2Li
I
l2-I
I
the multiplicity of regimes made any detailed classification unprofitable. For descriptive purposes, therefore, the patients have been divided into two major groups designated “Fluids” and “No Fluids”; the former were those who had received any intravenous solutions irrespective of type, volume or duration of infusion. Plasma glucose. The samples were collected into oxalate-fluoride tubes and stored at 4 “C. until assayed by a modified oxidase-peroxidase technique (Trinder, 1969). Previous work in this Unit had shown little glucose loss under these conditions. Plasma insulin. The samples were collected into heparin-lithium tubes and again kept at 4 “C. until collected by the laboratory staff. After centrifugation the plasma was snap-frozen in liquid nitrogen and stored at -25 “C. until assayed by a very sensitive single-antibody immunoassay technique as previously described (Lind et al., 1972).
RESULTS The distribution of the 854 cord plasma insulin values is shown graphically in Figure 1 ; the distribution of maternal and cord glucose and
I
I
I
2
6
10
14
563
18
22
26
30
Umbikal cord plasma insulin pU/mt FIG.1 Percentage distribution of cord plasma insulin values (854 cases).
>30
564
LIND, GILMORE AND MCCLARENCE
insulin values are given in Table I under subheadings according to whether the mothers did, or did not, receive intravenous fluids. The centile values for these variables are given in Table 11. It is clear that the giving of intravenous fluids does have some effect since the maternal and cord values for glucose and insulin were elevated in this group. The mean glucose and insulin values together with the standard deviations for both groups are shown in Table 111. From Table 1 we see that the main effect of giving the mother intravenous fluids was to exaggerate the skewness rather than move the whole distribution to the right. Thus the differences between the
-
- _ _ ~.
mean and median values (Tables I1 and 111) for maternal and cord glucose are 1 - 4 and 0 - 9 mg. per 100 ml. in the “No Fluid” group, but 1 1 -3 and 3.4 mg. per 100 ml. in the “Fluid” group. The corresponding insulin mean-median differences are 2.4 and 1.0 pU./ml. for the “No Fluids” and 3.1 and 2.2 for the “Fluid” group. Within the “Fluid” group there was no relation between the volume or duration of fluid given and the glucose and insulin concentrations in either the mother or infant. The effect of infusion is therefore difficult to quantify and is probably influenced by considerable variation in response between individuals.
TABLEI Distributions of maternal and cord glucose and insulin values with cumulative per cent ____ Maternal plasma Cord plasma No fluids Fluids No fluids Fluids ___No. Per cent No. Per cent No. Per cent No. Per cent
Glucose
(ms./
100 nil.) -40 -50 -60 -70 -80 -90
1 -
- 120 -130 -140 -150 150- __ Total
2 31 51 70 50 34 16 7 3 2 267
Insulin (&J./ml.) -2 -4 -6 -8 - 10 -12 -14 -16 -18 -20 -22 -24 -26 -28 -30 30 Total
2 23 49 40 35 28 24 10 15 8 6 5 6 1 3 11 266
- 100 -~I10
_-
0. 4 0.4 1. 1 12.7 31.8 58.1 76.8 89.5 95.5 98.1 98.1 99.3 -
1 13 19 48 38 38 19 16 11 7 26
0.4 5.9 14.0 34.3 50.4 66.5 74.6 81.4 86.0 89-0 -
3 20 71 101 132 88 32 14 10 3 8 1
-
483
~.
-
236
0.8 9. 4 27.8 42-9 56.0 66.5 75.6 79.3 85.0 88.0 90.2 92.1 94.4 94.7 95.9 -
7 28 31 31 28 17 17 12 13 12 10 5
-
5
2 18 236
-
3.0 14.8 28.0 41.1 53.0 60.2 67.4 72.5 78.0 83.1 87.3 89.4 91.5 92.4 -
13 132 120 110 48 17 17 7 7 2 4 1 -
-
483
5
0.8 4.8 19.5 40.4 67.7 85.9 92.6 95,5 97.5 98.1 99.8 100.0 -
2.1 30.0 54.9 71.6 87.6 91.1 94.6 96.1 97.5 97.9 98.8 99.0 99.0 99.0 99.0
-
1 5 24 56 73 67 53 40 15 14 7 4 12 371
0.3 1.6 8.1 23.2 42.9 60.9 75.2 86.0 90.0 93.8 95.7 96.8 -
6 94 108 48 34 20 16 8 10 7 7 1 4 1 1 6 371
1.6 27-0 56.1 69.0 78.2 83.6 87.9 90.0 92.7 94.6 96.5 96.8 97.8 98.1 98.2
-
-
-
MATERNAL AND FETAL GLUCOSE AND INSULIN
565
TABLE I1 Centile values for maternal and cord nlucose and insulin concentrations Centile
Determination
Glucose (mg.llO0 ml.) Maternal (no fluid) Maternal (fluid) Cord (no fluid) Cord (fluid) insulin (pu./ml.) Maternal (no fluid) Maternal (fluid) Cord (no fluid) Cord (fluid)
5
10
25
50
75
90
95
63.3 68.3 50.2 55.2
67.7 75.1 53.6 61.3
76.4 85.4 62.7 70.9
86.9 99.7 73-5 84.0
99.1 120.6 84-0 99.9
110.8 162.8 96.2 119.0
119.2 202.1 108.5 136.4
3.0
4.1 5.2 2.5 2.7
5.7 7.6 3.6 3.9
9.1 11.5 5.6 5.6
13.9 18.9 7.8 9.3
21.8 26.6 11.4 16.0
4.3 2.2 2.3
_
28.5 34.8 14.5 20.4 _ _
TABLE I11 Mean values and standard deviations for plasma glucose and insulin concentrations in mother and infant according to whether the mother received intravenous fluids ____ Determination “NO fluids” group “Fluids” group -_ 111.0*44.2 (236) 88.3&18.1 (267)* Maternal glucose (mg./100 ml.) 14.6&11.0 (236) Maternal insulin (pU./ml.) 11.5+ 8.9 (266) Cord glucose (mg./100 ml.) 74.4117.4 (483) 87.3k24.3 (371) Cord insulin (pU./ml.) 7 . 7 + 6.4 (371) 6.6+ 5 . 2 (483) * Figures in parentheses denote numbers in each group.
The effect of fluid therapy on the selection of a “cut-off’ point, if cord-insulin concentration is to be used as a clinical screening procedure, can be seen in Figure 2. If the insulin levels of infants in the top 10 per cent of each group were thought to indicate those mothers with potential diabetes, then for the “No Fluid” group the cord insulin concentration would be 12 pU./ml. and above whereas for the “Fluid” group it would be 16 pU./ml. and above. Variables considered as possibly influencing the concentration of cord insulin were cord glucose levels, maternal glucose levels, maternal insulin concentration and birthweight. The coefficients of correlation between these factors, and other variables selected for probable influence, are given in Table IV. In 12 women the non-pregnant weight was not available and the data are therefore drawn from the 490 paired samples for which all the variables were known. The correlations did not differ in any significant degree for those where the total numbers could be used. The log,, transformation was used for maternal and cord insulin values to overcome the effects of skewness on tests of significance and correlations. As might be anticipated, there is a moderately high correlation between the
c W Z
CK V
a W w c 2 1 Q 3
f0
INSULIN @U/rnl)
FIG.2 Cumulative distribution of cord plasma insulin values comparing 371 patients who received intravenous glucosecontaining fluids during labour with 483 patients who did not (solid line).
~
566
LIND, GILMORE AND MCCLARENCE
TABLEIV Correlations between selected variables for the 490 paired maternal and fetal samples for which all data were available. (The significance level of individual correlations is indicated as: 5 per cent=*; I per cent=* * and 0 . 1 per cent=* * *) Variables
“No fluid”
“Fluid”
0.268*** 0.119 0.126* 0.228***
0.285** * 0.197** 0.022 0.267***
x maternal glucose x log maternal insulin
-0-092 0.714*** 0.322***
-0-021 0.421***
Log maternal insulin x non-pregnant weight x maternal glucose
0.231*** 0.504***
0.123 0.318***
Maternal glucose x non-pregnant weight
0.054
Log cord insulin x birthweight x cord glucose x maternal glucose x log maternal insulin
Significance of the differences between the two groups N.S. N.S. N.S. N.S.
Cord glucose
x birthweight
glucose concentrations in cord and maternal plasma. There is also a degree of correlation between cord glucose and cord insulin levels; however, it can be calculated that given the maternal glucose value the correlation between cord insulin and cord glucose virtually disappears in the “NO Fluid” group (r = 0.041) although a significant, but small, correlation remains in the “Fluid” group (r = 0.207). The correlations for most variables are similar whether or not the mothers received intravenous fluids (Table IV). Where the two groups do differ the change is probably explicable on the grounds that mothers not receiving fluids are virtually in a basal or “fasted” state and reasonable correlations between plasma glucose and insulin can be anticipated. The mothers given sugar-containing fluids are in a “fed” state and less correlation between glucose and insulin in both mother and fetus at any given moment is to be expected. Accepting those mothers who did not receive intravenous fluids as the more “physio1ogical” group the next factor considered as likely to influence the cord glucose and insulin concentrations was the method of delivery. There was no significant difference in either of these concentrations whether the baby was delivered as
N.S. 1 per cent 5 per cent
0.510***
N.S. 5 per cent
-0.042
N.S.
a spontaneous vaginal delivery, as a breech or by forceps. Interestingly, those babies born by Caesarean section had higher insulin and lower glucose levels in their cord blood and this difference achieved significance for the 35 patients so delivered (Table V). It could be argued that, as many of the Caesarean sections were “elective”, this method was less stressful to the babies than any form of vaginal delivery and hence represents more truly the intrauterine state. If so, it might be anticipated that the longer the duration of labour preceding vaginal delivery the more “stressful” the situation. This TABLEV Cord plasma glucose and insulin values according to the method of delivery in those women who did not receive intravenousfluids during labour Type of delivery
Glucose (mg./lOO ml.)
Vertex Breech Forceps delivery Caesarean section
74-7&16-6 (411) 96.3526.5 (4) 81.2rt21.7 (33) 61.8f13.8* (35)
Insulin (dJ./ml.) 6-4455.14 6.28*3*11 6.76*6.50 8.3154-74*
Significant differences from the values for vertex deliveries are indicated by * (0.1 per cent and 1 per cent for glucose and insulin respectively).
MATERNAL AND FETAL GLUCOSE AND INSULIN
does not seem to be the case because for those women having spontaneous vaginal deliveries and not receiving intravenous fluids, the cord insulin values were 7 - 4 5 6 . 7 (62 cases), 6 . 3 f 4 - 6 (155 cases), 6 . 3 h 5 . 6 (117 cases), 6 . 4 & 4 * 1 (53 cases) and 6 . 2 f 3 - 5 (24 cases) for labour durations of under three hours, 3 to 6, 6 to 9, 9 to 12 and more than 12 hours respectively. The differences between insulin values for those labouring for less than three hours and the other groups is nowhere significant.
DISCUSSION The distribution of cord plasma insulin concentrations in a large group of infants shows a markedly skewed distribution. The mean value is 7 pU./ml., or 8 pU./ml. if the mother received intravenous fluids, but concentrations of 30 pU./ml. or more were recorded in 12 infants. While variables such as the giving of intravenous fluids to the mother, birthweight and whether delivery was by Caesarean section influenced the cord insulin level to a small degree none of these factors could account for the wide range of values recorded. It therefore seems reasonable to infer that some factor other than maternal or fetal plasma glucose levels, or any “stress” situation, is stimulating some fetuses to produce much more insulin than others. What cannot be decided is whether those babies with high levels were producing increased amounts of insulin in utero for days, or even weeks, prior to labour or whether it arose for the first time during labour itself. It is interesting to note, however, that a significant positive correlation was found between cord insulin concentration and birthweight and it has been postulated that insulin might have a growth promoting role for the fetus in utero.
567
If labour caused “stress” in the fetus, a situation more likely to arise in the larger babies delivered vaginally, one might expect that any catecholamine-type of response would suppress rather than stimulate insulin secretion. In this context we have noted that babies born by the vaginal route, whether assisted or spontaneous, have similar cord insulin values while others delivered by the perhaps less stressful method of Caesarean section have slightly higher cord values. The original premise which stimulated this particular work was that high cord insulin values might indicate those mothers with disordered carbohydrate metabolism but who had no clinical symptoms. At the present time all those mothers who delivered infants with a cord insulin value of 15 pU./ml. or more or in whom the fetal level was at least twice as high as the maternal concentration are being asked to return for a 50 g. oral glucose tolerance test. As many women as possible in the group who had spontaneous vaginal deliveries, and not in the above categories, are also being asked to return for a similar test to act as a control series. We hope to report the results of this study in the near future. ACKNOWLEDGEMENTS We are grateful to our medical colleagues a d especially our nursing colleagues in the labour ward, for their co-operation in this study. REFERENCES Baird, J. D., and Farquhar, J. W. (1962): Lancer, 1, 71. Lind, T., Billewicz, W. Z., and Brown, G . (1973): Journal of Obstetrics and Gynaecology of the British Commonwealth, 80, 1033. Lind, T., Van de Groot, H. A., Brown, G., and Cheyne, G . A. (1972): British Medical Journal, 3, 320. Trinder, P. (1969): Annals of Clinical Biochemistry, 6, 24.
CORD PLASMA GLUCOSE AND INSULIN CONCENTRATIONS AND MATERNAL-FETAL RELATIONS BY
T. LIND ELIZABETH A. GILMORE AND
MAUREEN MCCLARENCE M.R.C. Reproduction and Growth Unit Princess Mary Maternity Hospital Newcastle upon Tyne Summary The concentration of glucose and insulin was estimated in 854 samples of umbilical cord blood plasma and 503 concurrently collected maternal blood samples. The mean cord insulin concentration, excluding all infants born to known diabetic mothers, was 7 pU./ml., but the distribution was skewed with 10 per cent of infants having a value of 12 pU./ml. or more. The giving of intravenous sugar-coating fluids to the mother during labour tended to exaggerate the skewness rather than move the whole distribution to the right. Babies born by the vaginal route, whatever the presentation or method of delivery, had slightly lower mean insulin values than those born by Caesarean section. It thus seems unlikely that “stress” is a factor causing high cord insulin values at birth. Other data concerning maternalfetal glucose and insulin relations are discussed.
THEplasma insulin concentration in the cord blood of babies born to insulin-treated diabetic mothers is higher than that found in babies of metabolically normal mothers. It is not immediately obvious why this should be so; it seems unlikely to be due to persistent maternal hyperglycaemia as, with good antenatal supervision, most diabetic patients can be maintained in a relatively normoglycaemic state. Again, while induced maternal anti-insulin gamma globulins can cross the placenta and interferewith immunoassay techniques for insulin determination, the original observations of this phenomenon were based on bioassay methods not so affected (Baird and Farquhar, 1962). In the course of a previous study (Lind et al., 1973) oral glucose tolerance tests (GTTs) were performed at intervals during pregnancy in a group of apparently healthy women selected as
being free of any stigmata of diabetes. One such volunteer patient was found to have an abnormal curve and shortly after delivery became an insulin-treated diabetic. The cord insulin concentration of her infant was 47 pU./ml. by a sensitive immunoassay technique (Lind et al., 1972). No insulin-binding activity could be demonstrated in the plasma. A further patient was found to be a gestational diabetic in that a markedly abnormal GTT at 32 weeks gestation returned to normal ten weeks after delivery; the cord insulin concentration of her infant was 54 pU./ml. The question we considered was whether a raised cord plasma insulin concentration might precede overt clinical diabetes in the mother and thus offer a cheap and easy method of screening large numbers of women for the early stage of diabetes mellitus. 562
MATERNAL AND FETAL GLUCOSE AND INSULIN
To begin to answer this question it was first necessary to define the range of cord plasma insulin concentrations in a large number of infants and to determine the various factors which might influence this concentration. In this paper we describe the plasma glucose and insulin concentrations from 854 cord blood samples and 503 concurrently collected maternal blood samples.
METHODS No selection of cases was attempted other than the exclusion of known diabetic patients, and the 854 patients represented virtually consecutive deliveries. The nursing staff collected the cord blood samples but the taking of maternal samples was a medical staff responsibility; as it was not practicable for a doctor to be present at every delivery there were fewer maternal blood samples available for analysis. A variety of infusion fluids were in current use: 5 per cent dextrose, mainly as a carrier for oxytocin and 5 per cent laevelose and Ringer lactate as part of the management of epidural analgesia. Although most patients receiving an infusion were having some intravenous sugar,
m w
2Li
I
l2-I
I
the multiplicity of regimes made any detailed classification unprofitable. For descriptive purposes, therefore, the patients have been divided into two major groups designated “Fluids” and “No Fluids”; the former were those who had received any intravenous solutions irrespective of type, volume or duration of infusion. Plasma glucose. The samples were collected into oxalate-fluoride tubes and stored at 4 “C. until assayed by a modified oxidase-peroxidase technique (Trinder, 1969). Previous work in this Unit had shown little glucose loss under these conditions. Plasma insulin. The samples were collected into heparin-lithium tubes and again kept at 4 “C. until collected by the laboratory staff. After centrifugation the plasma was snap-frozen in liquid nitrogen and stored at -25 “C. until assayed by a very sensitive single-antibody immunoassay technique as previously described (Lind et al., 1972).
RESULTS The distribution of the 854 cord plasma insulin values is shown graphically in Figure 1 ; the distribution of maternal and cord glucose and
I
I
I
2
6
10
14
563
18
22
26
30
Umbikal cord plasma insulin pU/mt FIG.1 Percentage distribution of cord plasma insulin values (854 cases).
>30
564
LIND, GILMORE AND MCCLARENCE
insulin values are given in Table I under subheadings according to whether the mothers did, or did not, receive intravenous fluids. The centile values for these variables are given in Table 11. It is clear that the giving of intravenous fluids does have some effect since the maternal and cord values for glucose and insulin were elevated in this group. The mean glucose and insulin values together with the standard deviations for both groups are shown in Table 111. From Table 1 we see that the main effect of giving the mother intravenous fluids was to exaggerate the skewness rather than move the whole distribution to the right. Thus the differences between the
-
- _ _ ~.
mean and median values (Tables I1 and 111) for maternal and cord glucose are 1 - 4 and 0 - 9 mg. per 100 ml. in the “No Fluid” group, but 1 1 -3 and 3.4 mg. per 100 ml. in the “Fluid” group. The corresponding insulin mean-median differences are 2.4 and 1.0 pU./ml. for the “No Fluids” and 3.1 and 2.2 for the “Fluid” group. Within the “Fluid” group there was no relation between the volume or duration of fluid given and the glucose and insulin concentrations in either the mother or infant. The effect of infusion is therefore difficult to quantify and is probably influenced by considerable variation in response between individuals.
TABLEI Distributions of maternal and cord glucose and insulin values with cumulative per cent ____ Maternal plasma Cord plasma No fluids Fluids No fluids Fluids ___No. Per cent No. Per cent No. Per cent No. Per cent
Glucose
(ms./
100 nil.) -40 -50 -60 -70 -80 -90
1 -
- 120 -130 -140 -150 150- __ Total
2 31 51 70 50 34 16 7 3 2 267
Insulin (&J./ml.) -2 -4 -6 -8 - 10 -12 -14 -16 -18 -20 -22 -24 -26 -28 -30 30 Total
2 23 49 40 35 28 24 10 15 8 6 5 6 1 3 11 266
- 100 -~I10
_-
0. 4 0.4 1. 1 12.7 31.8 58.1 76.8 89.5 95.5 98.1 98.1 99.3 -
1 13 19 48 38 38 19 16 11 7 26
0.4 5.9 14.0 34.3 50.4 66.5 74.6 81.4 86.0 89-0 -
3 20 71 101 132 88 32 14 10 3 8 1
-
483
~.
-
236
0.8 9. 4 27.8 42-9 56.0 66.5 75.6 79.3 85.0 88.0 90.2 92.1 94.4 94.7 95.9 -
7 28 31 31 28 17 17 12 13 12 10 5
-
5
2 18 236
-
3.0 14.8 28.0 41.1 53.0 60.2 67.4 72.5 78.0 83.1 87.3 89.4 91.5 92.4 -
13 132 120 110 48 17 17 7 7 2 4 1 -
-
483
5
0.8 4.8 19.5 40.4 67.7 85.9 92.6 95,5 97.5 98.1 99.8 100.0 -
2.1 30.0 54.9 71.6 87.6 91.1 94.6 96.1 97.5 97.9 98.8 99.0 99.0 99.0 99.0
-
1 5 24 56 73 67 53 40 15 14 7 4 12 371
0.3 1.6 8.1 23.2 42.9 60.9 75.2 86.0 90.0 93.8 95.7 96.8 -
6 94 108 48 34 20 16 8 10 7 7 1 4 1 1 6 371
1.6 27-0 56.1 69.0 78.2 83.6 87.9 90.0 92.7 94.6 96.5 96.8 97.8 98.1 98.2
-
-
-
MATERNAL AND FETAL GLUCOSE AND INSULIN
565
TABLE I1 Centile values for maternal and cord nlucose and insulin concentrations Centile
Determination
Glucose (mg.llO0 ml.) Maternal (no fluid) Maternal (fluid) Cord (no fluid) Cord (fluid) insulin (pu./ml.) Maternal (no fluid) Maternal (fluid) Cord (no fluid) Cord (fluid)
5
10
25
50
75
90
95
63.3 68.3 50.2 55.2
67.7 75.1 53.6 61.3
76.4 85.4 62.7 70.9
86.9 99.7 73-5 84.0
99.1 120.6 84-0 99.9
110.8 162.8 96.2 119.0
119.2 202.1 108.5 136.4
3.0
4.1 5.2 2.5 2.7
5.7 7.6 3.6 3.9
9.1 11.5 5.6 5.6
13.9 18.9 7.8 9.3
21.8 26.6 11.4 16.0
4.3 2.2 2.3
_
28.5 34.8 14.5 20.4 _ _
TABLE I11 Mean values and standard deviations for plasma glucose and insulin concentrations in mother and infant according to whether the mother received intravenous fluids ____ Determination “NO fluids” group “Fluids” group -_ 111.0*44.2 (236) 88.3&18.1 (267)* Maternal glucose (mg./100 ml.) 14.6&11.0 (236) Maternal insulin (pU./ml.) 11.5+ 8.9 (266) Cord glucose (mg./100 ml.) 74.4117.4 (483) 87.3k24.3 (371) Cord insulin (pU./ml.) 7 . 7 + 6.4 (371) 6.6+ 5 . 2 (483) * Figures in parentheses denote numbers in each group.
The effect of fluid therapy on the selection of a “cut-off’ point, if cord-insulin concentration is to be used as a clinical screening procedure, can be seen in Figure 2. If the insulin levels of infants in the top 10 per cent of each group were thought to indicate those mothers with potential diabetes, then for the “No Fluid” group the cord insulin concentration would be 12 pU./ml. and above whereas for the “Fluid” group it would be 16 pU./ml. and above. Variables considered as possibly influencing the concentration of cord insulin were cord glucose levels, maternal glucose levels, maternal insulin concentration and birthweight. The coefficients of correlation between these factors, and other variables selected for probable influence, are given in Table IV. In 12 women the non-pregnant weight was not available and the data are therefore drawn from the 490 paired samples for which all the variables were known. The correlations did not differ in any significant degree for those where the total numbers could be used. The log,, transformation was used for maternal and cord insulin values to overcome the effects of skewness on tests of significance and correlations. As might be anticipated, there is a moderately high correlation between the
c W Z
CK V
a W w c 2 1 Q 3
f0
INSULIN @U/rnl)
FIG.2 Cumulative distribution of cord plasma insulin values comparing 371 patients who received intravenous glucosecontaining fluids during labour with 483 patients who did not (solid line).
~
566
LIND, GILMORE AND MCCLARENCE
TABLEIV Correlations between selected variables for the 490 paired maternal and fetal samples for which all data were available. (The significance level of individual correlations is indicated as: 5 per cent=*; I per cent=* * and 0 . 1 per cent=* * *) Variables
“No fluid”
“Fluid”
0.268*** 0.119 0.126* 0.228***
0.285** * 0.197** 0.022 0.267***
x maternal glucose x log maternal insulin
-0-092 0.714*** 0.322***
-0-021 0.421***
Log maternal insulin x non-pregnant weight x maternal glucose
0.231*** 0.504***
0.123 0.318***
Maternal glucose x non-pregnant weight
0.054
Log cord insulin x birthweight x cord glucose x maternal glucose x log maternal insulin
Significance of the differences between the two groups N.S. N.S. N.S. N.S.
Cord glucose
x birthweight
glucose concentrations in cord and maternal plasma. There is also a degree of correlation between cord glucose and cord insulin levels; however, it can be calculated that given the maternal glucose value the correlation between cord insulin and cord glucose virtually disappears in the “NO Fluid” group (r = 0.041) although a significant, but small, correlation remains in the “Fluid” group (r = 0.207). The correlations for most variables are similar whether or not the mothers received intravenous fluids (Table IV). Where the two groups do differ the change is probably explicable on the grounds that mothers not receiving fluids are virtually in a basal or “fasted” state and reasonable correlations between plasma glucose and insulin can be anticipated. The mothers given sugar-containing fluids are in a “fed” state and less correlation between glucose and insulin in both mother and fetus at any given moment is to be expected. Accepting those mothers who did not receive intravenous fluids as the more “physio1ogical” group the next factor considered as likely to influence the cord glucose and insulin concentrations was the method of delivery. There was no significant difference in either of these concentrations whether the baby was delivered as
N.S. 1 per cent 5 per cent
0.510***
N.S. 5 per cent
-0.042
N.S.
a spontaneous vaginal delivery, as a breech or by forceps. Interestingly, those babies born by Caesarean section had higher insulin and lower glucose levels in their cord blood and this difference achieved significance for the 35 patients so delivered (Table V). It could be argued that, as many of the Caesarean sections were “elective”, this method was less stressful to the babies than any form of vaginal delivery and hence represents more truly the intrauterine state. If so, it might be anticipated that the longer the duration of labour preceding vaginal delivery the more “stressful” the situation. This TABLEV Cord plasma glucose and insulin values according to the method of delivery in those women who did not receive intravenousfluids during labour Type of delivery
Glucose (mg./lOO ml.)
Vertex Breech Forceps delivery Caesarean section
74-7&16-6 (411) 96.3526.5 (4) 81.2rt21.7 (33) 61.8f13.8* (35)
Insulin (dJ./ml.) 6-4455.14 6.28*3*11 6.76*6.50 8.3154-74*
Significant differences from the values for vertex deliveries are indicated by * (0.1 per cent and 1 per cent for glucose and insulin respectively).
MATERNAL AND FETAL GLUCOSE AND INSULIN
does not seem to be the case because for those women having spontaneous vaginal deliveries and not receiving intravenous fluids, the cord insulin values were 7 - 4 5 6 . 7 (62 cases), 6 . 3 f 4 - 6 (155 cases), 6 . 3 h 5 . 6 (117 cases), 6 . 4 & 4 * 1 (53 cases) and 6 . 2 f 3 - 5 (24 cases) for labour durations of under three hours, 3 to 6, 6 to 9, 9 to 12 and more than 12 hours respectively. The differences between insulin values for those labouring for less than three hours and the other groups is nowhere significant.
DISCUSSION The distribution of cord plasma insulin concentrations in a large group of infants shows a markedly skewed distribution. The mean value is 7 pU./ml., or 8 pU./ml. if the mother received intravenous fluids, but concentrations of 30 pU./ml. or more were recorded in 12 infants. While variables such as the giving of intravenous fluids to the mother, birthweight and whether delivery was by Caesarean section influenced the cord insulin level to a small degree none of these factors could account for the wide range of values recorded. It therefore seems reasonable to infer that some factor other than maternal or fetal plasma glucose levels, or any “stress” situation, is stimulating some fetuses to produce much more insulin than others. What cannot be decided is whether those babies with high levels were producing increased amounts of insulin in utero for days, or even weeks, prior to labour or whether it arose for the first time during labour itself. It is interesting to note, however, that a significant positive correlation was found between cord insulin concentration and birthweight and it has been postulated that insulin might have a growth promoting role for the fetus in utero.
567
If labour caused “stress” in the fetus, a situation more likely to arise in the larger babies delivered vaginally, one might expect that any catecholamine-type of response would suppress rather than stimulate insulin secretion. In this context we have noted that babies born by the vaginal route, whether assisted or spontaneous, have similar cord insulin values while others delivered by the perhaps less stressful method of Caesarean section have slightly higher cord values. The original premise which stimulated this particular work was that high cord insulin values might indicate those mothers with disordered carbohydrate metabolism but who had no clinical symptoms. At the present time all those mothers who delivered infants with a cord insulin value of 15 pU./ml. or more or in whom the fetal level was at least twice as high as the maternal concentration are being asked to return for a 50 g. oral glucose tolerance test. As many women as possible in the group who had spontaneous vaginal deliveries, and not in the above categories, are also being asked to return for a similar test to act as a control series. We hope to report the results of this study in the near future. ACKNOWLEDGEMENTS We are grateful to our medical colleagues a d especially our nursing colleagues in the labour ward, for their co-operation in this study. REFERENCES Baird, J. D., and Farquhar, J. W. (1962): Lancer, 1, 71. Lind, T., Billewicz, W. Z., and Brown, G . (1973): Journal of Obstetrics and Gynaecology of the British Commonwealth, 80, 1033. Lind, T., Van de Groot, H. A., Brown, G., and Cheyne, G . A. (1972): British Medical Journal, 3, 320. Trinder, P. (1969): Annals of Clinical Biochemistry, 6, 24.
