BtitishJouraal ofHaematology, 1975,29,
221.
Haernoglobin Bart’s in Saudi Arabia M. E. PEMBREY, D. J. WEATHERALL,* J. B. CLEGG,* c. BUNCHAND R. P. PERRINEf Departrncnt of Haenzafology, Uriiversity of Livcrpool, arid tArabiarz Americari Oil C O M I ~Dhahrari, U I Z ~ , Sartdi Arabia
SUMMARY. The liaemoglobin (Hb) pattcrns of 345 Shiite Saudi Arab cord bloods wcrc examined by alkaline starch-gel electropliorcsis. A fast-moving component, identified by structural analysis as Hb Bart’s, was found in 52% of cases, the highest incidence of this variant yet rccordcd. Tlic levels of Hb Bart’s ranged from 0.5 to 16% ofthe total haenioglobin. The relative rates of synthcsis of the a , P and ?-chains, measured by [3H]lcuci~ie incorporation, were estimated in IZiicwborii Arab infants. There was an excellent corrclatioii between the amount of Hb Bart’s and tlie a/nona-globin-chain production ratio. Furthermore tlierc was a significant correlatioii between tlic lcvel of Hb Bart’s and morpliological abnorinalitics of the red cells and the mean cell haemoglobiii (MCH). These findings indicate that elevated levels of Hb Bart’s in this population are due to tlie presence of a thalassacinia. The absciice of hydrops fetalis and the rarity of Hb-H disease despitc the intense iiibrccding in tliis population, points to an a-tlialassacmia genotype that is, in tcrins of phenotypic cxprcssion, ititcrnicdiatc bctwccii tlic lieterozygous state for a-tlialassacmia I and Hb-H discasc. A possible ~iiolecularbasis for this genotype is suggcsted. Iluring tlic pcriiiatal pcriod liaemoglobin A (Hb A, a2Pz) replaces liaenioglobin F (Hb F, tlie main liaenioglobin. Normally this process proceeds smoothly without any gross imbalance of globin-chain production, although trace aniounts of Hb Bart’s, a variant with molecular formula y 4 (Hunt & Lchmanii, I959), are found in most normal cord bloods, indicating a slight degrec of chain imbalance during the switching ovcr process. It is now well cstablislied that tlic presence of increased amounts of Hb Bart’s in the iicoiiatal period is the most sciisitivc iiidicator of the presence of an a-thalassaciiiia gene (Weatherall, 1963 ; Wasi c’t (I\, 1964, 1969; Lchniann, 1970). O n the other hand, Esaii (1970) has suggestcd that iiicreascd amounts of Hb Bart’s do not always indicate the presence o f a tlialassaeniia and may represent a developmental abiiorniality. That a tlialassaemia may occur in Saudi Arabia was suggestcd by reports of the occurrcncc of Hb-H discasc or its variants in Shiite Arabs by Weatherall et nl(1969a) and McNcil (1971). Since littlc is kiwwii of tlie incideiicc or clinical features of a tlialassaeniia in tlic Middlc East wc decided to iet up a survey of tlic liacnioglobins of iicwborii infants in Eastern Saudi Arabia. This d s carried out from 1967 until 1973 and the results indicate that this population has the higlielt iiicideiicc of clcvated levels of Hb Bart’s yet rccorded. The results of this
a 2 y 2 )as
* Present address: Nufield Department of Clinical Medicine, University of Oxford. Correspondence: Dr M. E. Pembrey, Department of Medicine, Guy’s Hospital Medical School, London, S.E.1. 221
222
M. E. Pembrey et
a1
survey and our further attempts to determine the significance of this unexpected finding form the basis of this paper.
SUBJECTS STUDIED A N D METHODS
Stibjects. The investigations were confined to Shiite Arabs from the Eastern Province of Saudi Arabia, who represent the ancient population from the Qatif and A1 Hasa Oases. All subjects were family dependents of the Arabian American Oil Company (ARAMCO) employees and were ascertained through the company’s own hospital or hospitals in the local town, A1 Khobar, when attending for routine maternal and child care. Over 50 ooo ‘dependents’, about 25% of whom are Shiites, come under the company’s health program and there is no indication that ‘dependents’ differ genetically from other Arabs in the Eastern Province. Cord bloods were taken routinely at the delivery of all Shiite babies. The majority of deliveries take place in hospital, the only factor selecting against this being the difficulty of transport from some outlying villages. Thc cord bloods were therefore a representative sample from Shiite Arabs living in the vicinity of Dhahran. The subsample on which hacmoglobin synthesis studies were carried out was similarly unselected. Infants were followed up at the company’s routine child-care clinics. In order to examine the levels of Hb Bart’s and the relativc rates of synthesis of the CL,/i’ aiid y chains of normal British Infants, cord-blood sainplcs were obtained from babies born at the Liverpool Maternity Hospital. Samples were collected into acid-citrate-dextrose (ACD) and thosc which wcre used for biosyiithetic experiments were studied within I h of collection. Collcctioir arid tmrzrport. 5-10 in1 of cord blood was takcii into ACD and one drop of 1% KCN added. The samples were kept at 4’C for 110 loiiger than I week before being flown to Liverpool as chilled cargo. The vast majority of samples arrived in excellent condition whilst the occasional partially haemolyscd sample was discarded. 10in1 of cord blood was collected into ACD within 14 h of delivery for iri vitro hacmoglobin synthesis studies and, after incubation, the red cells were stored at - 2o°C and transported on dry ice. Haemoglobiri arrulysis. Red-cell 1ysatcs were prepared by standard methods aiid aiialyscd by starch-gel electrophoresis using a tris-EDTA-borate system, pH 8.5, or a phosphatcbuffer system, pH 7.0 (Weatherall & Clcgg, 1972). For quantitation of tlie level of Hb Bart’s duplicate samples were elcctrophorescd 011 starch block using an 0.05 M phosphate buffer system, pH 7.0. The fractions containing Hb Bart’s and the remaining haemoglobins were eluted scyaratcly aiid tlieir relative proportions determined as described by Weatherall & Clegg (1972). In some cases haemoglobiiis were aiialysed by Amberlite IRC-50 chromatography under tlie conditions of Allen ct aI (1958). The samples were dialyscd against developer 2 for 24 11 at 4°C and then applied to 3 by 30 cin columiis of Ambcrlite IRC 50 equilibrated against the same developer at 4°C. The fiactions which were uiiretaiiicd at 4°C were eluted completely and then the column was warnicd to 25°C to obtain the remainder of the liaemoglobiii com-
Haerrioglobiii Bart’s in Saudi Arabia
2-23
pormits. Followiiig chromatography each fraction was coiicciitrated in vaciio and aiialyscd by starch-gel electrophoresis. In ordcr to determine the Structure of the fast-moving haemoglobins, presumed to be Hb Bart’s, they wcre isolated by starcli-block electrophoresis and coiivcrtcd to globin, aftcr wliicli they were digested witli trypsiii and fingerprintcd as described by Clcgg et a1 (1966). Iinerrtoglobin synthesis. Cord blood was collected into ACD and the cells wcre washed three times and incubated as described by Wcatlierall et a1 (1969b). After approximately 10 min 50 pCi of [3H]lcuciiic was added aiid the incubation allowed to contiiiuc for 2 11. The cells were then waslied three times and frozen. After thawing tlic lysatc was coiivertcd to ‘globin’ without any furtlier purification. The ‘globin’ was applied to a CM-cellulose column equilibrated witli 8 M urca/nicrcaptoctliaiiol, pH 6.8, as described by Clcgg et a/ (1966), and the CI, fl aiid y chains were separated. Tlie radioactivity incorporated into tlic cliains was detcriiiincd as previously described (Wcatlierall et al, 1969b). Tlie tubes coiitaiiiiiig tlie chains were then pooled and dialyscd against 0.5% formic acid aftcr which their specific activities wcre dcterinined.
RESULTS
Detlcrriiiizatioii of the Structure of the Rapidly-MigrafifigHaeriioglobiti in Saridi Cord Bloods Early on in this study it became obvious that a liigli proportion of the cord bloods coiltained a rapidly-migrating liaemoglobin. This fraction migrated in tlic position of Hb Bart’s 011 starch-gel or starch-block electrophoresis, pH 8.6 and 6.5. It was isolated by starch-block clectroplioresis from two samples containing 5% and 1% rcspectively. Thcsc components wcre clicckcd for purity by starch-gel clcctrophorcsis and then coiivcrted to globin and finger-printed. Thc rcsultaiit pcptidc imps showed y-chain pcptidcs only, iiidicatiiig that the rapidly-migrating hacmoglobiii is liaciiioglobiii Bart’s. of Hb Unrt’s by Starch-Gel Elcctroyhoresis As a prcliiiiinary to this survey a scrics of cxpcriiiiciits wcrc pcrforiiicd to coiifirin the adcquacy of the scrcciiiiig technique for tlic prcsciicc of Hb Bart’s in cord-blood aiid to asccrtaiii the lowest level of H b Bart’s which can bc discriminated by iiictliods which arc currently available. Scrial dilutions of a cord blood liaeiiiolysatc coiitainiiig about 5% Hb Bart’s wcrc iiiadc with a normal adult liacniolysatc of the same coiiccntratioii and the rcsulting niixturcs wcrc analyscd by starch-gcl clcctrophoresis at pH 8.5 and 6.5, as describcd in the Methods section. The gcls wcrc staincd with a bciizidinc stain (Wcathcrall & Clcgg, 1972) aiid tlic dilution at which Hb Bart’s was clearly visible was dctcrinincd. It was found that at pH 8.6 Hb Bart’s was easily distinguishable wlicii prcsciit as 0.5% of tlic total liacnioglobin; similar results were obtaiiicd on tlic pH 6.5 gel. Tlius in tlie screening program all cord bloods wcre studied at a liaciiiolysatc coiiccntration of 8-10 g/dl, using starch-gcl clcctrophoresis witli a tris-EDTA-borate system, pH 8.6. Using this system Hb Bart’s was not see11 in 50 coiisccutivc British cord-blood samples. A furtlicr scries of experiments were pcrforiiied to dctcrniiric tlie accuracy of tlic determination of relatively small amounts of Hb Bart’s by starch-block electrophoresis. Ten
The Der?iotistratioiz
M. E. Pembrey et a1
224
determinations were carried out on a cord-blood sample containing approximately Bart’s. The values raiigcd from 0.9 to 1.4% (mean 1.15%; SD+o.27%).
I
.OXHb
Incidence and Levels of Hb Bart’s 345 cord bloods were run on starch gel and 182 samples had a visible Hb-Bart’s band making an overall iticideiice of 52.6%. In the first 325 samples 176 had Hb Bart’s visible on starch gel and in each case the level of Hb Bart’s was estimated in duplicate by starch-block electrophoresis; these arc suiniiiarizcd in Fig I . 169 samples (52%) had more than 0.5% Hb Bart’s. 149 samples (45.8%) had no visible Hb-Bart’s band and caii be assumed to have contained less than 0.5% Hb Bart’s, and seven cases (2.2%) had a measured level of less than 0.5%, briiigiiig tlic total with less than 0.5% Hb Bart’s to 156 (48%). 24
-
21 18
-
15 -
,“
129-
63-
0Hb Bart’s (%)
FIG I. Distribution of Hb Bart’s in 176 Shiite Saudi ncwboriis. Total survey = Incidence H b Bart’s = 54.2%.
325
cord bloods
Distribiition o f H b Bart’s in 176 Cord Bloods The distribution of the Hb-Bart’s levels foulid in the cord-blood samples shown in Fig I is clearly skewed to the right, with a discreet upper tail comprising 14 cases with levels between 10 and 17%. O n inspection the main body ofthe distribution appears to be bimodal, or possibly even trimodal, with an aiitiniode at 2.2% and possibly one at 6.0%. Attempts were made to fit one to four ‘normal curves’ to the whole distribution employing the method of maxiiiiuni likelihood using the MINUIT program of F. James and M. Roos (Cern computer program library). Within the constraints imposed by requiring one, two, three or four normal curves, the ‘best’ curves were fitted and the mean, standard deviatioii slid proportion of the total for each curve were estimated. The reliability of these estimates was somewhat reduced by the relatively small total number of observations. When more than one curve was concerned the program also provided the probability of correct classification of any given Hb-Bart’s level into one or other of the groups. In each case the corn-
Haemoglobin Bart’s in Saudi Arabia TABLE I. The best resolution ofthe Hb-Bart’s % distribution into one to four normal distributions No. of normal curves fitted
Means
Standard deviations
Percentage of srrbjects of each ‘curve’
One
3.8
3.1
I00
Two
2.8 8.8
1.7 3.8
83 I7
1.2
4.1 10.6
0.4 1.S 3.2
34 55
I .2
0.4
3.9 6.9
1.1
37 47 8 8
Three
Four
P
~0.0001 088
FIG 2. Distribution of 176 Hb-Bart’s levels divided into three and four groups. Above: the ‘best’ three normal curves are fitted to the Hb-Bart’s data and the chi-squared valuc for the goodness of fit is given. The probability of correct classification of any single Hb-Bart’s value within the three horizontal arrows is given by P. Below: the same data is given for the ‘best’ four normal curves.
226
M. E. Pembrey et aE
puted curve (or curves) was compared with the actual Hb-Bart’s data obtained using a twotailed chi-squared, goodness-of-fit test. The data on the curves fitted are given in Table I and Fig 2 and it is clear that a mixed series of three normal distributions approximates best to the cord-Hb-Bart’s levels we have obtained, with four curves being the next best fit. A population made up of either one or two iiornial distributions would gciicrate our observed data on less than one occasion in 10 000; whereas a population consisting of three normal distributions would do so apyroximately I in 50 times.
Siblings and Twins Selective family and twin studies are underway but tlic eight sibling pairs ciicouiitcrcd in the survey will be rcvicwed hcrc. Six sibling pairs were coiicordaiit for the level of Hb Bart’s in cord blood. One pair of dizygotic twins had 3.4% and 4.2% Hb Bart’s aiid two sibling pairs liad 4.7% aiid 3.2%, and 4.8% aiid 5.1% rcspcctively. In a further three sibling pairs all liad uiidetcctablc lcvels (less than 0.5%). Two sibling pairs were discordant. One sibling pair liad 2.6% and undetectable (lcss than 0.5%) Hb Bart’s aiid a pair of likc-scxcd twins liad 3.9% and undetectable levels in their respective cord bloods. These twins are probably iiioiiozygotic; they shared a placenta aiid at 2 ycars look so alike that nciglibours caiiiiot distinguish them. The parents differ in only three blood groups aiid so tlic finding of identical groups in the twins provides little cvidciice of moiiozygosity. Sex Ratio
For much of tlie survey tlie cord bloods were initially idciitified by tlic cmployec’s number, date of birth and the mother’s name, the latter bciiig important since an employee may liave more than one wife. The iiaiiie axid sex of tlie iiifaiit was gciicrally obtained retrospcctively from a ‘depciidents roster’ produced quarterly by the coniyaiiy. This roster iiicludes all dependents who liave at some time preseiitcd tlicmselvcs at one of the company’s clinics. In this respect the 109 infants ideiitificd from a rccciit roster are selected, possibly in favour of sons, since 60 ( 5 5 % ) were male. 52 infants wcrc kiiowii to have undetectable (lcss than 0.5%) Bart’s at birtli aid of these 28 (54%) werc daughters. However, iiiales prcdoiiiiiiated in 57 infants known to have Hb Bart’s in their cord blood, 36 (63%) being male. These sex differenccs are not sigiiificaiit at the 5% level (x2 = 2.53, Yates correction), aiid wlicre H b Bart’s was present, high, intcrmediate and low levels occurrcd in both sexes.
Globin-Chain Ratios and Hb Bart’s 7; (Table I1 and Fig 3) Satisfactory mcasurcineiit of the synthesis of a , P aiid y-globiii chains was obtained in 12 unselected cases. Seven samples Bad Hb Bart’s ranging from 2.9 to 13.3% whilst five had undetectable levels. Thus these cascs are fairly reyreseiitativc of the Sliiite Saudi newborn population, except that Bart’s levels between I and 2% are unreprcsented. Tlic a to iioii-a (p i - y ) chaiii-synthesis ratios ranged from 1.11 in an infant with no Hb Bart’s to 0.64 in tlie case with 13.3% Hb Bart’s (Table 11). Fig 3 deiiioiistratcs a significant iiegative correlation between the level of Hb Bart‘s and the a/noii-a-chain ratio (P = 0.002). Thus the amount of
Haemoglobin Bart‘s in Satrdi Arabia
227
TABLE 11. Haematological atid chain synthesis data on Shiite Saudi cord blood samples Shiite Saudi sanrple
Hb Bart’s
Red cell nrorphology* ajnon a
(%)
_
~
I 2
3
4 5 6 7 8 9
MCV
y/B+y
4.8 5.4 8.7 13.3 10.1
0.79 0.79 0.67 0.64 0.68 0.78 0.84
2.8
4.7 0
1.11
0
0.9
I0
0
I1
0
I2
0
I3
0
0.92 0.83 0.88 -
29 63 71 60 40 34 58 53 42 48 62 36
(P)
MCH (Pd
I04 113 128 91
25.5 28.5 28.5 I94
24.5 25.5 22.5
-
-
-
-
-
-
113
28.0
113
30.0
24.5 26.5
119 121 I21
I
-
Poikilocytosis Hypochrornia
~
++ ++ +++ +++ +++ ++ +++ ++ ++ +
21.5
-
-
32.0
27.0 28.5
32.5 30.5 33.2
115
-
Anisocytosis
_
++ ++
2s.s
~
+ I +
27.8
I
+++ ++ +++ ++ ++++ +++ +++ ++ + +++ +++ ++
--
++
* Abnormalities graded from normal ( 0 ) to markedly abnoirmal (+ +) ; the score given is the inean of that obtained by two indepcndent observers working ‘blind’ (see text). Hb Bart’s present appears to be a good indication of the degree of globin-chain imbalance, altliough tlie cc/non-cc-cliain ratio itself gives 110 information on whether there is an cc-chain deficiency or non-cc-chain excess. The a/iioii-a-chain ratios in cord samples from ‘seven’ full-term British neonates ranged from 0.84 to 1.14with a mean of 0.95; SDko.10. Hb Bart’s, Gcstational Age and 7-Chain Productioiz 111tlic same 12 unsclcctcd cases as abovc, no significant correlation was detected between the Hb Bart’s and gestational age based on dates or birth weight. Bard ct al (1970) and Bard (1973) have shown that the proportion of Hb F and Hb A 1.2 r = -079 P = 0.002
I
0
I
I
2
I
I
4
I
I
I
I
€I Hb Bart’s (%i
6
I
I
10
I
I
12
FIG 3, H b Bart’s % atid globin-chain ratios in Shiite Saudi cord bloods.
I
I
14
M. E. Penibrey et al
228
synthesized in cord-blood reticulocytcs correlates well with gestational age, and in the absence of Hb Bart’s or Hb H the proportion of y and p-chain synthesis would be exactly comparable. Despite the complicating factor of Hb Bart’s in some cases it is worth recording that the y-chain percentage of the total non-a-chain synthesis ( y % ) ranged from 71% in a newborn delivered prematurely at 32f weeks gestation to 29% in an infant born at 424 weeks by datcs. The 12 y”/o values fell broadly within the lower half of the range for % Hb F synthcsis published by Bard, and there was no significant correlation with the percentage of Hb Bart’s prcsent.
Hb Bart’s % and Red-Cell Indices and Morphology (Table I1 and Fig 4) The red-cell indices and other haematological data on 10 unselected Shiite Saudi cord bloods are given in Table 11. It is clear that those cases with Hb Bart’s tend to have a reduced mean cell haemoglobin (MCH) and abiiorinal red-cell morphology. The most striking correlation, shown in Fig 4, exists between the Hb Bart’s level and the rcductioii in MCH.
I4IIe
13
\
I
=-0.89
\ \
m
e
. i
019
MCH (pg)
FIG4. Hb Bart’s and rnean-corpuscular-haemoglobin (MCH) levels in Shiite Saudi cord bloods.
Erythrocyte morphology is more di$cult to assess quantitatively and in this study a numerical score for each blood smcar for anisocytosis, poikilocytosis and hypochromia was derived from scores obtained by two independent observcrs who assessed the smears ‘blind’. Of the three morphological characters considered only anisocytosis showed a statistically significant positive correlation with the level of Hb Bart’s ( r = 0.7, P
221.
Haernoglobin Bart’s in Saudi Arabia M. E. PEMBREY, D. J. WEATHERALL,* J. B. CLEGG,* c. BUNCHAND R. P. PERRINEf Departrncnt of Haenzafology, Uriiversity of Livcrpool, arid tArabiarz Americari Oil C O M I ~Dhahrari, U I Z ~ , Sartdi Arabia
SUMMARY. The liaemoglobin (Hb) pattcrns of 345 Shiite Saudi Arab cord bloods wcrc examined by alkaline starch-gel electropliorcsis. A fast-moving component, identified by structural analysis as Hb Bart’s, was found in 52% of cases, the highest incidence of this variant yet rccordcd. Tlic levels of Hb Bart’s ranged from 0.5 to 16% ofthe total haenioglobin. The relative rates of synthcsis of the a , P and ?-chains, measured by [3H]lcuci~ie incorporation, were estimated in IZiicwborii Arab infants. There was an excellent corrclatioii between the amount of Hb Bart’s and tlie a/nona-globin-chain production ratio. Furthermore tlierc was a significant correlatioii between tlic lcvel of Hb Bart’s and morpliological abnorinalitics of the red cells and the mean cell haemoglobiii (MCH). These findings indicate that elevated levels of Hb Bart’s in this population are due to tlie presence of a thalassacinia. The absciice of hydrops fetalis and the rarity of Hb-H disease despitc the intense iiibrccding in tliis population, points to an a-tlialassacmia genotype that is, in tcrins of phenotypic cxprcssion, ititcrnicdiatc bctwccii tlic lieterozygous state for a-tlialassacmia I and Hb-H discasc. A possible ~iiolecularbasis for this genotype is suggcsted. Iluring tlic pcriiiatal pcriod liaemoglobin A (Hb A, a2Pz) replaces liaenioglobin F (Hb F, tlie main liaenioglobin. Normally this process proceeds smoothly without any gross imbalance of globin-chain production, although trace aniounts of Hb Bart’s, a variant with molecular formula y 4 (Hunt & Lchmanii, I959), are found in most normal cord bloods, indicating a slight degrec of chain imbalance during the switching ovcr process. It is now well cstablislied that tlic presence of increased amounts of Hb Bart’s in the iicoiiatal period is the most sciisitivc iiidicator of the presence of an a-thalassaciiiia gene (Weatherall, 1963 ; Wasi c’t (I\, 1964, 1969; Lchniann, 1970). O n the other hand, Esaii (1970) has suggestcd that iiicreascd amounts of Hb Bart’s do not always indicate the presence o f a tlialassaeniia and may represent a developmental abiiorniality. That a tlialassaemia may occur in Saudi Arabia was suggestcd by reports of the occurrcncc of Hb-H discasc or its variants in Shiite Arabs by Weatherall et nl(1969a) and McNcil (1971). Since littlc is kiwwii of tlie incideiicc or clinical features of a tlialassaeniia in tlic Middlc East wc decided to iet up a survey of tlic liacnioglobins of iicwborii infants in Eastern Saudi Arabia. This d s carried out from 1967 until 1973 and the results indicate that this population has the higlielt iiicideiicc of clcvated levels of Hb Bart’s yet rccorded. The results of this
a 2 y 2 )as
* Present address: Nufield Department of Clinical Medicine, University of Oxford. Correspondence: Dr M. E. Pembrey, Department of Medicine, Guy’s Hospital Medical School, London, S.E.1. 221
222
M. E. Pembrey et
a1
survey and our further attempts to determine the significance of this unexpected finding form the basis of this paper.
SUBJECTS STUDIED A N D METHODS
Stibjects. The investigations were confined to Shiite Arabs from the Eastern Province of Saudi Arabia, who represent the ancient population from the Qatif and A1 Hasa Oases. All subjects were family dependents of the Arabian American Oil Company (ARAMCO) employees and were ascertained through the company’s own hospital or hospitals in the local town, A1 Khobar, when attending for routine maternal and child care. Over 50 ooo ‘dependents’, about 25% of whom are Shiites, come under the company’s health program and there is no indication that ‘dependents’ differ genetically from other Arabs in the Eastern Province. Cord bloods were taken routinely at the delivery of all Shiite babies. The majority of deliveries take place in hospital, the only factor selecting against this being the difficulty of transport from some outlying villages. Thc cord bloods were therefore a representative sample from Shiite Arabs living in the vicinity of Dhahran. The subsample on which hacmoglobin synthesis studies were carried out was similarly unselected. Infants were followed up at the company’s routine child-care clinics. In order to examine the levels of Hb Bart’s and the relativc rates of synthesis of the CL,/i’ aiid y chains of normal British Infants, cord-blood sainplcs were obtained from babies born at the Liverpool Maternity Hospital. Samples were collected into acid-citrate-dextrose (ACD) and thosc which wcre used for biosyiithetic experiments were studied within I h of collection. Collcctioir arid tmrzrport. 5-10 in1 of cord blood was takcii into ACD and one drop of 1% KCN added. The samples were kept at 4’C for 110 loiiger than I week before being flown to Liverpool as chilled cargo. The vast majority of samples arrived in excellent condition whilst the occasional partially haemolyscd sample was discarded. 10in1 of cord blood was collected into ACD within 14 h of delivery for iri vitro hacmoglobin synthesis studies and, after incubation, the red cells were stored at - 2o°C and transported on dry ice. Haemoglobiri arrulysis. Red-cell 1ysatcs were prepared by standard methods aiid aiialyscd by starch-gel electrophoresis using a tris-EDTA-borate system, pH 8.5, or a phosphatcbuffer system, pH 7.0 (Weatherall & Clcgg, 1972). For quantitation of tlie level of Hb Bart’s duplicate samples were elcctrophorescd 011 starch block using an 0.05 M phosphate buffer system, pH 7.0. The fractions containing Hb Bart’s and the remaining haemoglobins were eluted scyaratcly aiid tlieir relative proportions determined as described by Weatherall & Clegg (1972). In some cases haemoglobiiis were aiialysed by Amberlite IRC-50 chromatography under tlie conditions of Allen ct aI (1958). The samples were dialyscd against developer 2 for 24 11 at 4°C and then applied to 3 by 30 cin columiis of Ambcrlite IRC 50 equilibrated against the same developer at 4°C. The fiactions which were uiiretaiiicd at 4°C were eluted completely and then the column was warnicd to 25°C to obtain the remainder of the liaemoglobiii com-
Haerrioglobiii Bart’s in Saudi Arabia
2-23
pormits. Followiiig chromatography each fraction was coiicciitrated in vaciio and aiialyscd by starch-gel electrophoresis. In ordcr to determine the Structure of the fast-moving haemoglobins, presumed to be Hb Bart’s, they wcre isolated by starcli-block electrophoresis and coiivcrtcd to globin, aftcr wliicli they were digested witli trypsiii and fingerprintcd as described by Clcgg et a1 (1966). Iinerrtoglobin synthesis. Cord blood was collected into ACD and the cells wcre washed three times and incubated as described by Wcatlierall et a1 (1969b). After approximately 10 min 50 pCi of [3H]lcuciiic was added aiid the incubation allowed to contiiiuc for 2 11. The cells were then waslied three times and frozen. After thawing tlic lysatc was coiivertcd to ‘globin’ without any furtlier purification. The ‘globin’ was applied to a CM-cellulose column equilibrated witli 8 M urca/nicrcaptoctliaiiol, pH 6.8, as described by Clcgg et a/ (1966), and the CI, fl aiid y chains were separated. Tlie radioactivity incorporated into tlic cliains was detcriiiincd as previously described (Wcatlierall et al, 1969b). Tlie tubes coiitaiiiiiig tlie chains were then pooled and dialyscd against 0.5% formic acid aftcr which their specific activities wcre dcterinined.
RESULTS
Detlcrriiiizatioii of the Structure of the Rapidly-MigrafifigHaeriioglobiti in Saridi Cord Bloods Early on in this study it became obvious that a liigli proportion of the cord bloods coiltained a rapidly-migrating liaemoglobin. This fraction migrated in tlic position of Hb Bart’s 011 starch-gel or starch-block electrophoresis, pH 8.6 and 6.5. It was isolated by starch-block clectroplioresis from two samples containing 5% and 1% rcspectively. Thcsc components wcre clicckcd for purity by starch-gel clcctrophorcsis and then coiivcrted to globin and finger-printed. Thc rcsultaiit pcptidc imps showed y-chain pcptidcs only, iiidicatiiig that the rapidly-migrating hacmoglobiii is liaciiioglobiii Bart’s. of Hb Unrt’s by Starch-Gel Elcctroyhoresis As a prcliiiiinary to this survey a scrics of cxpcriiiiciits wcrc pcrforiiicd to coiifirin the adcquacy of the scrcciiiiig technique for tlic prcsciicc of Hb Bart’s in cord-blood aiid to asccrtaiii the lowest level of H b Bart’s which can bc discriminated by iiictliods which arc currently available. Scrial dilutions of a cord blood liaeiiiolysatc coiitainiiig about 5% Hb Bart’s wcrc iiiadc with a normal adult liacniolysatc of the same coiiccntratioii and the rcsulting niixturcs wcrc analyscd by starch-gcl clcctrophoresis at pH 8.5 and 6.5, as describcd in the Methods section. The gcls wcrc staincd with a bciizidinc stain (Wcathcrall & Clcgg, 1972) aiid tlic dilution at which Hb Bart’s was clearly visible was dctcrinincd. It was found that at pH 8.6 Hb Bart’s was easily distinguishable wlicii prcsciit as 0.5% of tlic total liacnioglobin; similar results were obtaiiicd on tlic pH 6.5 gel. Tlius in tlie screening program all cord bloods wcre studied at a liaciiiolysatc coiiccntration of 8-10 g/dl, using starch-gcl clcctrophoresis witli a tris-EDTA-borate system, pH 8.6. Using this system Hb Bart’s was not see11 in 50 coiisccutivc British cord-blood samples. A furtlicr scries of experiments were pcrforiiied to dctcrniiric tlie accuracy of tlic determination of relatively small amounts of Hb Bart’s by starch-block electrophoresis. Ten
The Der?iotistratioiz
M. E. Pembrey et a1
224
determinations were carried out on a cord-blood sample containing approximately Bart’s. The values raiigcd from 0.9 to 1.4% (mean 1.15%; SD+o.27%).
I
.OXHb
Incidence and Levels of Hb Bart’s 345 cord bloods were run on starch gel and 182 samples had a visible Hb-Bart’s band making an overall iticideiice of 52.6%. In the first 325 samples 176 had Hb Bart’s visible on starch gel and in each case the level of Hb Bart’s was estimated in duplicate by starch-block electrophoresis; these arc suiniiiarizcd in Fig I . 169 samples (52%) had more than 0.5% Hb Bart’s. 149 samples (45.8%) had no visible Hb-Bart’s band and caii be assumed to have contained less than 0.5% Hb Bart’s, and seven cases (2.2%) had a measured level of less than 0.5%, briiigiiig tlic total with less than 0.5% Hb Bart’s to 156 (48%). 24
-
21 18
-
15 -
,“
129-
63-
0Hb Bart’s (%)
FIG I. Distribution of Hb Bart’s in 176 Shiite Saudi ncwboriis. Total survey = Incidence H b Bart’s = 54.2%.
325
cord bloods
Distribiition o f H b Bart’s in 176 Cord Bloods The distribution of the Hb-Bart’s levels foulid in the cord-blood samples shown in Fig I is clearly skewed to the right, with a discreet upper tail comprising 14 cases with levels between 10 and 17%. O n inspection the main body ofthe distribution appears to be bimodal, or possibly even trimodal, with an aiitiniode at 2.2% and possibly one at 6.0%. Attempts were made to fit one to four ‘normal curves’ to the whole distribution employing the method of maxiiiiuni likelihood using the MINUIT program of F. James and M. Roos (Cern computer program library). Within the constraints imposed by requiring one, two, three or four normal curves, the ‘best’ curves were fitted and the mean, standard deviatioii slid proportion of the total for each curve were estimated. The reliability of these estimates was somewhat reduced by the relatively small total number of observations. When more than one curve was concerned the program also provided the probability of correct classification of any given Hb-Bart’s level into one or other of the groups. In each case the corn-
Haemoglobin Bart’s in Saudi Arabia TABLE I. The best resolution ofthe Hb-Bart’s % distribution into one to four normal distributions No. of normal curves fitted
Means
Standard deviations
Percentage of srrbjects of each ‘curve’
One
3.8
3.1
I00
Two
2.8 8.8
1.7 3.8
83 I7
1.2
4.1 10.6
0.4 1.S 3.2
34 55
I .2
0.4
3.9 6.9
1.1
37 47 8 8
Three
Four
P
~0.0001 088
FIG 2. Distribution of 176 Hb-Bart’s levels divided into three and four groups. Above: the ‘best’ three normal curves are fitted to the Hb-Bart’s data and the chi-squared valuc for the goodness of fit is given. The probability of correct classification of any single Hb-Bart’s value within the three horizontal arrows is given by P. Below: the same data is given for the ‘best’ four normal curves.
226
M. E. Pembrey et aE
puted curve (or curves) was compared with the actual Hb-Bart’s data obtained using a twotailed chi-squared, goodness-of-fit test. The data on the curves fitted are given in Table I and Fig 2 and it is clear that a mixed series of three normal distributions approximates best to the cord-Hb-Bart’s levels we have obtained, with four curves being the next best fit. A population made up of either one or two iiornial distributions would gciicrate our observed data on less than one occasion in 10 000; whereas a population consisting of three normal distributions would do so apyroximately I in 50 times.
Siblings and Twins Selective family and twin studies are underway but tlic eight sibling pairs ciicouiitcrcd in the survey will be rcvicwed hcrc. Six sibling pairs were coiicordaiit for the level of Hb Bart’s in cord blood. One pair of dizygotic twins had 3.4% and 4.2% Hb Bart’s aiid two sibling pairs liad 4.7% aiid 3.2%, and 4.8% aiid 5.1% rcspcctively. In a further three sibling pairs all liad uiidetcctablc lcvels (less than 0.5%). Two sibling pairs were discordant. One sibling pair liad 2.6% and undetectable (lcss than 0.5%) Hb Bart’s aiid a pair of likc-scxcd twins liad 3.9% and undetectable levels in their respective cord bloods. These twins are probably iiioiiozygotic; they shared a placenta aiid at 2 ycars look so alike that nciglibours caiiiiot distinguish them. The parents differ in only three blood groups aiid so tlic finding of identical groups in the twins provides little cvidciice of moiiozygosity. Sex Ratio
For much of tlie survey tlie cord bloods were initially idciitified by tlic cmployec’s number, date of birth and the mother’s name, the latter bciiig important since an employee may liave more than one wife. The iiaiiie axid sex of tlie iiifaiit was gciicrally obtained retrospcctively from a ‘depciidents roster’ produced quarterly by the coniyaiiy. This roster iiicludes all dependents who liave at some time preseiitcd tlicmselvcs at one of the company’s clinics. In this respect the 109 infants ideiitificd from a rccciit roster are selected, possibly in favour of sons, since 60 ( 5 5 % ) were male. 52 infants wcrc kiiowii to have undetectable (lcss than 0.5%) Bart’s at birtli aid of these 28 (54%) werc daughters. However, iiiales prcdoiiiiiiated in 57 infants known to have Hb Bart’s in their cord blood, 36 (63%) being male. These sex differenccs are not sigiiificaiit at the 5% level (x2 = 2.53, Yates correction), aiid wlicre H b Bart’s was present, high, intcrmediate and low levels occurrcd in both sexes.
Globin-Chain Ratios and Hb Bart’s 7; (Table I1 and Fig 3) Satisfactory mcasurcineiit of the synthesis of a , P aiid y-globiii chains was obtained in 12 unselected cases. Seven samples Bad Hb Bart’s ranging from 2.9 to 13.3% whilst five had undetectable levels. Thus these cascs are fairly reyreseiitativc of the Sliiite Saudi newborn population, except that Bart’s levels between I and 2% are unreprcsented. Tlic a to iioii-a (p i - y ) chaiii-synthesis ratios ranged from 1.11 in an infant with no Hb Bart’s to 0.64 in tlie case with 13.3% Hb Bart’s (Table 11). Fig 3 deiiioiistratcs a significant iiegative correlation between the level of Hb Bart‘s and the a/noii-a-chain ratio (P = 0.002). Thus the amount of
Haemoglobin Bart‘s in Satrdi Arabia
227
TABLE 11. Haematological atid chain synthesis data on Shiite Saudi cord blood samples Shiite Saudi sanrple
Hb Bart’s
Red cell nrorphology* ajnon a
(%)
_
~
I 2
3
4 5 6 7 8 9
MCV
y/B+y
4.8 5.4 8.7 13.3 10.1
0.79 0.79 0.67 0.64 0.68 0.78 0.84
2.8
4.7 0
1.11
0
0.9
I0
0
I1
0
I2
0
I3
0
0.92 0.83 0.88 -
29 63 71 60 40 34 58 53 42 48 62 36
(P)
MCH (Pd
I04 113 128 91
25.5 28.5 28.5 I94
24.5 25.5 22.5
-
-
-
-
-
-
113
28.0
113
30.0
24.5 26.5
119 121 I21
I
-
Poikilocytosis Hypochrornia
~
++ ++ +++ +++ +++ ++ +++ ++ ++ +
21.5
-
-
32.0
27.0 28.5
32.5 30.5 33.2
115
-
Anisocytosis
_
++ ++
2s.s
~
+ I +
27.8
I
+++ ++ +++ ++ ++++ +++ +++ ++ + +++ +++ ++
--
++
* Abnormalities graded from normal ( 0 ) to markedly abnoirmal (+ +) ; the score given is the inean of that obtained by two indepcndent observers working ‘blind’ (see text). Hb Bart’s present appears to be a good indication of the degree of globin-chain imbalance, altliough tlie cc/non-cc-cliain ratio itself gives 110 information on whether there is an cc-chain deficiency or non-cc-chain excess. The a/iioii-a-chain ratios in cord samples from ‘seven’ full-term British neonates ranged from 0.84 to 1.14with a mean of 0.95; SDko.10. Hb Bart’s, Gcstational Age and 7-Chain Productioiz 111tlic same 12 unsclcctcd cases as abovc, no significant correlation was detected between the Hb Bart’s and gestational age based on dates or birth weight. Bard ct al (1970) and Bard (1973) have shown that the proportion of Hb F and Hb A 1.2 r = -079 P = 0.002
I
0
I
I
2
I
I
4
I
I
I
I
€I Hb Bart’s (%i
6
I
I
10
I
I
12
FIG 3, H b Bart’s % atid globin-chain ratios in Shiite Saudi cord bloods.
I
I
14
M. E. Penibrey et al
228
synthesized in cord-blood reticulocytcs correlates well with gestational age, and in the absence of Hb Bart’s or Hb H the proportion of y and p-chain synthesis would be exactly comparable. Despite the complicating factor of Hb Bart’s in some cases it is worth recording that the y-chain percentage of the total non-a-chain synthesis ( y % ) ranged from 71% in a newborn delivered prematurely at 32f weeks gestation to 29% in an infant born at 424 weeks by datcs. The 12 y”/o values fell broadly within the lower half of the range for % Hb F synthcsis published by Bard, and there was no significant correlation with the percentage of Hb Bart’s prcsent.
Hb Bart’s % and Red-Cell Indices and Morphology (Table I1 and Fig 4) The red-cell indices and other haematological data on 10 unselected Shiite Saudi cord bloods are given in Table 11. It is clear that those cases with Hb Bart’s tend to have a reduced mean cell haemoglobin (MCH) and abiiorinal red-cell morphology. The most striking correlation, shown in Fig 4, exists between the Hb Bart’s level and the rcductioii in MCH.
I4IIe
13
\
I
=-0.89
\ \
m
e
. i
019
MCH (pg)
FIG4. Hb Bart’s and rnean-corpuscular-haemoglobin (MCH) levels in Shiite Saudi cord bloods.
Erythrocyte morphology is more di$cult to assess quantitatively and in this study a numerical score for each blood smcar for anisocytosis, poikilocytosis and hypochromia was derived from scores obtained by two independent observcrs who assessed the smears ‘blind’. Of the three morphological characters considered only anisocytosis showed a statistically significant positive correlation with the level of Hb Bart’s ( r = 0.7, P
