1393

bacteria by livestock must be the rule, and where the useof antibiotics as feeding additives must exert a greatselective influence on the gut flora, their incidence amongthe normal faecal bacteria of animals can be very high(Smith and Halls 1966). As long ago as 1957 Smith andCrabb observed the appearance of tetracycline-resistantE. coli in pigs when tetracycline had been used as a feedingadditive, and more recently Mitsuhashi et al. (1967)isolated multi-resistant E. coli from all of 151 pigs on onefarm in Japan on which feeds supplemented with tetra-cyclines had been used. This situation is also reflected in avery high incidence of resistant strains among peoplelooking after farm animals (Smith and Halls 1966) andmust represent a formidable source of human infectionwith drug-resistant E. coli able to transfer resistance intohuman enterobacteria (Anderson 1967).

Lederberg and Tatum (1953) reported that, of nearly2000 independent isolations of E. coli from various sources,more than 50 strains were able to transfer chromosomalmarkers to E. coli K12. It is unlikely that any of thistransfer was mediated by transferable colicine factors;killing of a sensitive K12 recipient by the colicine pro-duced by a colicinogenic donor can mask even quite highfrequencies of resistance transfer. It is also reasonable toassume that transferable drug resistance did not figuresignificantly in this series, since streptomycin resistancecould be used as a selective marker in the recipient strain.The number of strains cross-fertile with K12 thereforealmost certainly gives a measure of the incidence of strainsin that series carrying simple transfer factors. Theincidence reported (rather more than 2-5%) agrees

closely with the incidence (3%) demonstrated in the

present series by using mobilisation of resistance markersin a triple cross. However, as fi + factors are much moreefficient than fi- factors in promoting chromosomaltransfer, the series of Lederberg and Tatum was probablybiased in favour of fi+ factors.Among their recent proposals for a uniform nomen-

clature in bacterial genetics Demerec et al. (1966) suggestthat strains carrying autonomous transfer factors should allbe termed F +. As this term has for many years beenused to designate bacterial strains carrying the F factorof E. coli K12 (Hayes 1953 a, b) it seems unsuitable togeneralise it in the way suggested. Among the strainsdescribed here there are at least three transfer factorsdifferent from F and from each other. As more suchfactors are described, and especially when means of

classifying them are further developed, adequate differen-tial labels for them will have to be devised.

It is a pleasure to thank Dr. E. S. Anderson for his helpful interestin this work, and Miss Mary Woodyard for valuable technical assis-tance.

REFERENCES

Anderson, E. S. (1965a) Br. med. J. ii, 1289.- (1965b) Nature, Lond. 208, 1016.- (1967) Annls Inst. Pasteur, Paris, 112, 547.- (1968) A. Rev. Microbiol. (in the press).— Lewis, M. J. (1965a) Nature, Lond. 206, 549.— — (1965b) ibid. 208, 843.

Cooke, E. M. (1968) J. Path. Bact. 95, 101.Cowan, S. T., Steel, K. J. (1965) Manual for the Identification of Medical

Bacteria; p. 114. London.Demerec, M., Adelberg, E. A., Clark, A. J., Hartman, P. E. (1966) Genetics,

54, 61.Hayes, W. (1953a) J. gen. Microbiol. 8, 72.- (1953b) Cold Spr. Harb. Symp. quant. Biol. 18, 75.

Lederberg, J., Tatum, E. L. (1953) Science, N.Y. 118, 169.Lewis, M. J. (1967) Lancet, ii, 953.Mitsuhashi, S., Hashimoto, H., Suzuki, K. (1967) J. Bact. 94, 1166.Smith, H. W., Crabb, W. E. (1957) Vet. Rec. 69, 24.- Halls, S. (1966) Br. med. J. i, 266.

Tarr, H. A. (1958) Mon. Bull. Minist. Hlth, 17, 64.Watanabe, T., Nishida, H., Ogata C., Sato S. (1964) J. Bact. 88, 716.

SEVERE HYDROPS FŒTALIS TREATED WITH

PERITONEAL DIALYSIS AND

POSITIVE-PRESSURE VENTILATION

ERLING NATHANM.D. Aarhus, Denmark

From the Departments of Pœdiatrics and Obstetrics A and B,Rigshospitalet, Copenhagen, Denmark

Summary Survival from hydrops fœtalis is extremelyrare, though some success has been

reported with venesections, laparocentesis, positive-pressure ventilation, antibiotics, and digitalis. A methodof treating hydropic infants is described. It consists of

positive-pressure ventilation and removal of the œdemaby peritoneal dialysis (which has not previously beendescribed in the treatment of this condition). Four

patients all severely rhesus immunised were treated. Twosurvived without any complications. Two died of

respiratory problems at a time where the dialysis hadbeen completed and the œdema had been removed.These patients were very premature, born 9 and 16 weeksbefore the calculated date of birth.

Introduction

SURVIVAL from hydrops foetalis is uncommon. Up to1966 the total number of published cases was thirteen.David et al. (1966) treated fifty-one out of ninety-fivecases, and eight survived; but they gave no details aboutthe infants’ weight or the degree of the oedema. Cook

(1956) and Dyggve (1960) each described a survivor whohad had oedema and ascites at birth. In the case describedby Shapiron and Cohen (1939) and in one of the twocases described by Jacobi et al. (1946) oedema developedafter the birth. It might then be questioned if all thepublished cases really had hydrops foetalis. In additionother cases seem to have survived, some after treatmentby intrauterine transfusions.On the basis of good experience with the removal of

oedema in adults by means of peritoneal dialysis I havetried this treatment on hydropic infants. This does notseem to have been described previously.

TechniqueWhen possible the patient was delivered by planned caesarean

operation. Immediately after the birth the baby was intubatedby nasotracheal tube. Exchange transfusion through an umbili-cal-vein catheter was started at once. Continual positive-pressure ventilation was instituted. The problems of respiratortreatment are reviewed by Cooke et al. (1967).As soon as the haemoglobin was brought up to about 100%

laparocentesis was carried out, either with a syringe or by aperitoneal dialysis catheter. After this, exchange transfusionwas completed.

Peritoneal dialysis was carried out along the lines outlined byBoen (1964). The catheter was inserted to the left of themedian line in order not to injure the umbilical arteries andurachus, and just distal to the umbilicus; it was directedtowards the sacral region so as not to injure the enlarged spleen.Disposable catheters (’Trocath’, McGaw Laboratory Inc.,Milledgeville, Georgia, U.S.A.) were shortened before use.Other peritoneal dialysis catheters could, of course, be used.The solutions used for the first part of the dialysis contained

sodium in concentration of 135 meq. per litre, potassium 2 meq.per litre, and glucose 4-25 g. per 100 ml. This removed extra-cellular and intracellular water and prevented hyperkaloemiafrom the haemolysis and from the blood used for the exchangetransfusions. Once most of the oedema were removed thedialysis was continued with solutions containing sodium 140meq. per litre, potassium 4 meq. per litre, and glucose 1-5 g.

1394

per 100 ml. Water extraction continued, but at a less violentrate, and the serum-electrolytes were normalised. The con-centrations (meq. per litre) of other electrolytes in the solutionswere: magnesium 1-5, calcium 45, lactate 45. The rest of theanions consisted of chloride. Ampicillin 25 mg. per litre andheparin 500 i.u. per litre were added to the solutions.The amount of solution instilled varied between 60 and

120 ml., depending on the amount of ascites primarilydrained. The numbers of instillations varied from two to three

per hour, depending on how the patients stood up to thedialysis. The dialysis solutions did not stay longer in theperitoneum than it took to flow in and out. During the dialysis,human albumin 20 g. per 100 ml. was supplied 5 ml. per kg.body-weight, slowly intravenously, repeated each 10th hour.

10 hours after delivery the second exchange transfusion wasdone, independently of the serum-bilirubin. The dialysis hadto be stopped during this procedure. All the patients weretreated with frusemide 10 mg. intravenously every 8th hour,and broad-spectrum antibiotics were also given.

Case-reportsCase 1The mother was 26 years old and in good health. Blood-

group was A Rh negative. Two previous pregnancies resultedin a normal delivery and a miscarriage between the 2nd and3rd months. The third pregnancy was at first complicated byparoxysmal tachycardia, later on oedema and rapidly increas-ing rhesus-antibody titres developed. Therefore amniocentesiswas done. The bilirubin level in the amniotic fluid was 8.1

ILmole per litre (Brodersen and Vind 1963). Amniographyindicated probable foetal hydrops. The infant was deliveredby a caesarean operation, approximately 4 weeks before theexpected date of birth. He weighed 3400 g. He was asphyxial,heart-rate below 100 per minute; he was cyanotic with lack ofreflexes and without spontaneous respiration. The abdominalwall was distended by ascites, the face was severely swollen,the penis difficult to see because of oedema of the scrotum andin addition there were severe widespread oedema (fig. 1).The child was treated as outlined above, but the dialysis

was first started 14 hours after birth. The patient was treatedwith digitalis. By the first exchange transfusion a deficit of50 ml. blood was established. By laparocentesis 260 ml. asciteswas primarily removed. By dialysis 480 ml. water was extracted.Frusemide treatment was stopped during the 5th day of life,when the patient reached his minimum weight of 2190 g.

Fig. 1-Case 1: in incubator shortly after the birth.The structure in the middle of the picture is a plug in the incubator,

The structure over his nose and right part of his forehead is th(tubing to and from the respirator.

Fig. 2-Case t: 5 days old.

after having lost 1210 g. or approximately 40% of his weightat birth (fig. 2).During dialysis attacks of heart arrest took place. They

were successfully treated by external heart massage. Electro-cardiography showed atrioventricular block and digitalis effect.The cause was probably that the amount of digitalis was toolarge in proportion to the rapid decrease in weight in connectionwith hypokalaemia.

After this bilateral pneumonia developed, Pseudomonas wasfound in the trachea. He was discharged at age 6 weeks. Twobig inguinal hernias and a large umbilical hernia remained.Apart from this the patient seemed to be normal. At 2 monthshe was readmitted because of anaemia and treated with onetransfusion. At 3 months he was operated on for the hernias.At 1 year old he seemed to have developed normally; electro-encephalogram, chest X-ray, and leucocyte and erythrocytecounts were normal.

Case 2The mother had been previously operated on for struma,

but apart from this she was healthy. Her blood-group wasA Rh negative. Two previous pregnancies resulted respec-tively in a normal delivery and a miscarriage in the 3rd month.The third pregnancy was complicated by a haemorrhage lasting1 day in the 3rd month. After 5 months’ pregnancy there werefound to be rapidly increasing antibody titres for anti-D andanti-C. Amniography showed hydrops fcetalis. A 60 ml.

erythrocyte suspension was given by intrauterine foetal peri-toneal transfusion. This was 16 weeks before the expecteddate of birth and a caesarean operation was not performed.However, the delivery commenced spontaneously 4 days later,and the patient was born in an occipital-anterior position.Weight was 1700 g. The infant was asphyxial with slowheart-rate, cyanosis, lack of reflexes, and without spontaneousrespiration. The infant was extremely hydropic. The thoraxwas almost hidden by distended skin caused by ascites andoedema. The patient was intubated by nasotracheal tube andtreated as described. 430 ml. ascites was removed. 300 ml.water was extracted by dialysis. A deficit of 55 ml. blood wasestablished by exchange transfusions. 2 days after birth hisweight was 900 g. ; this being a loss in weight of 800 g. orapproximately 47% of the weight at birth. The patient wasvery ill and had to be treated with steroids. He had spon-taneous movements, but died at 21 /, days old during an exchangetransfusion. There was at that time no oedema and the dialysishad been finished 36 hours previously. Necropsy showedatelectasis of the lungs and a 2-3 mm. large remainder of ahaemorrhage in the left caudate nucleus of the brain.Case 3The mother was healthy. Her blood-group was A Rh

negative. The one previous pregnancy had been normal.There had been no complications in the second pregnancy,but the mother had neglected to attend prenatal examinations.Delivery was spontaneous in occipito-anterior position 1 weekbefore the estimated date of birth. (Edema at birth was not

noticed, and it is doubtful if the infant actually had hydropsfoetalis, but during the 15 hours after birth the child developedsevere oedema. Birth-weight 3100 g. The mother had highantibody titres.

At birth the central venous pressure was high. During thefirst 12 hours a total of 100 ml. blood was removed by exchangetransfusions and venesections. At 15 hours old treatment byrespirator was started because of pulmonary cedema. At thattime there was also peripheral oedema. Treatment with

peritoneal dialysis and frusemide as described previously wasstarted. 110 ml. water was extracted by the dialysis, and afew hours later the nasotracheal tube could be removed.The child was examined at 4 months of age and seemed to

be healthy and had developed normally.Case 4The mother of this patient was 19 years old and in good

health. Her blood-group was 0 Rh negative. Two previouspregnancies resulted in a stillbirth and a living child (treatedwith three exchange transfusions because of rhesus immunisa-

1395

Fig. 3-Case 4: shortly after the birth.

tion). The third pregnancy was complicated by mild oedemaand 14 weeks before the expected date of delivery there werefound to be rhesus antibodies.

9 weeks before the expected date she was admitted havinguterine contractions. Hydrops fcetalis was suspected and acxsarean operation was performed. Weight at birth was2600 g. The child was asphyxial, heart-rate below 60 perminute, no spontaneous movements, lack of muscle tone andreflexes, and cyanosis. The child had extreme cedema and theabdominal wall was distended by ascites (fig. 3).

215 ml. ascites was removed by laparocentesis, and sub-sequently by implacement of a peritoneal dialysis catheter afurther 100 ml. By dialysis 640 ml. water was extracted (esti-mated on the basis of weight differences). By exchangetransfusions a total deficit of 95 ml. blood was established.The weight after dialysis (36 hours after the birth) was 1550 g.-i.e., a loss in weight of 1050 g. or approximately 40% of theweight at birth. 11 hours later the patient became ill, cyanotic,and had increased heart-rate. A thoracentesis was done be-cause of suspected hydrothorax, but no fluid was found. Thepatient died aged 2 days.Necropsy revealed double-sided pneumothorax, hypoplasia,

and atelectasis of the lungs. The weight of the lungs wasonly a third of normal. It is uncertain if the reason for thiswas the pneumothorax, a congenital hypoplasia, or the

hydropic condition.

THE IMPORTANT FINDINGS IN THE FOUR PATIENTS

Dead.

The findings of the four patients are compared in thetable.

Discussion

The xtiology of the development of severe oedemaduring foetal life is still not clear. The following explana-tions still find support: low serum-protein (Jacobi et al.1946, Richdorf and Cady 1948), liver damage (Silveret al. 1958), capillary damage (Zollinger 1946), heartdecompensation with backward failure (O’Neille and

Gordon 1959), or problems concerning the water trans-port through the placental membrane (Minkowski et al.1957, Drogendijk 1963). This last possibility, which inall cases must play a role, depends on one or more of theprevious mentioned factors. The good results with waterextraction by dialysis indicate that the xtiology is theaction of one or more factors chiefly on the water

transport through the placenta.In the three cases in which serum-albumin was

measured at birth, the level was low. The total amountof albumin may, however, have been rather high. Theserum-protein concentration increased rapidly under thetreatment (see table) and it is not clear whether the lowvalue at the birth was a result of, or a cause of, the waterretention.David et al. (1966) are probably the only group to have

had any success in treating hydropic infants actively.Their treatment consisted of laparocentesis, venesections,prolonged exchange transfusions with increasing deficits,positive-pressure ventilation, digitalis, and steroids. Themortality-rate in the long term was more than 90%, butthe severity of the oedema was not described.

Dyggve’s (1960) patient was treated by laparocentesis,but apart from this the oedema decreased spontaneously.The aetiology in his case was probably paroxysmaltachycardia. Cook’s (1956) patient was rhesus immunisedand was treated with two exchange transfusions. Deficitswere not established and laparocentesis was not done.

Intrauterine foetal peritoneal transfusions cannot,according to Liley (1966), be used when the foetus is

hydropic. This is partly in accordance with my experiencein case 2-although the hoemoglobin concentration wasperhaps higher than it would have been without theintrauterine transfusion.

Peritoneal dialysis, as done in these four patients, doesnot differ from the method used in adults, apart from thesize of the catheters, the amount of solutions, and theanatomical differences. But it is necessary, as in adults, tocooperate with the dialysis expert when using this treatment.The electrolyte concentration in the dialysing solutions

was chosen, because of the need to have a concentrationnot very different from that of normal serum. Neverthe-less there should be a potassium-lowering effect. Further-more hypervolwmia should be avoided, so the sodiumconcentration should be below normal. Glucose, whichis in the main responsible for the water extraction, shouldnot be used in concentrations higher than 4-4-5g. per100 ml. With higher concentrations the capillaries arealmost completely emptied and there is a great risk ofvascular collapse. With lower concentrations the extrac-tion is too slow.

The solutions used for the peritoneal dialysis are allstandard solutions in the Rigshospitalet, Denmark. In thefirst part of the dialysis two solutions were mixed; onesolution with 140 meq. sodium per litre, 1-5% glucose,and no potassium, the other with 4 meq. potassium perlitre, 130 meq. sodium per litre, and 7% glucose. Inthe second part of the dialysis the solution described inTechnique was used. All the solutions contained mag-nesium, calcium, lactate and, chloride.

Small variations in the concentrations of the electrolytesand glucose cannot be avoided, but this is not critical. Asodium concentration below 135 meq. per litre could beuseful throughout the entire dialysis. In all the patientsthe serum concentration tended to rise too much towardthe end of the dialysis.

1396

As the table shows, water extraction was very effective.No directions can be given for the duration of the dialysis,neither with the high glucose concentration nor with theensuing more isotonic solution. In each case the durationmust be based on a clinical estimation of the amount ofthe oedema present as the dialysis proceeds.No complications were seen, but they can be expected

- infections, intraperitoneal haemorrhages, leakages wherethe catheter passes through the abdominal wall, anddehydration of the patients are possible, and difficultymay be experienced in measuring the amounts of solutionsflowing in and out.Albumin infusions were given to replace protein

removed from the peritoneum by the dialysis. Furtherdetails concerning peritoneal dialysis are referred to byBoen (1964).The low complication risk is demonstrated in the two

very premature infants who withstood the very rapidwater extraction without any problems.The indications for this treatment could be rather wide,

perhaps as wide as was the case in patient 3. More

experience with the treatment in hydropic infants isneeded before limitations can be prescribed. I feel that

peritoneal dialysis should be tried no matter how bad thecondition of the hydropic child.

Frusemide in very high doses has been used in thetreatment of anuric adults, and increasing doses in new-born infants has not given any toxic reactions. No toxicreactions were seen in these four hydropic infants.

Unfortunately, measuring the diuresis failed, but the

impression was that the effect was good.In all patients a secondary exchange transfusion was

carried out not later than 10 hours after birth independentlyof a low serum-bilirubin. This was done because thebilirubin-binding capacity of the serum from the first

patient was very low. The serum-bilirubin would onlyhave confused an estimate of the timing of the secondexchange transfusion.

Peritoneal dialysis has not previously been describedas a method of treatment for hydrops foetalis. Preventionof the disease naturally has to be considered first, andanti-Rh gamma-globulin could be given to the motherafter the delivery of Rh-positive foetus (Freda et al. 1964).When severe immunisation is found intrauterine trans-fusion could be given (Charles et al. 1966, Liley 1966).When hydrops foetalis is present the technique of peritonealdialysis that I have described is effective. The method issimple, but demands experience with peritoneal dialysis.It should only be done with the help of an expert. Further-more, operators must be able to cope with the very greatproblems presented by respirator treatment of intubatednewborn infants. This is necessary because of the presentpulmonary oedema. Close cooperation with the obstetriciansis essential so that the treatment of affected infants can

begin immediately; and haemoglobin concentration mustbe increased immediately, and dialysis started very shortlyafterwards to remove oedema of brain and lungs.The method does not remove bilirubin. In case 1 only

a 9-29 mg. bilirubin was removed by dialysis: this accordswith the findings of other workers (Devantier andHobolth 1967).The number of hydropic infants treated by peritoneal

dialysis is small. However, these were the only four suchpatients born at the Rigshospitalet, Copenhagen, betweenJanuary, 1967, when this method was first applied, andApril, 1968. Two patients survived, but it is perhaps

doubtful if patient 3 had hydrops fcetalis. The two whodied were very premature, being born 9 and 16 weeksbefore the expected date, and with minimum weights of1600 and 900 g. Death was not caused by complicationsof the dialysis or by their water balance, but was due torespirator and exchange transfusion problems. Most

probably, dialysis treatment in patient 3 shortened therespirator treatment conisderably.

REFERENCES

Boen, S. T. (1964) Peritoneal Dialysis in Clinical Medicine. Springfield,Illinois.

Brodersen, O., Vind, I. (1963) Scand J. clin. Lab. Invest. 3, 15.Charles, A. G., Alpern, W. M., Friedman, E. A. (1966) Obstet. Gynec. 28,

182.

Cook, J. (1956) Archs Dis. Child. 31, 92.Cooke, R., Lunding, M., Lomholt, N. F., Yssing, M., Zachau-Christiansen

B., Friis-Hansen, B. (1967) Acta pœdiat. scand. 56, 498.David, G., Larroche, V. C., Maigret, P., Lacomme, M. (1966) Bibl. Gynœc.

38, 114.Devantier, M., Hobolth, N. (1967) Ugeskr. Lœg. 26, 871.Drogendijk, A. C. (1963) Ned. Tijdschr Verlosk. Gynœc. 63, 347.Dyggve, H. (1960). Acta pœdiat. scand. 49, 437.Freda, V. J., Gorman, J. G., Pollack, W. (1964) Transfusion, 4, 26.Jacobi, M., Litvak, A., Gruber, S. (1946) J. Pediat. 29, 177.Liley, A. W. (1966) Bibl. Gynœc. 38, 146.Minkowski, A., Sainte-Anne Dargassies, S., Armaganidou, D., Bourgonnier

L., Herbert-Jouas, C., Pellerin, D. (1957) Étud. neonat. 6, 25.O’Neill, E., Gordon, R. R. (1959) Archs Dis. Childh. 34, 174.Richdorf, L. F., Cady, L. H. (1948) J-Lancet, 68, 188.Shapiron, L. M., Cohen, P. (1939) Am. J. Dis. Child. 58, 1050.Silver, H. K., Huffman, P. J., Nakashima, I. I. (1958) ibid. 96, 268.Zollinger, H. U. (1946) Helvet. pœdiat. Acta. suppl. II, p. 127.

NEONATAL LYMPHOCYTE REACTIVITY

AS AN INDICATOR OF INTRAUTERINE

BACTERIAL CONTACT

JEROME I. BRODYM.D. Jefferson

ASSOCIATE PROFESSOR

OF MEDICINE

FRANK A. OSKIM.D. PennsylvaniaASSISTANT PROFESSOR

OF PEDIATRICS

EDWARD E. WALLACHM.D. Cornell

ASSISTANT PROFESSOR OF OBSTETRICS AND GYNECOLOGY

From the Department of Medicine, the Graduate Hospital ofthe University of Pennsylvania, the Departments of Pediatricsand Obstetrics and Gynecology, the Hospital of the University ofPennsylvania, and the University of Pennsylvania School of

Medicine, Philadelphia, Pennsylvania.

Summary The in-vitro response of neonatal lympho-cytes to a microorganism was studied to

determine whether it would serve as an indicator of intra-uterine fœtal-antigen contact in infants born of motherswhose pregnancies were complicated by urinary-tractinfection. The experimental model consisted of harvestingperipheral-blood lymphocytes from neonates born to twogroups of mothers: those who were known to have hadsignificant Escherichia coli bacteriuria or pyelonephritis atsome time during gestation; and those who, on oneoccasion at least, had normal urine cultures, and werewithout documented disease of the urinary tract. The

lymphocytes were grown in cell culture with autologousplasma and in the presence of a cell-free extract of E. colias the stimulating antigen. On the fourth day, mitoticarrest was produced by colchicine, the cultures wereharvested by centrifugation of the cellular sediments, andcoverslips were treated with Wright’s stain. Mitoses,estimated quantitatively, were considered the experimentalendpoint and an indicator of an anamnestic immune

response. Immunoglobulins (IgA and IgM) were