Plasma a fetoprotein reference ranges in infancy: effect of prematurity 363 intestinal tract, and rarer ..... contained in Table 3. This latest statistical elabora- ..... Miss J Hyslop and Mr A Irwin, Department of Medical Illustration,. RHSC, for ...
Archives of Disease in Childhood, 1987, 62, 362-369
Plasma a fetoprotein reference ranges in infancy: effect of prematurity J I BLAIR, R CARACHI, R GUPTA, F G SIM, E J MCALLISTER, AND R WESTON Departments of Paediatric Surgery and Biochemistry, Royal Hospital for Sick Children, and Department of Pathological Biochemistry, Gartnavel General Hospital, Glasgow, Scotland
SUMMARY The dearth of plasma a fetoprotein reference ranges for preterm infants often impairs the clinical interpretation of plasma a fetoprotein data collected from ill babies. This study tested our hypothesis that meaningful plasma reference ranges could be established for preterm infants by a simple correction of patient age at sampling date for gestational age deficit at birth. Using a modified radioimmunoassay kit method, determinations of a fetoprotein were performed on capillary and venous blood samples collected from 56 babies aged from birth to 5 months with gestational ages ranging from 26 weeks to 43 weeks. Unmodified plasma a fetoprotein values were grouped according to patient age and examined statistically using established normal theory methods, but these yielded excessively wide reference intervals and non-Gaussian distribution parameters. Acceptable reference ranges were derived using logarithmic transformation of plasma a fetoprotein values and rearrangement against patient age corrected for gestational age deficit. These provisional reference ranges for plasma a fetoprotein in preterm (and term) infants are applied to groups of previously meaningless a fetoprotein results and used to test the potential usefulness of plasma a fetoprotein determination as a diagnostic marker in biliary atresia, hepatitis, and yolk sac derived tumours.
Alpha fetoprotein is a single chain a globulin, of disturb the usual fall in plasma a fetoprotein molecular weight 68 000, normally unique to the concentration. Such a mechanism has been posfetus but often found in increased concentrations in tulated in neonatal viral hepatitis,6 where complasma of children and non-pregnant adults with pensatory proliferation of hepatocytes produces liver disease and some forms of malignant tumours. rapid regeneration of liver tissue and restoration of Alpha fetoprotein is produced in the developing hepatic a fetoprotein synthesis. fetus in equal amounts by the yolk sac and the fetal Since the onco-fetoprotein a fetoprotein was first liver, but at about 12 weeks after conception the identified electrophoretically in the cord blood of yolk sac degenerates and the fetal liver becomes the human infants, its merit as a general marker of main site of synthesis. Fetal plasma a fetoprotein perinatal disease has been studied intensively. The concentration reaches a peak value of 3 000 000 demonstration that this alpha migrating serum U/ml at a gestation of about 14 weeks. ' Alpha protein (otherwise detected in serum only during fetoprotein synthesis almost ceases at parturition pregnancy) was identical in structure to a protein and the infant plasma concentration falls exponen- found in mice bearing transplanted hepatocellular tially from a mean of 50 000 U/ml to a typical adult carcinoma ' was a major stimulus to interest in blood concentration of less than 10 U/ml at 6 to 8 months a fetoprotein as a potential tumour marker. This of age.2 association was confirmed when patients with liver The observation that a fetoprotein synthesis does tumours, in particular hepatocellular carcinoma and not cease entirely at term'-5 may be explained by the hepatoblastoma, were shown to have increased presence of fetal hepatocytes, which, though de- serum a fetoprotein concentrations.9 creasing in number, sustain transient production of Plasma a fetoprotein is now universally estaba fetoprotein during the early postnatal period. Any lished as a reliable biochemical marker of maligclinical condition that interferes with this physio- nancy not only for hepatic tumours but also for yolk logical decline in fetal protein biosynthesis will sac derived tumours, some tumours of the gastro362
Plasma a fetoprotein reference ranges in infancy: effect of prematurity 363 intestinal tract, and rarer teratocarcinomas with Assay method. Plasma a fetoprotein was measured yolk sac elements, such as malignant sacrococcygeal using a routine 'tumour marker' assay based on the teratoma. 1(}13 A valuable feature of plasma a Amersham International AFP kit IM.984,'6 with the fetoprotein monitoring is its reported specificity'3 in following modifications: (1) Preparation of six additional standards by detecting a small malignant yolk sac focus located in dilution of kit standards with the 'zero' the midst of a 'benign teratoma'. standard to give 10 standards in the range Clinical interpretation of plasma a fetoprotein 0-110 U/ml. concentrations in our term and preterm infants has (2) Preincubation of standards and patient been based predominantly on the graph of reference samples with anti-AFP serum for three hours ranges published by Tsuchida et al, which takes no at room temperature before addition of tracer. account of gestational age.'4 We have suspected for (3) Aspiration of supernatants under mild some time, however, that correction for gestational vacuum (Venturi pump). age deficit at birth would greatly enhance the clinical value of many plasma a fetoprotein results obtained (4) All results with an a fetoprotein concentration of 3-4 U/ml or less reported as less than 4 in early infancy. U/ml. (5) Appropriate dilution with the 'zero' standard Patients and methods and re-analysis of all specimens with a fetoprotein values greater than 100 U/ml. Alpha fetoprotein results were reported in U/ml Fifty six babies admitted to the neonatal surgical unit, Royal Hospital for Sick Children, and the based on the International Agency for Research on paediatric department, Queen Mother's Hospital, Cancer Standard 72/225 (1 ng a fetoprotein=1-09 from October 1984 to January 1985 were included in International Units). Interassay imprecision expressed as percentage this study. At the time of sampling the infants were aged from birth to 5 months, with estimated coefficient of variation was 3-9% at a mean a gestational age at birth ranging from 26 to 43 weeks. fetoprotein concentration of 50 U/ml and 3.5% at a Thirty seven preterm infants were studied. Fifty mean concentration of 25 U/ml. babies were singletons, four others comprised two sets of twins, and the remaining two were from two Statistical methods. Data reduction was achieved further twin pregnancies. A detailed list of the range with the aid of a desk top minicomputer, using of clinical conditions with disease incidences is established normal theory methods on age grouped available on request from the authors. Babies with a fetoprotein results, which yielded the following conditions associated previously with abnormal statistics together with their standard errors and plasma a fetoprotein concentrations-for example, 95% confidence limits'7: mean, median, standard six infants with sacrococcygeal teratoma-were ex- deviation, and percentage coefficient of variation. Previous studies on paediatric and adult a cluded. In the case of two infants with gastrointestinal disorders and increasing serial plasma a fetoprotein reference ranges prompted a simple fetoprotein concentrations only the initial a fetopro- logarithmic transformation of uncorrected a fetoprotein results and a fetoprotein results cortein value was included for statistical analysis. Residual plasma (typically 50-100 RI) from rected for gestational age with re-application of the capillary and venous blood samples collected for data reduction procedure. Linear (Y=mX+c) and routine biochemical analyses essential to patient exponential (ln(Y)=mX+ ln(c)) least squares regrescare was used for estimation of a fetoprotein. These sion analyses were performed on ungrouped a residual samples provided a single plasma a fetopro- fetoprotein data paired with both uncorrected and tein result for 34 patients; plasma was obtained corrected patient ages. twice from further blood samples in 13 others, and three to six serial (weekly intervals) a fetoprotein Results plasma samples were obtained for the remaining nine babies. Heparinised blood was spun, separ- An essential prerequisite to the lengthy derivation ated, and stored at 4°C until the primary analyses of age related plasma a fetoprotein reference ranges were completed and the plasma stored at -200C, was the verification of underlying relations between pending a fetoprotein analysis.'5 Study samples plasma a fetoprotein concentration and patient age. were included with batches of routine a fetoprotein Linear and exponential least squares regression determinations and handled in exactly the same analyses were performed on the entire set of 97 manner as a fetoprotein requests for general diag- plasma a fetoprotein values matched in the first nosis. instance with patient age at sampling and then with
364 Blair, Carachi, Gupta, Sim, McAllister, and Weston patient age corrected for gestational age deficit at birth (Table 1). Results of linear regression analysis quickly confirmed earlier assumptions of a significant negative correlation (r=-0.42, n=97, p