PRENATAL DIAGNOSIS OF MONOGENIC DISORDERS BY DNA ANALYSIS:
Fifteen years after the first prenatal DNA diagnosis was carried out, by Kan and Dozy in 1978 (for sickle cell anaemia (7)), the DNA-based diagnosis of monogenic disorders is finally becoming considered a routine service, in general because of the rapid advance of knowledge in the field of human molecular genetics. More particularly, the invention and development of the polymerase chain reaction (PCR) has made prenatal detection faster, cheaper, and more sensitive. Prenatal DNA testing is commonly provided by specialized laboratories attached to clinical cytogenetic services, whose experience in genetic counselling is of great importance.
At present, prenatal DNA testing is restricted to monogenic disorders—those diseases in which the clinical phenotype is a direct consequence of the mutation of a single gene. In the long term, it is possible that diagnosis will also become available for multigenic diseases, when their genetic aetiology is better understood.
Tests are available for the majority of the common monogenic disorders, as well as for a great number of rarer ones. Table 1 shows a selection of seven diseases for which prenatal testing is available and commonly requested. All of these diseases, with the exception of congenital adrenal hyperplasia, are relatively common and severe. Cystic fibrosis is the most common lethal genetic disease of childhood in populations of white European origin, fragile X syndrome is one of the most common causes of mental retardation in males, and both sickle cell anaemia and type I (Portuguese-type) familial amyloidotic polyneuropathy (FAP1) are lethal diseases which reach endemic levels in some regions.
It is no longer meaningful (nor indeed possible) to produce an exhaustive catalogue of diseases for which DNA testing is available. In the most recent comprehensive review of this subject, Connor (2) lists forty-seven diseases for which DNA-based prenatal diagnosis had been reported, and a further sixty-nine for which testing would be technically possible if desired. In the last year, this list has expanded by perhaps one quarter.
For DNA testing to be made available for a particular disease, only two basic conditions must be fulfilled:
The former point initially seems self-evident, but incorporates the important concept of genetic heterogeneity: a disease phenotype can be caused by defects in different genes. In the worst cases, different genetic defects simply cannot be distinguished on clinical grounds, potentially leading to the wrong gene being analysed. Furthermore, the clinician must always be ready to reconsider diagnoses that were made years earlier, perhaps because new genes have been found; this can be difficult or impossible if no patients are available for examination and if the clinical documentation is incomplete.
The latter point illustrates one of the major reasons for the recent rapid increase in the clinical significance of molecular genetics: more and more human genes are being characterized and their roles in disease investigated, providing the basic tools for clinical diagnosis. In particular, one of the principal aims of the Human Genome Project is the production of a map of all human genes, which will be a tool of immeasurable value in clinical medecine (1).
In general, two conditions are required to justify prenatal diagnosis of any sort:
Ethical considerations are of great importance in prenatal diagnosis, because for the majority of disorders termination is the intended course of action; the interested reader is referred to the excellent study of Fletcher and Wertz (5).
Prenatal DNA diagnosis is not a screening test
At present, prenatal DNA analysis is a test performed exclusively on indication—in contrast to physical, cytogenetic, and enzymatic examinations, it is not regarded as a screening test.
Prenatal diagnosis is generally contraindicated when the risk of performing the test exceeds the risk of the disease being present. The nature of the molecular genetics techniques used in prenatal diagnosis imposes a high degree of specificity—one test examines one (defined) disorder. Given the incidence of most monogenic disorders, it is evidently only in exceptional circumstances that the risk to a fetus of inheriting a defined disease will exceed the risk of performing the test (which necessarily involves amniocentesis, chorionic villus sampling (CVS), or fetal blood sampling), and therefore that DNA diagnosis is indicated.
In contrast, the more general tests currently employed in prenatal screening can detect the effects of a wide range of different disorders; although each one of these may be relatively infrequent, the cumulative risk is often significant, notably with increased maternal age.
Two situations can be foreseen where prenatal DNA screening might be justified. Firstly, for the rare situation where lethal genetic diseases are endemic in local populations, such as FAP1 in northern Sweden or in certain regions of Portugal (about 1 in 30 affected), and sickle cell anaemia in some regions of the Mediterranean (1 in 3 carriers).
Secondly, amongst women who have requested amniocentesis or CVS for the diagnosis of chromosomal anomalies and have concomitantly accepted the risk of an invasive technique. The incidences cited in Table 1 show that, amongst 10’000 cytogenetic prenatal diagnoses (e.g. for maternal age) in Switzerland, there will be 5 fetuses with cystic fibrosis, 4 males with fragile X syndrome, and 1-2 males with Duchenne muscular dystrophy. In certain populations, there might in addition be 30 with sickle cell anaemia or FAP1. In all these cases, the result of " cytogenetically normal " would be given.
In conclusion, in some situations there may be a clear argument for offering a DNA-based screening of a few monogenic disorders, selected according to their frequency in the population in question.
The purpose of a prenatal diagnosis is to act on the result
The procedure of a prenatal diagnosis, for a monogenic disorder or for any other, should be undertaken only in the aim of taking positive action in the case of an unfavourable result. For the great majority of disorders, at present the only possible action is termination. The parents should be counselled about the different courses of action before starting the diagnostic procedure, to allow them to take an informed decision.
In the future, many diseases may be susceptible to treatment in utero, but at present this is unfortunately a rare option. One common disorder amenable to such treatment is congenital adrenal hyperplasia, where prenatal diagnosis has a proven positive value in avoiding the virilizing effects associated with the defect of the enzyme steroid 21-hydroxylase.
If neither termination nor in utero treatment is being considered, there is rarely an indication for prenatal diagnosis.
It depends on the risk
As was described above, prenatal diagnosis is generally indicated for parents for whom the risk of having an affected fetus are greater than the risk of undergoing the test, or in the rare cases where an in utero treatment is available. Consequently, the major consideration when answering the above question is the precise risk of having an affected child.
Three factors are involved in defining the genetic risk for a couple:
Example: Cystic fibrosis. To illustrate the practical application of risk calculations in combination with DNA testing, we will consider a family with one member affected by cystic fibrosis (CF), an autosomal recessive disorder which is very common in populations of white European origin (Table 2).
The medical genetics of CF are well known:
One boy of the family is affected by the disease (individual III-1). The known autosomal recessive inheritance of the disease indicates that his parents (II-1 and II-2) are obligate carriers of mutations of the CFTR gene. The risk ® for each subsequent child of this couple is a function of four probabilities:
p(father carrier) x p(mother carrier) x p(father transmits) x p(mother transmits)
R = 1 x 1 x ½ x ½ = ¼
With such a high risk, a prenatal diagnosis would of course be indicated. But is a prenatal similarly indicated for the aunt of the affected child and her husband (II-3 and II-4)?
The prior risk that the aunt is a carrier—before any DNA analysis has been performed—is approximately ½, and the risk of her husband is that of the normal population. Consequently,
Rprior = ½ x 1/23 x ½ x ½ = 1/184
The prior risk for the fetus is perhaps low in absolute terms, but is nonetheless sufficiently high (over ten times that of the normal population) to provoke concern.
DNA testing of the parents is useful to modify this risk, and to give a better indication of the options in terms of prenatal diagnosis. Let us suppose that the most common mutations of CFTR are tested in the couple: the aunt II-3 is shown to be a carrier, but her husband III-4 is not. These mutations account for 80% of all CFTR mutations (in terms of frequency), and so the residual risk that the husband is a carrier is only 20% of the prior risk.
Rresidual = 1 x (1/23 x 20/100) x ½ x ½ = 1/460
Even though the one of the couple is a proven carrier, testing of her husband has led to a reduced calculated risk for the fetus, and to a greatly reduced indication for prenatal testing.
These calculations can be utilised for other autosomal recessive disorders, simply by modifying the carrier frequency.
It should also be noted that, in the case of the couple II-3 and II-4, it is not possible to guarantee an informative prenatal diagnosis, because although no mutation has been found in the father, the presence of an undetected mutation cannot be excluded. There are two possible outcomes to a prenatal diagnosis:
The former result is obviously of positive value, and the latter equally obviously has a considerable negative psychological effect. Despite this, in our experience approximately half of couples request prenatal diagnosis in these circumstances after being informed of the possible outcomes.
Two different types of analysis are used for in DNA diagnosis.
Direct analysis is the method of choice
Direct analysis of the mutation responsible for a disease permits a rapid diagnosis, with 100% accuracy, and generally at reasonable cost. In diseases with a unique mutation, such as sickle cell anaemia, the test can be used without any preliminary family studies. In those diseases with multiple mutations, such as cystic fibrosis, it is necessary to study an affected index patient (or carrier), to identify the particular mutation(s) which will be sought during the prenatal diagnosis.
Example: sickle cell anaemia. Fig. 1 shows a family with one child affected by sickle cell anaemia, who is by definition homozygous for the mutant gene haemoglobin (HB) S. The parents are heterozygous, with one mutant and one normal gene. Future pregnancies for this couple can be tested immediately, by directly testing for the presence or absence of the mutation in a CVS or amniocentesis, with the polymerase chain reaction (PCR).
Briefly, the ·ß-globin gene is amplified to near-purity with the PCR, and then tested with the enzyme Dde1. This enzyme recognizes the DNA sequence of the normal gene and cuts it into two fragments (two bands on the gel), but is unable to recognize the mutant gene, leaving it uncut (one band). The pattern of the fragments after gel electrophoresis provides the diagnosis. The result can be obtained within 24 hours of the reception of the sample.
Indirect analysis can be used for many disorders, but is <100% accurate
Direct analysis is frequently impossible, either because the disease gene is not completely characterized or because the exact mutation in a family cannot be identified. In such cases, indirect analysis is the only option, on the condition that the position of the gene on the chromosomes is known and that a linked marker is available. A linked marker is a sequence of DNA—perhaps a gene, perhaps a sequence which codes for nothing—which is physically near to the gene of interest, and which is polymorphic in the population. It may have only two different forms (such as the Rhesus blood group), or many (for example the HLA system of major histocompatibility antigens).
Fig. 2 shows an indirect analysis of a family with a child with the autosomal recessive disease congenital adrenal hyperplasia (CAH). The disease gene, steroid 21-hydroxylase, has many different mutations, and so indirect testing is generally necessary. 21-hydroxylase is adjacent to the HLA genes on chromosome 6, providing a perfect polymorphic marker for diagnosis.
In this family, three different variants of HLA (alleles) are present. Because the HLA genes are so closely linked to the disease gene, the HLA alleles in the affected child can be used as labels to identify the mutant chromosomes in the parents. Thus the father’s mutation is associated with B7, and the mother’s with B2. DNA analysis of the fetal HLA genes indicates that the fetus has inherited a B7 allele from each parent, and by inference a mutant 21-hydroxylase gene from the father but a normal one from the mother. The diagnosis is therefore that the fetus is an unaffected carrier of CAH.
Unfortunately, this analysis carries a small risk of error. During meiosis, every chromosome undergoes at least one crossover event with its homologue. Because the marker used in the diagnosis is not precisely at the site of the mutation, there is a possibility that during the maternal meiosis there was a crossover between the HLA and 21-hydroxylase genes. The outcome of such a recombination would be a pair of chromosomes with the B2 allele on the healthy chromosome and the B7 on the mutant, and subsequently a false diagnosis.
This risk is proportional to the distance between the polymorphic marker and the disease gene, and so it is important to use the most closely-linked markers available. The frequency of recombination should be determined before using a marker, to permit an accurate assessment of the risk of error (in the above example, the risk of error is approximately 1%). In addition, this risk can be almost eliminated by using a marker to each side of the gene: in this case, an error could only arise if two crossovers, one just to each side of the gene, were to occur, which is very unlikely.
Finally, it must be noted that it is essential to have a DNA sample from an affected individual to perform an indirect analysis, to determine which " label " is associated with the mutation in each family member. It is consequently often necessary to maintain a " DNA bank " of samples from affected patients, in case a diagnosis will ever be required in a family.
Prenatal diagnosis of monogenic disease is now accepted as a routine service in clinical genetics, although it should be noted that the analyses are technically demanding and not generally amenable to a " kit " approach but require a specialized laboratory. Many monogenic diseases can be tested now, and more are being added to the list every month. The message to the practising clinician is: if in doubt, ask!
DNA technology is also being applied to cytogenetic analyses, to increase their sensitivity and informativity. Fluorescent in situ hybridization can be used clinically to identify marker chromosomes, to characterize complex rearrangements, to detect microdeletions, or simply to provide rapid detection of chromosomal aneuploidies without cell culture (8). In addition, analysis of DNA polymorphisms has been used in the author’s laboratory to exclude uniparental disomy in some families with balanced translocations (4).
In the future, it is possible that DNA analysis will be offered as a screening test for several of the most frequent disorders, particularly in pregnancies where chromosomal analysis has already been requested.
In the very long term, prenatal diagnosis in very high risk pregnancies may be superseded by preimplantation diagnosis, in which single cells from a number of embryos are tested, and only non-affected embryos implanted. This approach has already been successfully used for cystic fibrosis (6).
I would like to thank my colleagues in the Division of Medical Genetics for their help and for many interesting discussions. The views expressed in this article are the views of the author, and not necessarily policy of this Division.
Edited by Aldo Campana,