Cell-Free Fetal DNA in Maternal Blood: Title and subTitle BreakEvolving Clinical Applications

In this issue of THE JOURNAL, the findings reported in the study by Dhallan and colleagues¹ on enhancing recovery of cell-free DNA in maternal blood have major clinical implications. Developing a reliable, transportable technology for cell-free DNA analysis impacts 2 crucial areas—prenatal genetic diagnosis and cancer detection and surveillance. In prenatal genetic diagnosis, detecting a fetal abnormality without an invasive procedure (or with fewer invasive procedures) is a major advantage. Likewise in cancer surveillance (eg, in patients with leukemia), monitoring treatment without having to perform a bone marrow aspiration for karyotype also would be of great benefit.

Prenatal genetic diagnosis has been available in the United States since 1968, when chromosomal abnormalities and metabolic traits proved detectable by analysis of amniotic fluid cells obtained by amniocentesis. The most common indication for prenatal genetic testing is maternal age older than 35 years; other indications include prior trisomic offspring, balanced parental chromosomal rearrangements, or increased risk for a mendelian trait. The standard approach is to offer women either second-trimester amniocentesis or first-trimester chorionic villus sampling, which are comparable or nearly comparable in safety. The high diagnostic accuracy of these techniques must be balanced against the risks of undergoing an invasive procedure.² Estimates for procedure-related fetal loss following amniocentesis range from 1 per 200 to 1 in 400 to 500.^{3 - 5} Nonetheless, because all invasive procedures have some risk, considerable effort has gone into developing noninvasive means of prenatal diagnosis.

Second-trimester maternal serum analyte screening (for alpha fetoprotein, unconjugated estriol, human chorionic gonadotropin [hCG], and inhibin A) can identify perhaps 70% to 75% of fetuses with Down syndrome, based on identifying those 5% of pregnant women having risk equal to that of a 35-year-old.⁶ In Europe, maternal serum analyte screening is commonly offered to all pregnant women, whereas in the United States it tends to be offered routinely only to women younger than 35 years; older women initially are offered an invasive procedure.

First-trimester noninvasive screening for trisomy 21 is also available, using ultrasound measurement of nuchal translucency, as well as measurement of levels of maternal serum pregnancy-associated plasma protein A, and maternal serum hCG. A recent collaborative study from the US National Institute of Child Health and Human Development (NICHD) has shown sensitivity as high or higher in the first trimester as in the second trimester.⁷ Screening sequentially in both the first and the second trimester has been advocated and indeed should have the highest detection rate. However, this approach requires withholding results in the first trimester and is limited by the risk of losing patients to follow-up. For instance, in the UK Serum, Urine and Ultrasound Screening Screening Study (SURUSS) cohort of 47 053 patients, 15 278 did not return for second-trimester testing.⁸ Thus, sequential (integrated) screening seems unlikely to be the most practical approach, and first-trimester noninvasive screening will probably become the more common strategy.

Despite these promising noninvasive approaches, sensitivity does not approach the 100% possible with an invasive procedure. Thus, the search continues for noninvasive methods independent of or complementary to maternal serum analytes or ultrasound. One such method involves detection and analysis of fetal cells recovered in maternal blood. In 1991 and 1992, our group demonstrated that fetal trisomy 21 and 18 could be detected by fluorescent in situ hybridization (FISH) using chromosome-specific probes on flow-sorted intact fetal cells.^{9 - 11} One intact fetal cell is estimated to be present per cubic centimeter of maternal blood. The collaborative NICHD trial demonstrated an aneuploidy detection rate of 74%.¹² Unfortunately, robust, reproducible, intact fetal cell recovery has proved difficult; therefore, this cannot yet be recommended routinely.

While methods for analysis of intact fetal cells continue to improve, correlative approaches are also being pursued. In 1998, Lo et al¹³ demonstrated cell-free fetal DNA in plasma from healthy pregnant women, using quantification through polymerase chain reaction (PCR). Surprisingly high concentrations of fetal DNA—nearly 5% of total maternal DNA—were detected in plasma of pregnant women. Mean fetal DNA was 25.4 genome equivalents (GE) per milliliter in early pregnancy, increasing to 292.2 GE/mL in late pregnancy. Such large amounts of cell-free DNA cannot simply be the result of degradation of the rare fetal cell. Presumably, cell-free fetal DNA is derived from the placenta, and indeed mRNA for fetal hCG and for human placenta lactogen is recoverable in maternal plasma.¹⁴ That cell-free fetal DNA is consistently detected in maternal blood during pregnancy raises diagnostic potential.

Initial diagnostic application of cell-free DNA involved detection of fetuses who had inherited a mutant allele from an affected father. If the father has a DNA sequence that the mother lacks, presence of that sequence in maternal blood must be of fetal origin; thus, the fetus has inherited the mutant paternal allele. Detection of a paternally transmitted autosomal dominant trait (ie, a fetus with hemoglobin Lepore disease) was reported initially by Camaschella et al.¹⁵ Analysis of cell-free DNA is potentially valuable in managing Rh(D) isoimmunization. Rh-negative women (genotype d/d) lack the Rh(D) D antigen as result of a deletion. Mothers may develop antibodies (anti-D) if exposed to Rh(D) erythrocytes. If antibodies are present and the father is heterozygous (D/d), there is 50% chance of transmitting the D allele, leading to isoimmunization. Presence of the D allele in maternal blood can be detected and has been shown to indicate that the fetus has inherited the D allele from its father.^{16 - 17} This information can be invaluable early in pregnancy. A converse strategy becomes applicable later in pregnancy. Rather than all Rh-negative (d/d) women having to receive Rh immune globin at 27 weeks' gestation (the current recommendation), only those women whose fetuses are definitively Rh(D) might be treated. Again, the Rh status could be determined by studying maternal blood for the (paternal) D allele.

Analysis of cell-free fetal DNA has other clinical applications. Total cell-free fetal DNA levels are increased 2-fold when the fetus has trisomy 21, even though the fetal DNA need not be derived from a gene locus on chromosome 21.¹⁸ Cell-free fetal DNA could thus serve as an additional (and perhaps independent) maternal serum analyte for aneuploidy screening.

Analysis of cell-free DNA also may be applicable in monitoring complications of pregnancy. Concentrations of cell-free DNA correlate with gestational age and are low in the first trimester but increase in the second and third trimesters.¹³ A sharp increase in fetal DNA levels in maternal plasma during the last 8 weeks of pregnancy has been demonstrated, presumably indicating breakdown of the maternal fetal interface and placental barrier.¹⁹ Any pathological process that disturbs the placenta should be accompanied by increased levels of cell-free fetal DNA in maternal blood. Cell-free DNA has already been shown to be increased in pregnancies complicated by preeclampsia.²⁰ This increase should become evident earlier in pregnancy than other physical or biological disturbances.

Cancer detection and surveillance is the second general area in which cell-free DNA analysis is likely to become applicable clinically. Nucleic acids (DNA and RNA) in plasma were first observed more than 50 years ago. However, the source, fate, and usefulness of this DNA was not determined until tumor-specific extracellular DNA fragments could be studied in the plasma of cancer patients. In 1989, Stroun et al²¹ indirectly verified this by demonstrating that the DNA found in the plasma of cancer patients displayed neoplastic characteristics, such as DNA strand instability. Subsequent groups confirmed the principle through mutation analysis, loss of heterozygosity, and microsatellite testing, which correlated cell-free DNA from the blood of patients with cancer with the DNA from their primary tumor. Anker et al²² showed that more than 80% of the cell-free DNA in plasma of patients with colorectal cancer was representative of the tumor DNA, judged by specific mutation in the K-ras oncogene. Mutations of p53 and APC also have been detected.^{23 - 24}

All of these observations raise tantalizing diagnostic prospects. If cancer-derived cell-free DNA can be robustly analyzed, and if detection is possible with only a few genome equivalents per milliliter of blood, a sea change in the current approach to monitoring cancers could occur. Detection or surveillance might no longer depend on imaging technology (eg, magnetic resonance imaging) or surgery. Rather, less expensive, more convenient, and perhaps even more sensitive cell-free DNA analysis could be used to identify and analyze cancer-specific sequences.

Thus, the findings of Dhallan et al demonstrating an increased percentage of cell-free DNA in maternal blood hold tremendous promise. In their study, the percentage of cell-free DNA was increased in 7 of 10 matched samples: 20.2% in treated vs 7.7% in untreated samples. The improved recovery shown with use of formaldehyde is plausible. Three sources of circulating DNA have been hypothesized: (1) dying cells (necrotic or apoptotic), (2) active DNA secretion, and (3) terminal differentiation. Apoptosis seems the most likely source. If so, circulating cell-free DNA should exist in the form of apoptotic bodies or nucleosomes. Irrespective of the type of bound vesicles in which fetal DNA resides, treatment with formaldehyde should have a salutary effect in its stabilization. Formaldehyde-tested plasma may specifically prove more resilient to variations introduced by delayed processing time, varying centrifugation speeds, and storage conditions.

The report by Dhallan et al also raises several other important questions. For instance, if cell-free DNA methods were to be incorporated clinically, would treatment of blood with formaldehyde actually be needed? Given the relatively large amounts of fetal DNA present in the maternal plasma, are methods to stabilize cell-free DNA gratuitous? These approaches are, in fact, necessary because the amount of fetal DNA in early pregnancy, or the amount of cell-free circulating DNA derived from early neoplasia, will remain low compared with background levels. Whenever target DNA (ie, fetal or neoplastic) is present in low copy numbers (eg, 1-3 copies/mL), quantitative PCR is at the limit of its sensitivity; thus, detecting the target becomes hazardous, especially given the necessity to minimize false-negative results in cancer surveillance. In the current study, the authors used conventional PCR techniques and serial dilutions of each sample to quantify fetal Y-specific DNA sequences, the Y simply serving as a surrogate for interloping DNA of any type. DNA proved detected if as few as 3 fetal copies per milliliter of blood exist.

The report by Dhallan et al is an important step in improving detection of cell-free DNA. Further refinements in techniques should maximize recovery of cell-free DNA and facilitate practical application. With prospective studies focusing on clinical applications of these findings, profound clinical implications could emerge for prenatal diagnosis and cancer surveillance.