Implications of fetoplacental mosaicism on cell‐free DNA testing: a review of a common biological phenomenon

Cell-free DNA (cfDNA) testing, also known as non-invasive prenatal testing/screening, for fetal aneuploidy is one of the most important technical advances in prenatal care and its uptake has increased rapidly. Practitioners offering this screening test must provide comprehensive pretest prenatal counseling, including the risk of possible discordances between cfDNA results and the ultrasound examination and/or the actual fetal genetic chromosomal constitution. Although cfDNA-based screening is similar to traditional serum screening and is aimed at revising the risk of the pregnant woman, assessing the nucleic acids derived from the placenta poses unique interpretative challenges not commonly encountered with traditional serum screening. The purpose of this article is to translate the established knowledge of placental genetics, derived from decades of prenatal diagnostic cytogenetics, into the new cfDNA testing paradigm to assist in pretest prenatal counseling, when test limits and advantages are explained, and post-test counseling, when there is discordance between a screening result and the ‘truth’. cfDNA testing analyzes DNA fragments circulating in the maternal bloodstream that are derived from apoptosis of the external layer of the placenta, the trophoblast1-3. The aim of cfDNA testing is to detect pregnancies that are at an increased risk for fetal aneuploidies, namely trisomy 21 (T21), 18 (T18) and 13 (T13). Several testing strategies are available, including massively parallel shotgun sequencing, chromosome-selective sequence analysis or digital analysis of selected regions and single nucleotide polymorphism (SNP)-only-based analysis4-8. Other methods of examining fetoplacental nucleic acids in maternal blood have been described previously; however, their implementation is not currently widespread in clinical practice9-15. A recent meta-analysis has shown that cfDNA testing has a high sensitivity and specificity for detection of fetal aneuploidies16. The respective weighted pooled detection rates and false-positive rates (FPRs) in singleton pregnancies were: 99.2% and 0.09% for T21, 96.3% and 0.13% for T18, 91.0% and 0.13% for T13, 90.3% and 0.23% for monosomy X (MX), and 93.0% and 0.14% for sex chromosome aneuploidies (SCAs) other than MX. Less well-validated performances were also available for twin pregnancies; the detection rate for T21 was 93.7% and FPR was 0.23%16. The performance of cfDNA testing for other aneuploidies in twin pregnancies is unknown. These data are the result of a pooled analysis of a heterogeneous group of 37 published studies based on different molecular methodologies (massively parallel shotgun sequencing, chromosome-selective sequence and SNP), applied to different pregnancy populations (high and general/low risk), using different recruitment strategies (prospective and retrospective, operators blinded or not blinded) at variable gestational ages (7–40 weeks) and analyzed with different methods of diagnosis of the chromosome abnormality under investigation (prenatal and neonatal karyotyping or clinical examination at delivery on all or only a portion of the whole tested population). Nevertheless, even considering the best possible performance characteristics, cfDNA tests can fail; sensitivity and specificity are never 100% and, although rare, they can provide discordant (false-negative and -positive) results when compared with the real chromosomal constitution of the fetus. In addition to technical issues (such as sample swap or lack of sufficient internal quality control metrics)17, there are biological reasons for discordant results that can be of either fetal or maternal origin. Contributing fetal factors include insufficient or absent fetal fraction (FF)18, 19, fetoplacental mosaicism20-26 and the presence of a vanishing twin27. These may cause false-negative28, 29, as well as false-positive, results27, 30. Maternally derived chromosomal abnormalities (probably also the submicroscopic ones)31-34 are a rare case of maternal occult malignancy35 and usually cause false-positive results31, 34. Maternal organ/tissue transfer from a male donor can cause errors in fetal sex determination36. Because cfDNA testing targets a mixture of DNA fragments of maternal and fetoplacental origin (the so-called ‘total circulating cfDNA’), fetoplacental discrepancies are a known unavoidable biological limiting factor that categorize cfDNA tests as screening tools and clearly not diagnostic assays. Maternal chromosomal background is another limiting factor affecting all screening methods except those based on SNPs-only37. In addition, nearly 2% of samples show a FF < 4%18, 19, 38 and in these cases the rates of T18 and T13, but not T21, are higher than in those with a successful test39. Therefore, failed tests owing to low FF at second redraw have a higher a-priori risk for T18 and T13 and, as such, they necessitate a particular healthcare management including a detailed sonographic investigation and/or a non-cfDNA-based screening test (gestational age permitting) or an invasive procedure. Preliminary data also suggest an increased risk for other fetal chromosome abnormalities; however, systematic studies performing fetal karyotyping in all failed cases with a low FF have not yet been performed40-42. The first successful cytogenetic prenatal diagnosis by karyotyping of chorionic villi (CV) was performed in 1983 at the Mangiagalli Hospital in Milan by Professors Bruno Brambati and Giuseppe Simoni43. At that time the general assumption was that the chromosomal constitution of the placenta faithfully reflected the true fetal karyotype. Soon after the introduction of cytogenetic analysis of CV, scientists encountered a new biological phenomenon, i.e. fetoplacental discrepancies as a consequence of chromosomal mosaicism. The term ‘chromosomal mosaicism’ refers to the genetic composition of an individual derived from a single zygote that has two or more populations of cells with different karyotypes44. In prenatal diagnosis, the mosaicism generally involves abnormal cells with full aneuploidies (usually trisomies) and more rarely involves structural rearrangements. The high prevalence of trisomies found in mosaic karyotypes is explained by the underlying mechanism of mosaicism formation, which frequently involves a non-disjunction (NDJ) error during a mitotic postzygotic cell division or during meiosis I or II (usually maternal) followed by a postzygotic correction of the trisomy45. Depending on the timing and mechanism of the NDJ error, the abnormal cell line could be present in both fetus and placenta (in these cases the NDJ occurred before separation of the fetal and placental compartments), only in the fetus or only in the placenta. The two former conditions are true fetal mosaicism (TFM) and are therefore identifiable by amniocentesis. The latter is defined as confined placental mosaicism (CPM) and is identifiable by placental cytogenetic analysis46, 47. CPM is a ‘false alarm’ (false-positive finding) because it is associated with a karyotypically normal fetus. It is the most common type of mosaicism because a higher number of mitotic divisions are required in extrafetal developing tissues compared with fetal tissues during early development; hence, theoretically, a mitotic error in the early stages of development is more likely to occur in the placenta than in fetal cell lineages48. The gold standard method for cytogenetic analysis of the CV includes the combined karyotype analysis of the cytotrophoblast (by ‘direct’ preparation) and of the mesenchymal core (by long-term culture)49, 50. By performing this combined approach, the mosaics detected in the CV can be classified into three types: Type 1 in which the abnormal cell line is identified in the cytotrophoblast only; Type 2 in which it is in the mesenchyme only; and Type 3 in which both placental layers are affected51. These three types of mosaicism in the CV have different likelihoods of fetal involvement52. Due to the unpredictable distribution of the abnormal cell line in the fetoplacental compartment during embryonic development, when a mosaic condition in the CV is detected during cytogenetic prenatal diagnosis, a confirmatory karyotype analysis of amniocytes should be performed to discriminate between a CPM or TFM condition. Thus, sequential karyotyping of cytotrophoblast, mesenchyme and, in case of a mosaicism in CV, the corresponding amniocytes can identify the mosaicism as one of six different classes of fetoplacental discrepancy: CPM I and TFM IV, in which the chromosome aberration is present in the cytotrophoblast and not in the mesenchymal core; CPM II and TFM V, in which it is present in the mesenchyme and not in the cytotrophoblast; and CPM III and TFM VI, in which the abnormal cell lineage is generalized to both placental tissues (Table 1)52. Because the trophoblastic DNA fragments circulating in the maternal plasma are derived from the external placental layer, cytogenetic analysis of a direct preparation of the cytotrophoblast can further increase our understanding of discrepancies by cfDNA testing. Such a study was indeed performed in which 1001 mosaic placentas were sequentially characterized by karyotyping52. This large study of placental mosaicism elucidated two main findings. First, there is consistent evidence of chromosome-specific differences in the frequency of TFM and CPM among the common aneuploidies. In fact, in the presence of mosaicism in the CV, the likelihood of fetal involvement depends on a combination of different factors including not only the type (1, 2 or 3) and modality (abnormal mosaic or abnormal non-mosaic) of the distribution of the abnormal cell line in the CV, but also the involved chromosome and the type of chromosome abnormality in the abnormal cell line. For example, trisomies 2 and 7 are the most frequent abnormalities among the single trisomies involved in CV mosaicism (77 and 74 cases out of 504 mosaic trisomies, respectively), however they were not identified at follow-up amniocentesis. T13 is less frequently identified in CV (n = 42) and the likelihood of fetal involvement is 2.4% (n = 1). T21 (n = 56) and small supernumerary marker chromosomes (n = 65) show the highest risk of fetal involvement (34%; n = 19 and 22, respectively). However, when both T21 and small supernumerary marker chromosomes show a Type 3 distribution in the CV, with an abnormal mosaic cytotrophoblast and an abnormal non-mosaic mesenchyme, the risk of fetal confirmation increases to 100%52. Second, the karyotype of the cytotrophoblast does not always match that of the fetus. In about 1% of prenatal cases (514/60 347; 0.85% (95% CI, 0.78–0.93%)) the cytotrophoblast, but not the fetus, contains the aneuploid cell line (CPM Types 1 and 3) and vice versa (TFM Type 5)30, 52. These two opposing cytogenetic constitutions of the fetoplacental compartments can potentially result in false-positive (CPM Types 1 and 3) and false-negative (TFM Type 5) results by cfDNA testing (Figure 1)30. The false-negative risk has been calculated to be 1/136 abnormal cases for T13, 1/64 for T18, 1/135 for T21 and 1/14 for MX30. The false-positive risk depends on the level of mosaicism in the cytotrophoblast and on the sensitivity of the specific assay to detect lower levels. Nevertheless, considering a conservative cut-off of 30% detectable mosaic cell line (30% abnormal to 70% normal cells), the estimated FPR is 1/3754 for T13, 1/7472 for T18, 1/7372 for T21 and 1/1421 for MX30. Based on the above evidence that a high-risk cfDNA test result could possibly arise from either a truly abnormal conceptus or a falsely (homogeneous or mosaic) abnormal cytotrophoblast, a confirmatory invasive procedure is recommended to distinguish between the two conditions53-57. This is analogous to the recommended procedure when mosaicism in the CV is identified during first-trimester cytogenetic prenatal diagnosis, because cfDNA testing can be considered a different approach for assessing the same tissue, i.e. the cytotrophoblast. There is currently debate as to whether CV sampling (CVS) is appropriate or whether, due to the possibility of again obtaining a result of mosaicism in the CV, amniocentesis alone should be used for confirmation of fetal involvement. Karyotyping of amniotic fluid cells is the gold standard confirmatory procedure because analyzed cells are mainly fetal in origin and not placental, thus avoiding the issue of fetoplacental discrepancies. However, whereas cfDNA testing is frequently performed in the first trimester (≥ 9–10 weeks' gestation, depending on the test used), amniocentesis requires a minimum gestational age of 15 weeks for testing. Some women could perceive this delay as psychologically untenable, which might therefore lead to an unconfirmed voluntary termination of pregnancy58. Because there are chromosome-specific differences in the frequency of the different types of fetoplacental discrepancies among the common aneuploidies, there is also a chromosome-specific risk of obtaining a mosaic result from the CV if CVS is performed after a high-risk cfDNA result, which would require follow-up amniocentesis: 2% for T21, 4% for T18, 22% for T13 and 59% for MX59. These risks were studied and calculated based on a retrospective analysis of a prospectively collected database of cytogenetic CVS results. Despite various limitations, including limited follow-up at delivery, these are the only data presently available. This study suggests that, in the case of a high-risk cfDNA result for T21 and T18, it may be reasonable to offer CVS as the diagnostic procedure of choice as the risk of a obtaining a result of mosaicism in the CV is equal to or slightly higher than the overall risk of mosaics in the general pregnant population (∼2%)52. In the case of a high-risk cfDNA result for MX, the frequency of this mosaicism is high and prenatal diagnosis should be deferred to the second trimester when an amniocentesis can be performed even if the sonogram is normal. When T13 is involved, the frequency of mosaicism is also relatively high and likewise may warrant offering amniocentesis as the procedure of choice (especially if the sonogram is normal). As a general rule, if the sonographic examination detects anomalies, CVS could be offered to the patient regardless of the risk of mosaicism or the cfDNA result, because of the obvious benefits of an early if a CVS than amniocentesis as the confirmatory invasive procedure after a high-risk cfDNA test result for aneuploidies, the gold standard cytogenetic method for both the cytotrophoblast and the mesenchyme is to provide the most risk for fetal of fetal sex poses that from sex for Although for genetic diagnosis have sex similar have not been for cfDNA testing. studies have that, for the presence of a non-invasive testing can be used in the clinical for high-risk cases (such as in and to this approach as non-invasive prenatal The that this approach is is that, with even a tested genetic the presence of a chromosome can be detected the which should the of This approach does not require or However, most cfDNA tests are based on and not on by sex rates for methods were in two studies to be (95% CI, male (95% CI, and (95% CI, and male (95% CI, For methods they were to be (95% CI, and male (95% CI, and (95% CI, and male (95% CI, cfDNA testing can provide fetal results discordant from the ultrasound of fetal and/or from the confirmatory diagnostic There are possible depending on the of findings. sample swap when the cfDNA fetal sex is different from confirmatory karyotyping and ultrasound the likely cause of is a placental or the of when ultrasound of fetal and karyotyping a male and cfDNA a fetus, of a or cytotrophoblast should be The frequency of fetoplacental mosaicism with a cytotrophoblast in a male pregnancy has been to be When ultrasound and karyotyping a and cfDNA a male fetus, i.e. a fetoplacental mosaicism with a male cell line in the cytotrophoblast, a of a male or a maternal should be when fetal on the sonogram not match the confirmatory karyotyping and cfDNA a genetic of development should be when cfDNA and karyotyping a male and ultrasound of fetal show sex of a fetus should be male and of When cfDNA test results and karyotyping a and ultrasound show a male fetus, of a fetus should be presence of a or an (Figure for fetal submicroscopic number by cfDNA testing is a recent and of in to possible implementation the common aneuploidies. There are two types of cfDNA testing for of specific regions associated with and from to in and the whole which the presence of on the type of in rare trisomies) on all A of is also the for karyotyping fetal Some suggest that may be possible for cfDNA are in of normal with normal at all maternal in the main risk factor for fetal chromosomal abnormalities in women ultrasound In addition, these the implementation of cfDNA testing to include as targets as offering women non-invasive to pregnancy to invasive although there are the of amniocentesis in Although there are known and factors the performance of cfDNA testing for common aneuploidies, evidence suggests that fetoplacental mosaicism is one of the biological of discordant test results. of the of fetoplacental mosaicism on cfDNA testing is the performance of the actual placental cytotrophoblast in the true presence of the fetal aneuploidy 2 the performance of cfDNA testing and karyotyping of cytotrophoblast in the studies of and Although the number of samples in the is these studies a similar in the performance of cytotrophoblast karyotyping compared with that of cfDNA for T21, for T18 and for T13, and other in a study using a that can false-positive cfDNA results based on maternal and vanishing the are in line with the rates by cytotrophoblast The of cfDNA testing for MX, and more generally for is to be and and The diagnostic performance of cytotrophoblast karyotyping is lower for than for the trisomies (Table and to with the of cfDNA testing for MX. In addition to maternal mosaicism for a further possible could in the of the sex chromosomes to and mitotic NDJ errors that mosaics in the These three cell can be in the different fetoplacental compartments and in the two placental layers with different and cfDNA testing for there are and associated with population screening for these The most to the limited of data on fetoplacental discrepancies for and submicroscopic Although more rare than aneuploidies, structural and structural of which submicroscopic that are associated with a clinical can be involved in fetoplacental The frequency of in CV for structural is and the risk of fetal involvement for those in CV of Types 1 and 3) is and and show a risk of fetal involvement of and It is to of would be detected as false-positive cases by cfDNA testing for because the detection of even those that are depends on the specific FF and However, it is that a portion of these structural would be a of false-positive results when cfDNA testing for There are also mosaics in CV that the mesenchymal core only in CV Type show a risk of as a TFM of for for and for TFM Type for structural are of a false-negative risk because they not the cytotrophoblast, which is the of cfDNA As a and conservative if of the and of the that are associated with an abnormal due to the presence of in the marker the false-negative risk due to TFM Type for structural would be (TFM Type Type true with a = 1 data provide the true of fetoplacental discrepancies for submicroscopic thus it is to of would be or falsely detected due to mosaicism in CV, but there is evidence that they cfDNA testing for and rare trisomies are also As described there are two possible biological a high-risk cfDNA a truly abnormal conceptus placenta and fetus in a or mosaic or a falsely abnormal cytotrophoblast in the presence of a normal fetus (CPM Types 1 and When the first condition involves a rare the that there are rare cases described in the in which it was detected in amniotic fluid cells or fetal tissue, the presence of a condition in the conceptus is to result in a pregnancy or a In fact, rare trisomies in prenatal diagnosis on the common in of In the second (CPM Types 1 and the only rare mosaic trisomies in CV a likelihood of fetal confirmation are the mosaic trisomies and of a fetal mosaic or are unpredictable and on the mosaicism level in the As for and even for rare full chromosomal trisomies, their presence in fetal in a consistent level of mosaicism would fetal ultrasound abnormalities or a even if a diagnosis of CPM could be a for pregnancy and the of a high-risk cfDNA test result for rare trisomies confined to the placenta and depends on the level of mosaicism in the placenta. consequence of the introduction of targets in cfDNA testing is a high rate of invasive are a to for non-invasive and, invasive confirmatory These should lead to when cfDNA testing for to pregnant a is by a-priori risk for all possible fetal or likely of an fetal sample should be in combination with pretest counseling, including about rates to the invasive which is lower than the rate in Based on fetal development, cfDNA testing performed at early gestational ages weeks) to be of limited clinical to is of abnormal In addition, a comprehensive sonographic of a screening test to detect fetal abnormalities prenatal diagnosis be performed in cases in which an ultrasound abnormality is detected after a cfDNA women could be falsely by a cfDNA test result as common trisomies for only a portion of the chromosomal abnormalities associated with fetal ultrasound a of to 15 weeks' is to a diagnosis by which could cause and untenable, For the same the of for screening tests could be in the case of pregnancies obvious ultrasound abnormalities at the In fact, should be on the time when a cfDNA test is used in a after a serum screening Due to the of the underlying the cfDNA test and the of prenatal genetic diagnosis to the of comprehensive pretest genetic including the benefits and of screening diagnostic testing and of cfDNA testing must implementation of prenatal cfDNA testing be further to its than its important than the addition of new chromosomal targets that are and of clinical it be on the and of to an and of the the of and in the quality of performing cfDNA clinical studies for new and the of the cfDNA This should include of the FF in sample for tests that are not of the different methods of FF with the introduction of a common method and of the performance at FF lower than the standard This is a issue not only the blood redraw and test rates but also the performances of the test and, in the of the The FF level of detection can only be the of and by the clinical and standard and of the and test performances large of all with a low by and blinded clinical studies with follow-up at is a prenatal diagnosis performing cytogenetic analysis and chromosomal on prenatal as well as cfDNA testing and serum screening for fetal has