Counseling prior to cfDNA screening: are we giving the right numbers?

Abstract

Objective

In this modeled study, we aim to evaluate the test performance of cfDNA screening for trisomy 21, 18, and 13 individually and in combination to help with pre-test counseling.

Methods

We used the German birth statistics 2023 and calculated the expected number of trisomy 21, 18, and 13 for each maternal age. The number of screen-positive tests (true and false) for trisomy 21, 18, and 13 was estimated based on maternal and gestational age-related risks for the three trisomies. The positive predictive value was calculated for each trisomy individually and in combination according to maternal age. In addition, we applied an uninformative test rate of 1.2%.

Results

In 2023, 692,730 children were born in Germany and the predicted number of trisomy 21, 18, and 13 fetuses are 2,521.71, 1,084.70, and 344.31, respectively. In total, cfDNA screening in all pregnancies would result in 4,940.9 screen-positive pregnancies (screen-positive rate of 0.7%) and 13,253.69 uninformative tests. Up to 34 years of age, the screen-positive rate would be between 0.3 and 0.5%, but for older women, the screen-positive rate would be between 0.9 and 11.3%. The combined PPV for the three trisomies would be 48.7% in women of less than 20 years of age and increases to 98.8% in women who are 45 years old or older.

Conclusion

cfDNA screening is an excellent screening test for the most common trisomies. However, counseling prior to the screening test is essential. Our study provides useful numbers which can be used for this purpose.

What does this study add to the clinical work

cfDNA screening is an excellent screening test for the most common trisomies. However, counseling prior to the screening test is essential and needs to be based on the prevalence of the disease and the test performance of the screening test.

Introduction

In recent years, cell-free DNA screening (cfDNA) has become the gold standard for screening for trisomy 21 [1]. However, there are significant differences in how cfDNA screening is implemented in different healthcare systems [2]. Many countries, such as the UK, have introduced cfDNA screening as a second-line test after combined first trimester screening, while other countries, such as Belgium, offer cfDNA screening as a first-line test for all pregnant women [3–5]. In some countries, such as Germany, cfDNA screening can be carried out if the obstetrician in charge identifies a particular risk that justifies cfDNA screening [6]. However, regardless of the national screening policy, many women request cfDNA screening as first-tier screening test and are willing to pay for the test on a private basis if the national healthcare system does not cover it [7, 8].

Genetic counseling is essential before cfDNA screening is performed. One of the most important messages that needs to be conveyed is the non-diagnostic nature of the test and its specific capabilities. Generally, for this purpose, the results of the study of Rose et al. are used [9]. In this meta-analysis, the detection and false-positive rates in screening for trisomy 21 are 98.8% and 0.04%, respectively, and the corresponding positive predictive value (PPV) is 91.8%. However, this information is misleading.

First, in the overwhelming majority of cases, cfDNA screening is carried out not only for trisomy 21 but also for trisomy 18 and trisomy 13. The detection (DR) and false-positive rates (FPR) for trisomy 18 and 13 are similar at 98.8% and 0.07% and 100% and 0.04% but, due to the lower prevalence of trisomy 18 and 13, the PPV is only 65.8% and 37.2%, respectively [9]. Therefore, it may be more appropriate to use adjusted figures for counseling women that combine all three screening tests for trisomy 21, 18, and 13. This includes a sum FPR and an adjusted PPV that includes all three screening tests.

Second, the PPV depends on the prevalence of the disease for which the screening test is being done. Since the risk for the common autosomal trisomies increases with increasing maternal age, the PPV must be adjusted accordingly. A combined PPV for all maternal age groups overestimates the PPV in younger women and underestimates it in the older population.

Third, the fact that some cfDNA test results are not informative needs to be considered [10–17]. Commonly, a second test is performed sometime after the uninformative result is obtained. In about 1% of cases, a result cannot be obtained even after a second analysis. Most frequently, this is due to maternal or fetal factors, which may result in a decreased cfDNA maternal blood concentration to the point where it does not reach the threshold for analysis. These include excessive maternal weight, use of heparin, maternal autoimmune disease, and chromosome abnormalities associated with a small placental size, such as triploidy, trisomy 18 and 13, and monosomy X [17]. Other reasons are related to specific laboratory criteria and algorithms. For example, in cases where a genome wide cfDNA test is performed and an abnormal profile related to the maternal background is detected, laboratory may issue an inconclusive report. Therefore, cases where there is no result constitute a heterogeneous group where the risk of a fetal chromosome abnormality is variable and needs to be evaluated on a case-by-case basis. Grati et al. demonstrated that, in an unselected population with heterogeneous cfDNA test platforms and varied referral indication for cfDNA test, the risk for a chromosomal abnormality is up to 14% [18]. Therefore, a patient with a repeat uninformative cfDNA test should be offered further investigations, including invasive testing [19–21]. In this case, however, the likelihood of identifying a trisomic fetus by invasive testing is much lower, as most fetuses with uninformative tests are normal. Regardless of the cause of an uninformative test and its various outcomes, pre-test counseling should include this aspect of cfDNA screening and the potential psychological distress it may cause.

In this modeled study, we aim to evaluate the test performance of cfDNA screening for trisomy 21, 18, and 13 individually and in combination and stratify the performance according to maternal age to help with pre-test counseling.

Methods

Our modeled approach uses maternal age-related risks for trisomy 21, 18, and 13 that are adjusted for gestational age as well as the 2023 German birth statistics to compute the test performance for each of the three trisomies individually and in combination. The results are stratified according to maternal age.

Details of the approach are as follows:

1. The risks of trisomy 21, 18, and 13 were set at 12 week gestation for each maternal year. These were extracted from the studies of Cuckle and Snijders et al. [22–24]. Adjusting for gestational age is necessary due to the high miscarriage rate in the trisomic population. Miscarriages in the euploid population between 12 and 40 weeks are rare; therefore, they were not considered.

2. We took the distribution of births in Germany in 2023 and stratified it by maternal age at delivery. For each maternal age year, we calculated the expected number of pregnancies with trisomy 21, 18, and 13 as well as the number of pregnancies without the three trisomies (non-trisomy group) at 12 week gestation [25]. Artificial reproductive techniques that could carry an effect on the maternal age-related risk such as oocyte donation are not recorded in the German birth statistics and were therefore not addressed in the analysis.

3. Based on the results of the meta-analysis by Rose et al., which evaluates the performance of cfDNA testing (detection rate for trisomy 21, 18, and 13: 98.8%, 98.8%, and 100%), we estimated the number of screen-positive fetuses with trisomy 21, 18, and 13 [9]. The FPR in cfDNA screening for trisomy 21, 18, and 13 (0.04%, 0.07%, and 0.04%, respectively) were applied individually to the non-trisomy group to compute the overall numbers of false-positive cases and the sum was then used to compute the combined FPR. In pregnancies, where the maternal age was higher 50 years, the risk of a 50-year-old woman was used.

4. For each maternal age year, we calculated the total screen-positive rate and the PPV. For clarity’s sake, we grouped these results in the following categories: women < 20 years, 20–24 years, 25–29 years, 30–34 years, 35–39 years, 40–44 years, and > 45 years The number of screen-positive pregnancies in each group reflects the sum of screen-positive cases in each maternal age year.

5. We added an uninformative test rate of 1.2% to each maternal age group according to the systematic review of Palomaki et al. [10]. This group requires further investigations such as an additional ultrasound scan and even invasive testing, depending on the outcome of the ultrasound examination and the mother's expectations. Since the management of pregnancies with an uninformative test result must be developed on a case-by-cases basis, we grouped them separately from the high-risk results.

For the data analysis, Microsoft Excel for Mac 365 (Version 16.96.1, Redmond/Washington, United States) and IBM SPPS30 (Armonk/New York, United States) was used.

Results

In 2023, there were 692,730 children born in Germany. Mean maternal age was 31.7 (SD 5.3) years and the 25–75th interquartile was 28.0–35.0 years. The maternal age distribution is shown in Fig. 1.

Fig. 1.

Maternal age distribution at birth in Germany in 2023

Figure 2 shows the risk for trisomy 21, 18, and 13 according to maternal age and Fig. 3 shows the expected number of fetuses with trisomy 21, 18, and 13 also according to maternal age. Based on the maternal age distribution at birth and the maternal age-related risk for trisomy 21, 18, and 13 at 12 week gestation, the predicted number of affected fetuses is 2,521.71 (95% prediction interval: 2,423,28–2,620.13), 1,084.70 (95% prediction interval: 1,020.14–1,149.25), and 344.31 (95% prediction interval 307.94–380.68), respectively.

Fig. 2.

Maternal age-related risk for trisomy 21, 18, and 13 at 12 week gestation

Fig. 3.

Distribution of pregnancies with trisomy 21, 18, and 13 according to the maternal age

Table 1 includes the expected number and the proportions of screen-positive cfDNA results, which are stratified according to maternal age. For example, our model predicts that the maternal age group of less than 25 years old (65,705 pregnancies) would contain 99.1 (0.15%) trisomic pregnancies. In contrast, in the group of women who are 45 years or older (3,088 pregnancies), 349.2 (11.31%) pregnancies with trisomy are predicted.

Table 1.

Expected number of fetuses with trisomy 21, 18, and 13 as well as fetuses without the common autosomal trisomies according to the maternal age distribution in Germany. Furthermore, the table shows the expected number of true positive (fourth column), false positive (sixth column), and uninformative cfDNA screening tests (eighth column) as well as the overall number of cfDNA tests that require further examinations (ninth column)

Maternal age	All Pregnancies	Expected pregnancies with trisomy 21, 18 and 13	High-risk pregnancies by cfDNA screening (true positives)^	Expected pregnancies without trisomy 21, 18, and 13	High-risk pregnancies by cfDNA screening (false positives)^^	Total number of high-risk pregnancies (false + true positives)	Un-informative tests*	Total number of pregnancies with a need of further examinations**
(years)	(n)	(n)	(n)	(n)	(n)	(n)	(n)	(n)
< 20	8,074	11.60	11.47	8,062.40	12.09	23.56	96.89	120.45
20–24	57,631	87.49	86.54	57,543.51	86.32	172.85	691.57	864.43
25–29	156,663	291.47	288.30	156,371.53	234.56	522.86	1,879.96	2,402.82
30–34	257,142	818.39	809.49	256,323.61	384.49	1,193.98	308.70	4,279.68
35–40	168,478	1,336.62	1,322.09	167,141.38	250.71	1,572.80	202.74	3,594.54
40–44	41,654	1,055.94	1,044.46	40,598.06	60,90	1,105.35	499.85	1,605.20
> 45	3,088	349.21	34.,41	2,738.79	4.11	349.52	37.06	386.57
Total	692,730	3,950.71	3,907.76	688,779.29	1,033.17	4,940.93	8,312.76	13,253.69

^Assuming a detection rate in screening for trisomy 21, 18, and 13 of 98.8%, 98.8%, and 100%

^^ Assuming a false-positive rate in screening for trisomy 21, 18, and 13 of 0.04%, 0.07%, and 0.04%

^*Assuming 1.2% uninformative test results of all pregnancies in this age group

^**False + true positives + uninformative tests

In total, cfDNA screening in all pregnancies would result in 4,940.9 screen-positive pregnancies or in a screen-positive rate of 0.7%. Up to 34 years, the screen-positive rate would be between 0.3 and 0.5% but for older women, the screen-positive rate would be between 0.9 and 11.3% (Table 2).

Table 2.

Positive predictive value of cfDNA screening for trisomy 21, 18, and 13 stratified by maternal age

Maternal age	PPV for trisomy 21	PPV for trisomy 18	PPV for trisomy 13	PPV in screening for all three trisomies	Proportion of high-risk pregnancies
(years)	(%)	(%)	(%)	(%)	(%)
< 20	69.4%	35.8%	23.9%	48.7%	0.29%
20–24	70.6%	37.1%	24.9%	50.1%	0.30%
25–29	74.6%	41.9%	28.9%	55.1%	0.33%
30–34	83.4%	55.3%	41.0%	67.8%	0.46%
35–40	92.7%	75.6%	63.5%	84.1%	0.93%
40–44	97.6%	91.0%	85.0%	94.5%	2.65%
> 45	99.5%	98.0%	96.5%	98.8%	11.32%

In the fifth column, the combined positive predictive value is shown. The sixth column gives the proportion of high-risk pregnancies according to maternal age

Table 2 shows the PPVs in screening for trisomy 21, 18, and 13 individually as well as the combined PPV in screening for all three trisomies. The calculated PPVs for trisomy 21, 18, and 13 are between 69.4 and 99.5%, between 35.8 and 98.0%, and between 23.9 and 96.5%, respectively.

The combined PPV would increase from 48.7% in women of less than 20 years of age to 98.8% in women who are 45 years of age or older.

Table 1 also contains the expected number of uninformative tests (1.2% of all cfDNA tests). In total, there would be 8,312.76 cases with uninformative test results. Combining these with the screen-positive test results, there would be 13,253.66 (1.9%) pregnancies that would need further evaluation. The proportion of test results that need further evaluation varies significantly with maternal age (Fig. 4).

Fig. 4.

Proportion of the three test result types “true positive”, “false positive”, and “uninformative test result” relative to the total and stratified by maternal age

If all the pregnancies that need further evaluation would undergo invasive testing, the proportion of trisomic fetuses would be much lower than expected based on the PPV based on high-risk results only (Table 2). In the maternal age groups of less than 30 years, the proportion would be between 9.5 and 12%. Between 30 and 39 years, the proportion would be between 18.9 and 36.8%, and in women of 40 years and more, the proportion would increase from 65.1 to 89.4%, respectively (Fig. 4).

Discussion

Our modeled calculations arrived at some key figures that are useful for counseling women prior to performing a cell-free DNA screening test. For this purpose, we have used data from a normal Western population, in this case from the German society. Figure 3 shows the expected frequency of trisomy 21, 18, and 13 as a function of maternal age. For all 25-year-old pregnant women in Germany, for example, there are fewer than 25 women with a fetus with trisomy 21 per year. In terms of pre-test counseling, we suggest using combined test statistics and especially a combined PPV instead of individual rates when screening for trisomy 21, 18, and 13 as this informs better, from the subject’s perspective, about the test performance. In addition, this PPV should be stratified according to maternal age (Table 2).

The rate of uninformative tests must also be taken into account. Of all the pregnancies that need further investigation, about 60% are due to uninformative results, if one assumes a 1.2% flat rate for all maternal ages. This proportion is the largest in the younger age groups. Women with an uninformative test result require individualized management, which includes a detailed ultrasound examination and, for some women, invasive testing. Grati et al. assessed the outcome of 137 cases with two uninformative test results. An abnormal karyotype was found in 13.9% of the cases. The proportion of chromosomal defects varied substantially according to the reason of the uninformative test result. If it was due to low fetal fraction, 7.7% of the fetuses had an abnormal karyotype. However, if the sample did not pass the quality control, based on specific laboratory criteria, 15% of the fetuses were chromosomally abnormal [18]. Bardi et al. highlighted the importance of a detailed ultrasound examination in cases with two uninformative cfDNA tests. If there were no fetal defects, only 1.1% of the fetuses had a genetic defect. In contrast, in the group of fetuses with fetal structural defects, a third had a genetic anomaly [26].

Our calculations raise questions about the utility of cfDNA screening as a first-line test in the younger population. The risk of trisomy here is much lower than published figures would indicate, i.e., the meta-analysis by Rose et al. [9]. The authors reported PPVs for trisomy 21, 18, and 13 of 91.8%, 65.8%, and 37.2%, respectively, if screening is done in the entire population. From a healthcare cost perspective, it does not make much difference if younger people, i.e., those under 25 years, have direct access to cfDNA screening. Less than 10% of all pregnant women in Germany belong to this age group, and in total, less than 200 women would have a positive screening result. Even in this group, the PPV is better than for some other screening tests [27]. However, if the inconclusive test results are also considered as a reason for an invasive test, there are already five times more test results that with a need for further examinations. On an individual basis, however, there is a significant difference whether the screening test is carried out in a 20-year-old or a 35-year-old woman. This aspect is clearly demonstrated in Table 2 and Fig. 4. If the inconclusive tests are also considered as a reason for invasive testing, only one-tenth of the invasive tests would identify a trisomic karyotype in those women who are less than 25 years old. This number does not increase to over 50% until the age of 40. This is important, since pre-test counseling involves a specific patient, not the entire population.

We propose three possible strategies to address this dilemma. First, precision of the cfDNA screening tests could be improved to further reduce the false-positive rate and uninformative results and increase the detection rate. However, the test characteristics are already extremely good, fetal fraction is a relevant embedded quality metric, and it is difficult to imagine that there will be any major improvements in the near future. In addition, there are some biological limitations, such as mosaicism and vanishing twins, which cannot be eliminated [14, 28].

Second, uninformative test results require an alternative approach for further risk assessment than just invasive testing. In fact, if we assume a strategy of invasive procedure for all to resolve these cases, uninformative test results dilute the test performance of cfDNA screening, as they increase the screening positive rate by 1.2% [10]. Of note, in some studies, the rates of uninformative tests were shown to be even higher [11, 12, 29–31].

We acknowledge that an invasive testing by chromosomal microarray analysis (CMA) can detect pathogenic microdeletions/duplications, that are undetectable by current screening modalities, that account for a relevant burden of perinatal morbidity and mortality and for an overall greater prevalence than that accounted by common trisomies, particularly in younger women [32–34]. So that offering an invasive procedure by CMA would have wider benefit than just trisomies. We also acknowledge that there are chromosomal abnormalities in the group of uninformative tests other than just the common trisomies, such as triploidy, rare autosomal trisomies, and large segmental imbalances [18]. However, many of these cases would also have an identifiable fetal defect on ultrasound or would die in utero due to a life-limiting chromosomal abnormality such as triploidy [31, 35–37]. Therefore, it is widely accepted that cfDNA screening should be combined with a prior ultrasound examination and invasive testing if fetal defects are present and that an ultrasound scan re-evaluation should be performed after an uninformative result. In terms of screening for trisomy 21, one possible solution could be to take an additional blood sample for serum screening when blood is drawn for cfDNA testing, particularly when there are maternal characteristics associated with a no-call [30, 38, 39]. This sample could be stored, and the serum markers could be measured only in case of an uninformative test result. If cfDNA screening is embedded in a broader screening concept with an ultrasound examination at 11–13 weeks, the nuchal translucency thickness could be measured first followed by the blood test for cfDNA screening and the serum markers. If the cfDNA screening test remains inconclusive, the serum markers could be measured and combined with the nuchal translucency thickness to give a combined risk result. While the test performance of combined first trimester screening for trisomy 21 is not as good as cfDNA screening, the detection rate for the three trisomies is still around 95% with a false-positive rate of 2.5% [35]. An alternative solution could be to schedule cfDNA screening at 10 week gestation, leaving enough time for combined first trimester screening in the non-informative test cases. However, cfDNA testing so early in pregnancy increases the rate of uninformative testing and raises more questions regarding further management [38].

Third, cfDNA screening could be used as a second-line screening test that is preceded by a test that is broader in scope such as the first trimester combined screen. If the calculated risk is in the borderline risk group (i.e., 1:10 to 1:1000), the prevalence of the common aneuploidies is high enough to justify further testing using cfDNA. As combined first trimester screening focuses on more than just trisomies, the benefit for younger women is much greater. While the risk for trisomy 21, 18, and 13 is still low, the risk for structural anomalies is 1 in 40 or 2.5% regardless of the maternal age many of which can be identified on a late first trimester ultrasound. Inclusion of a first trimester ultrasound examination in the screening algorithm also has many additional benefits not the least of which is establishing the risk for preeclampsia. Regarding the detection of trisomy 21, we have shown that the overall testing performance of such a two-step approach is similar to cfDNA screening of the whole population [29]. In this study, we defined the first trimester borderline risk as 1:10 to 1:1,000, as it is currently done in Switzerland. This group of patients was then offered cfDNA testing. If the cfDNA test indicated high risk for a chromosomal abnormality or if it was not informative, invasive testing was offered. Invasive testing was also offered to those who combined first trimester screening result showed a risk > 1:10.

We acknowledge that our model has some limitations. First, we calculated the age-dependent risk for trisomy 21, 18, and 13 to determine the total number of affected fetuses using data from older studies conducted before the era of screening. The quality of evidence is lower for trisomy 18 and 13 than for trisomy 21, which could explain why the PPVs for cfDNA screening for trisomy 18 and 13 differ from the published data. However, the more likely reason is that the studies included in the meta-analysis have some bias, e.g., that they were not conducted as screening studies or that the ultrasound examination prior to cfDNA testing revealed an abnormality, leading to invasive testing instead of cfDNA screening.

To summarize, cfDNA screening is an excellent screening test for the most common trisomies. However, counseling prior to the screening test is essential. To this end, one should be aware of the prevalence of the screened-for diseases and the relevant test statistics stratified by maternal age.

Author contributions

T.E.: manuscript writing and editing. J.S.: manuscript writing and editing. N.P.: manuscript writing and editing. M.H.: manuscript writing and editing. F.R.G.: manuscript writing and editing. K.O.K.: manuscript writing and editing, data analysis, and project development.

Funding

None.

Data availability

No datasets were generated or analyzed during the current study.

Declarations

Conflict of interest

Karl Oliver Kagan, Tania Elger, and Jiri Sonek are editors of Archives of Gynecology and Obstetrics. Karl Oliver Kagan is the Editor-in-Chief. Francesca Romana Grati is full-time employee of Menarini Silicon Biosystems.

Ethical approval

This study is based on the German birth statistics in 2023. Therefore, an ethical approval is not necessary.

Consent for publication

All authors have agreed to publish this manuscript in Archives of Gynecology and Obstetrics.