Commentary: Expanded carrier screening: how much is too much?

C.S. Lewis once wrote, “What you see and what you hear depends a good deal on where you are standing.”¹ Such is the case with the introduction of expanded carrier screening to clinical care. To some, it is a marvelous opportunity to identify (or in the case of preconception screening prevent) the majority of pregnancies with Mendelian disorders. To others, it is a time consuming, expensive new responsibility that has been added to an already busy obstetrical practice without much consideration of how to implement it. Both authors of this commentary are obstetrician/gynecologists and clinical geneticists and as such have a unique perch from which to observe both vantage points.

It used to be so simple. Reproductive carrier screening programs began in the early 1970s with the discovery of the underlying biochemical cause of Tay–Sachs disease. Over the next 5 years, 50 programs were developed in Jewish communities. Several factors contributed to this rapid uptake and acceptance including the high rate of carriers in the Jewish populations, development of community-based screening centers, the severe nature of the disease, advocacy on the part of clinicians who were deeply involved with at-risk families, and the existence within the Jewish community of organizations willing to help coordinate early testing programs.² Tay–Sachs screening served as the model for other community and ethnic-based screening programs such as that for thalassemia in the Mediterranean region. These programs were successful in nearly eradicating well-described and severe disorders in educated and committed ethnic groups.³

Similar programs were not successful in other at-risk populations. Although prenatal testing was not yet available, carrier screening for sickle cell disease was begun in the 1970s in US African American communities but was ultimately viewed as a failure. There are likely many contributors to this outcome, including racial tensions, discriminatory insurance practices directed at sickle cell disease carriers, testing procedures that failed to distinguish between carriers and those with the disease, and problematic screening procedures. Some programs screened school-age children under legislative mandates. Educational, counseling, and follow-up procedures were often weak and sometimes absent; and many screening programs were initiated without community involvement or support.^4,5

Over time as the molecular underpinnings of additional disorders were discovered, prenatal screening expanded to other populations. This expansion began in 1997 (ref. ⁶) shortly after the identification of the gene for cystic fibrosis (CF) when a National Institutes of Health (NIH) consensus panel recommended that CF carrier testing should be offered to couples currently planning a pregnancy or those seeking prenatal testing but did not opine on which populations should be screened. Subsequently in 2001, the American College of Obstetricians and Gynecologists (ACOG) and the American College of Medical Genetics (ACMG) recommended that CF carrier screening be offered to non-Hispanic European Americans, for whom carrier detection was greater than 85% sensitive.⁷ Screening was to be “made available” to patients of other ethnicities in whom carrier detection had a lower detection rate.⁸ Since the majority of screening was being done as part of early prenatal care, this recommendation posed particular difficulties for the obstetrical community in identifying appropriate patients and in selectively offering them testing. An ACOG survey found that few of its members used ethnicity in deciding whether to offer CF carrier testing and a 2005 update endorsed offering screening to all couples regardless of race or ethnicity.⁹ In 2009, carrier screening for spinal muscular atrophy (SMA) was recommended by the ACMG for all racial and ethnic groups, but was not suggested by ACOG due to concerns about the complexities of the genetics, counseling on phenotype, and the variable carrier rate across ethnic groups.^10,11

With the introduction of next-generation sequencing into clinical care in the late 2000s, the ability to identify carriers for multiple disorders allowed broader population screening without the need to consider ethnicity.¹² Commercial laboratories shortly began offering “expanded” carrier screening panels for Mendelian diseases (often including 100 or more disorders) to physicians (infertility and obstetrical practices) as an alternative to ethnic-based screening approaches. As opposed to earlier screening approaches, this was not preceded by guidelines from national physician organizations, consensus with practitioners, or patient and community engagement. In an attempt to “catch up,” in March 2015 a Joint Statement of the Perinatal Quality Foundation, ACMG, ACOG, Society for Maternal–Fetal Medicine, and National Society of Genetic Counselors published “Points to Consider” when performing expanded carrier screening in reproductive medicine.¹² However, despite the increasing use of expanded carrier screening over the last 5 years and the development of laboratory guidelines for variant detection and analysis by ACMG,¹³ no general consensus exists on appropriate content of a panel or on the appropriate clinical implementation, still leaving clinicians with uncertainty.

Most recently, the American College of Obstetricians and Gynecologists issued a committee opinion that for the first time suggested that expanded carrier screening (ECS) was an appropriate alternative to standard testing and suggested that each individual practice should choose either standard or expanded screening for all of their patients.¹⁴ This left the obstetrical practitioner with two alternatives: (1) continue with the present approach, which screens based on ethnicity plus CF and SMA and will identify only approximately 30% of the disorders screened for on the typical expanded carrier panel; or (2) use an expanded carrier panel to screen for a broader and larger number of disorders. The standard approach also has an inherent racial and ethnic bias in carrier identification ranging from 35% in Northern European Caucasians, to only 6% of carriers of significant Mendelian disorders in East Asians.¹⁵

Despite this introduction of extended screening into many obstetrical practices, appropriate panel content remains unresolved. A recent study comparing 16 commercially available ECS offerings had panels ranging from 41 to 1792 conditions, with only 3 conditions screened by all panels.¹⁶ General guidelines for ECS panel design have been provided by professional societies, which stress the need to “maximize clinical utility over the quantity of conditions”,^12,13,17 but provide very few specific recommendations.¹⁶

Recent ACOG guidelines have suggested that disorders selected for inclusion should meet several consensus-determined criteria; one of which is a carrier frequency of 1 in 100 or greater (disease frequency of 1 in 40,000).¹⁴ This threshold was suggested to represent a reasonable balance between identifying carriers of the more common disorders and minimizing the anxiety and downstream costs associated with identifying excessive carriers. Since some available panels will identify over 75% of patients in specific ethnic groups as carriers, concern about the need for genetic counseling and partner testing is legitimate, especially given the limited resources available to accomplish this.

Two articles in this issue use different approaches to examine the impact of the 1 in 100 carrier threshold on the number of genes on a panel, the carrier detection rate in various ethnic groups, and the carrier couple rate. Guo et al.¹⁸ used population data from gnomAD and variant data from ClinVar to estimate rates of pathogenic and likely pathogenic variants identified when screening a representative US population as well as specific ethnicities. In this analysis, they demonstrated that a panel of 415 pathogenic or likely pathogenic genes would identify at least one variant in anywhere from 33% to 63% of individuals depending on ethnicity. The carrier couple rate would vary between 0.17% and 2.5%. Screening only genes with a carrier rate of 1 in 100 or higher would only include 40 genes and would identify approximately 3 of every 4 carrier couples. If, alternatively, ethnic-specific panels incorporating only the genes with an ethnic-specific carrier rate of 1 in 100 or higher in each population were used, then panels would only require 6 to 28 genes.¹⁸ However, this approach ignores the difficulty in determining an individual patient’s ethnicity and race.

In any population, a small majority of high-frequency genes account for the majority of carriers so that additional genes contribute proportionally less to the overall gene carrier rate and even less so to the identification of carrier couples. In other words, as genes infrequently associated with genetic disease in specific populations are added to expanded panels, the incremental yield for carrier identification decreases. Accordingly, raising the threshold to be more liberal in terms of genes screened would significantly increase the number of genes required on a pan-ethnic panel but would only modestly improve the identification of carrier couples.

Ben-Shachar et al.¹⁹ use a different approach to question the appropriate carrier rate for inclusion on a panel. Using data from a single lab’s experience of screening over 56,000 individuals using a 176-gene panel, they modeled the impact of the 1 in 100 threshold.¹⁹ Importantly, they point out the lack of clarity in the threshold recommendation since it doesn’t discern whether the 1 in 100 cutoff should be used for the entire pan-ethnic population or alternatively per ethnic group. They demonstrate that there are significant implications based on which assumption is used: 1-in-100 carrier rate in any ethnicity, 1-in-100 US-weighted carrier rate, and 1-in-100 carrier rate in all ethnicities. Based on the definition, a compliant panel could differ between 3 and 38 conditions and identified carrier rates could be reduced between 36% and 79% with identification of at-risk couples reduced between 11% and 92%. They also demonstrate that there is no natural cost–benefit ratio at this 1 in 100 cutoff.¹⁹ While the cost–benefit ratio grows as conditions become more rare, the relationship is roughly linear down to carrier rates as low as 1 in 1000 (ref. ¹⁹). Similar to Guo et al.,¹⁸ they also demonstrate that the at-risk couple rate saturation occurs at smaller panel sizes than the panel carrier rate. Using their 176-gene panel, 18 conditions would have a carrier rate of 1:100 or greater. A panel with these 18 conditions identifies 84.1% of at-risk couples. The addition of 73 conditions (carrier rate 1:500 or greater) to the panel increases the percentage of at-risk couples identified to 94.9%, while addition of the remaining 85 rare conditions to complete the 176-disease panel increases the percentage of at-risk couples identified by only 5.1% (ref. ¹⁹).

Both of these papers add important information to the discussion of panel content but still do not completely answer the question of “how much is too much.” There is general agreement that a basic panel should include disorders with a well-defined phenotype, with a detrimental effect on quality of life, that cause cognitive or physical impairment, that require surgical or medical intervention, and have an onset early in life.¹⁷ Additionally, screened conditions should be able to be diagnosed prenatally to optimize reproductive outcomes including opportunities for antenatal intervention to improve perinatal outcomes and early education of the parents about special care needs after birth. However, current expanded panels include additional conditions with the number of rare disorders continuing to increase as companies use the number of genes on a panel to compete in the market place. These additional disorders can have significant variation in their presentation, including variable expressivity, penetrance, and age of onset. Although some genetic variants on expanded panels have a relatively consistent phenotype, others are less clearly defined. Finally, for a portion of these disorders the precise carrier frequency as well as the proportion of condition-causing variants that can be detected may be unknown, making residual risk determination difficult and increasing the genetic counseling burden and potential patient anxiety.^20,21,22

In deciding the content of a panel perhaps we are asking the wrong stakeholders. In this modern era of medicine, as we strive to be more patient-centric, instead of asking medical professionals and professional societies to define what is an appropriate carrier screening panel, should we be asking our patients about their values and priorities with regard to genetic screening? Only a few studies have evaluated this issue, finding that the key motivations for those who choose screening are learning the condition of their fetus and preparing for the possibility of a child with a genetic condition.^{23,24,25,26,27} Additionally, a study by Gilmore et al.,²⁸ which surveyed 240 women planning a pregnancy who declined enrollment in a genomics-based carrier screening study, found that the most common reasons for declining the study were a lack of time or lack of interest in getting screening. However, a smaller number did indicate that they did not want to know the information or cause themselves anxiety. These data show that those who ultimately get or decline screening in a clinical setting are able to identify specific reasons for their decisions, suggesting that these are informed choices and they may also have opinions on which category of disorders they would elect to screen for.²⁵ At present this may be difficult, since compared with ancestry based high-risk groups, the general population is less familiar with genetic conditions and has less knowledge of or experience with families with an affected child.²⁹ However, as we endeavor to improve the overall health and well-being of our society, working toward this goal of informed decision making would be aspirational.

Finally, speaking as obstetricians, there really is no rational reason why such screening should occur during pregnancy. With most obstetrical visits lasting approximately 15 minutes, there is limited time for obstetric practitioners to explain pregnancy-related issues (e.g., risk of preterm birth, diet, weight gain, etc.) leaving little opportunity to appropriately explore carrier screening. Moreover, obstetricians are not prepared, nor expected, to counsel patients on the intricacies of many of these disorders. Screening during pregnancy also means that parents have little time to make decisions and limits reproductive options for carrier couples, such as preimplantation genetic diagnosis. In addition, for some conditions, family studies to better understand the significance of certain variants cannot be performed quickly thereby compromising the interpretative ability for carriers of these variants. Preconception carrier screening would avoid most of these issues and minimize much of the controversy regarding the conditions on the screening panels. There are many excuses given for why this might not work; but no organized studies have evaluated this option. It is time to do this. Perhaps we should reevaluate the most successful carrier screening yet (Tay–Sachs) where the majority of testing was done preconception by a community that had been educated in the risks of Mendelian disorders and the value of avoiding them.