Introduction

Constitutional genetic testing currently uses a wide range of standard-of-care (SOC) methods to detect possible pathogenic variants, extending from karyotyping to chromosomal microarray (CMA) and whole-exome sequencing (WES).1–3 In combination, these methods allow to detect all kinds of known variants. However, while the characterisation of chromosomal deletions and sequence variations is relatively uncomplicated, SOC methods usually fall short of resolving breakpoints of copy-number neutral (balanced) aberrations and copy-number gains. This is of disadvantage when the classification of a detected variant relies on its complete resolution. It particularly affects patients where clinical data are insufficient, for example, in prenatal diagnostics or pertaining diseases with an unspecific phenotype. Often, for lack of other possibilities, the identification of such variants is followed up by segregation analysis of close relatives to tentatively reach a clinical assessment.4 However, this approach is limited to variants that, if pathogenic, cause an observable phenotype and require informative carriers of the variant. Hence, this excludes a variety of scenarios (eg, most cases of egg or sperm donation, unavailable relatives, late-onset or low-penetrance or recessive, especially X linked recessive, diseases) which can only be addressed by a precise topological characterisation of the structural variant (SV). A promising, efficient way to achieve this task seems to be by optical genome mapping (OGM). It has been shown in singular cases to comprehensively characterise SVs and to eliminate ambiguities remaining in conventional diagnostics.5–7 Here, we present a workable strategy that integrates OGM in the pipeline of routine diagnostic analysis. Based on a sample cohort from a routine genetic testing background, we estimate the extent of clinical cases that may profit from the additional data provided by OGM. Furthermore, we showcase examples of the scenarios in which it is successfully applied. With emphasis on prenatal cases, these examples illustrate the utility of OGM in this particular setting in which its diagnostic value has been less well documented so far.8–12

Materials and methodsSamples

The prenatal and postnatal samples surveyed in this retrospective study were recruited from the patient cohort routinely analysed at the Institute of Medical Genetics during a 1-year period in 2023. All genetic tests were performed as requested by the patients’ supervising physicians to find the cause of a suspected genetic disorder by WES, CMA or karyotyping. For the exemplified cases, investigated material included blood, amniotic fluid or chorionic villi specimen (CVS) collected at the Division of Obstetrics and Feto-Maternal Medicine of the Vienna General Hospital, at the Institute of Medical Genetics or at a private local provider of specialised services in fetal medicine, FetoMed.

Optical genome mapping

Cultured amniotic or chorionic cells were pelleted by centrifugation and resuspended in 2 mL of cell culture medium (Amniogrow plus medium, Cytogen, Sinn, Germany, or Chang medium D, Fujifilm, Minato, Japan), followed by live cell counting on an automated cell counter (TC20, Bio-Rad, California, USA). For ultra-high-molecular-weight (UHMW) DNA extraction, between 1.0 and 1.5×106 viable cells were either processed directly or pelleted and processed after storage at −80°C according to the manufacturer’s ‘SP Fresh-’ or ‘SP Frozen Cell Pellet DNA Isolation Protocol’, respectively, using the SP Blood and Cell Culture DNA Isolation Kit (Bionano Genomics, California, USA). The same kit was used for the preparation of frozen peripheral blood samples using 1.0–1.5×106 of viable cells and following the manufacturer’s ‘SP Frozen Human Blood DNA Isolation Protocol’.

As described previously, UHMW DNA was labelled with DLE-1 enzyme (Bionano Genomics) and imaged on a Saphyr instrument (Bionano Genomics).13 Data collection per sample was aimed at >300 Gbp of UHMW DNA (including molecules of 150 kbp or larger). In analyses meeting the quality requirements, this corresponds to >80× coverage of the genomic reference and to >50× coverage of the genomic assembly, as recommended for the subsequent whole-genome de novo analysis. De novo analysis, as well as SV calling, was performed with the Bionano Access Suite using the respective software modules.

OGM data analysis also included automated calls as implemented by the manufacturer’s software (the SV calling algorithm for identification of breakpoints and the CNV calling algorithm for identification of gains and losses, both using default settings, without filtering by frequency). However, we primarily relied on direct visual assessment of data generated in the region of interest by an experienced human evaluator. This assessment was based on all OGM maps of the region, as well as the information on molecule quantity provided by the ‘CNV-Track’ of the Access Suite.

Routine diagnostic methods

The routine diagnostic methods used to identify or confirm the SVs described in this study (CMA; chromosomal analysis and FISH; PCR and sequencing; multiplex ligation-dependent probe amplification (MLPA)), as well as the corresponding DNA isolation protocols, are described in the online supplemental methods. WES analysis was performed as described previously.3

Criteria for follow-up analysis

Follow-up analyses were considered when SVs with a potential clinical relevance were detected or suspected by SOC methods. The follow-up analyses performed were aimed at: (1) distinguishing true SVs from possible detection artefacts, (2) assessing the SV’s clinical relevance or (3) estimating the risk of its reoccurrence in the patient’s offspring.

Results

A total of 3021 patients were referred to the Institute of Medical Genetics for genetic analysis in 2023 (online supplemental figure S1). Follow-up analyses for potentially pathogenic abnormalities were requested by the patients’ physicians according to the criteria detailed above. In the majority of cases (n=41), these analyses consisted of a secondary SOC method. This included the precise demarcation of gains or losses of genetic material by CMA, and the localisation of gained or translocated material, as well as inversions, by cytogenetic techniques. For a small number of SVs, present in seven patient cases (P1–P7, table 1), conventional follow-up approaches were unfeasible, requiring OGM instead.

Table 1

Overview of cases followed up by optical genome mapping (OGM), including the respective findings and their final classification on OGM analysis.

The common blueprint for addressing these cases with OGM is outlined in figure 1. Notably, despite presenting highly individual situations with different genetic and clinical constellations involved, the respective cases all qualified to OGM for the same reason: to interrogate the disruption of genes by the observed or suspected SV of interest, either by determining the location and orientation of duplicated material or by characterising the breakpoints in case of copy-number neutral aberrations. OGM was useful in all seven cases, as the data obtained from it either allowed direct interpretation of the corresponding SV or enabled a region-specific downstream analysis achieving that goal.

Figure 1

Diagnostic pipeline for genetic examinations. The schematic shows the conditions under which optical genome mapping (OGM) is used as a follow-up analysis (red box; green box, no follow-up by OGM).

Three of the cases, P1–P3, are discussed in more detail, as they are deemed to represent the most instructive examples for leveraging OGM in clinical practice. A further case, P4, describes the localisation of a breakpoint of a translocation in a child suffering from blepharophimosis. Based on the specific clinical picture and the cytogenetic location of the translocation, a disruption of the gene FOXL2 was queried by OGM, showing that the gene’s coding and promoter regions were unaffected by the breakpoint. Interestingly, however, the location was determined in close proximity to FOXL2 (ca 300 kbp upstream), which does not completely rule out impairment of regulatory elements of the gene.

Cases P5–P7 are examples of observed aberrations that, although not of particular scientific interest, could lead to ambiguous genetic reports that unnecessarily distress the patients. P5 and P6 illustrate cases of SVs suspected by fetal karyotyping, which were shown by OGM to be likely artefacts. In case P7, a duplication of parts of the gene CAMSAP1, which is associated with autosomal recessive cortical dysplasia, was followed up by OGM. This served to show that the duplication did not disrupt the gene and was hence classified as likely benign.

P1: localisation of additional genetic material

A woman pregnant with a male fetus was referred for amniocentesis at 21 weeks of gestational age due to a suspected fetal infection. An arbitrary genetic screening of the fetus, consisting of karyotyping and CMA, was performed. Karyotyping was unremarkable; however, CMA revealed two X-chromosomal duplications. One encompassed a ca 193 kbp region in Xp22.31, and the other encompassed a ca 654 kbp region in Xp21.2p21.1, hereupon referred to as duplications 1 and 2, respectively (table 1). Duplication 1 involved the terminal exon 11 of the STS gene and parts of its 3’-flanking region, potentially leading to a disruption of the STS gene (transcript ENST00000674429.1). Disruptive mutations leading to loss of function of STS are causative for X linked recessive ichthyosis.14

Duplication 2 involved terminal exons 56–79 of the DMD gene (transcript ENST00000357033.9). Duplications inserted into the DMD gene, as well as deletions of the DMD gene, are known to lead to muscular dystrophy-type Becker or Duchenne in male carriers, while females are often unaffected.15–17 Although to our knowledge, a duplication of exons 56–79 has not yet been reported, similar duplications including terminal exons have already been classified as pathogenic.18–20 Consequently, both duplications held potentially severe implications for the pregnancy. MLPA confirmed both duplications of maternal origin. However, considering the X linked recessive effect of pathogenic DMD and STS mutations, we were unable to immediately recruit further male relatives for carrier testing.

Hence, we aimed to localise the duplicated material by OGM. Prior to execution, the likelihood of success was determined by reviewing OGM data from other samples. Both regions in question contained sufficient OGM label sites for identifying the duplicated material, irrespective of the site of insertion (38 labels at duplication 1; 158 labels at duplication 2), exceeded the software’s minimum size for detection (30 kbp) and showed no propensity to assembly artefacts. Consequently, OGM was performed on fetal DNA. The obtained OGM data suggested a complex rearrangement on the p-arm of the X chromosome (figure 2; online supplemental figure S2). This is in contrast to what was anticipated based on available data from systematic surveys of duplications.21 The aberration involves material from duplication 2 (figure 2a, ‘segment D’) and duplication 1 (‘segment B’) in consecutive order, inserted directly ca 200 kbp downstream of the STS gene in a region not containing any known genes. However, the automatic SV caller could only identify one breakpoint and accordingly labelled the SV a 24.2 Mbp sized duplication (online supplemental table S1). On further inspection, we found the other breakpoint called with a confidence too low (0.02) for the SV caller’s standard settings. The CNV caller only detected the duplication of the ca 654 kbp sized region in Xp21.2p21.1. The ca 193 kbp sized duplication was not automatically called and could be barely spotted by visual inspection of the CNV track (online supplemental figure S2). Also, its size is below the recommended threshold of 500 kbp for CNV calls.

Figure 2

Complex rearrangement of the p-arm of the X chromosome. (a) Schematic of an unremarkable (above) and aberrant (below) p-arm of an X chromosome (the genes STS and DMD are shown in green boxes, with their orientation indicated by black arrows). The p-arm is subdivided into five segments for clarity (A–E). The rearrangement leads to the duplication of segments B and D (indicated as B* and D*), which include the terminal segments of the genes STS and DMD, respectively. (b) Breakpoints of segments B/D* and D*/B*. The junction of B/D* consists of only one homologous base. The junction of D*/B* contains a 714 bp-long homologous region (coordinates according to GRCh38, orientation of the genes STS and DMD indicated by dashed black arrows).

To identify the exact breakpoints, long-range PCR followed by Sanger sequencing of the amplicons was performed (figure 2b). Breakpoint B/D* featured one homologous base, indicating double-strand repair by non-homologous end joining. At breakpoint D*/B*, we found a 714 bp LINE (Long Interspersed Nuclear Element) region with >99% homology, fusing STS intron 10 with DMD intron 55. Both types of recombination are described as common mechanisms for structural aberrations of the DMD gene.7 22

Evidence for an intact copy of DMD by OGM, together with the localisation of the duplicated material outside gene-containing regions, eventually led us to assess both duplications as likely benign and ensured rapid and reliable counselling of the family. Later, this assessment was confirmed when a healthy male relative could be recruited and identified as a carrier of the aberration by MLPA. The mother continued the pregnancy and eventually gave birth to a healthy boy.

P2: breakpoint clarification of a balanced SV

A woman with a sonographically normal pregnancy was referred for CVS at 11 weeks of gestational age to examine the fetal carrier status of a disorder not pertaining to this study. Routinely performed karyotyping detected an SV on chromosome 7. It affected the region q11.2q22 and was suspected to be a paracentric inversion of the q-arm (table 1, figure 3a), as no CNVs were detected in this region by CMA. While inversions detected prenatally are often considered benign if inherited,23 similar SVs on chromosome 7 are rarely reported and have been described as causes of haematological malignancies.24 25 As carrier analysis is not reliable for assessing genetic predispositions to tumours, we considered characterising the inversion via OGM. We performed a prior review of OGM data of other samples of the cytobands q11.2 and q22 of chromosome 7, where the breakpoints were expected. Similar to P1, this was done to exclude that these regions were prone to low coverage or assembly artefacts. This review supported an informative outcome of OGM in advance.

Figure 3

Paracentric inversion of the q-arm of chromosome 7. (a) Ideogram of chromosome 7 (inversion indicated by pink arrow; target regions of FISH probes in red and green). (b) Maternal chromosome 7 by karyotyping with GTG banding. (c) Maternal chromosome 7 according to FISH analysis (RP11-213E22—green; RP11-577D9—orange; +/+, unremarkable chromosome 7; inv, chromosome 7 bearing the inversion). (d) Optical genome mapping (OGM) data of the breakpoint regions of the inversion. Labels on the reference (green track) are aligned via vertical matchlines (grey lines) to labels on two optical maps (blue tracks; above: unremarkable map; below: map showing inversion). The inversion breakpoints are shown to be located in areas of high homology (grey areas; inversion indicated by dotted lines; coordinates according to GRCh38).

The OGM analysis yielded a map that depicted an inversion involving cytobands 7q11.23 and 7q22.1 (online supplemental figure S3). Both breakpoints mapped to large regions containing homologous segmental duplications of ca 0.75 and 3.25 Mbp (figure 3d). Hence, the presence of the inversion could be inferred by a human evaluator despite the segmental duplications’ sizes and considerably lower OGM quality metrics than recommended (poor effective coverage of the reference due to flow cell leakage with 25× instead of recommended >80×; all other quality metrics were satisfactory). No automated identification of the inversion was made likely due to the low coverage (online supplemental table S1).

Parental karyotyping revealed that the fetus had inherited the inversion from its mother (figure 3b). Consequently, OGM was performed on maternal DNA, which yielded data with adequate quality metrics, confirmed the inversion’s breakpoints and allowed the correct automated identification of the variant (online supplemental table S1 and figure S3). Altogether, the breakpoint intervals could be sufficiently delimited to rule out the disruption of known genes. Although this does not completely rule out an association of the inversion with haematological malignancies, we could not substantiate the suspicion and consequently classified the variant as likely benign. The pregnancy was continued and eventually a healthy girl was born.

For confirmation, FISH analysis was conducted on fetal and maternal samples, as the homologous regions of the segmental duplications prohibited confirmation by PCR. Consistent with the OGM data, FISH analysis of the maternal and fetal samples revealed a normal chromosome 7 with the red and green probes located proximal and distal to the q-arm, respectively, and an inverted signal pattern for the aberrant chromosome 7 (figure 3c).

P3: confirmation of a cryptic SV

A newborn girl with suspected hereditary pseudohypoaldosteronism was analysed by WES, which yielded no coverage of exon 13 of the SCNN1B gene. As nonsense mutations in exon 13 of SCNN1B have been reported as pathogenic, a cryptic SV was suspected to be responsible for the consanguineous family’s pseudohypoaldosteronism, which also affected the newborn’s father.26 However, SVs with breakpoints within exons are considered rare and consequently unlikely.27 As at that time, OGM was unavailable to us, attempts were made to confirm the hypothesised SV by PCR of the girl’s DNA, which remained unsuccessful. When several years later, the child’s clinically unremarkable mother was again pregnant and presented to us, OGM had become available. It was performed after CVS at 14 weeks of gestational age. As at that point, the presence of an SV was hypothetical only, this case is not an example of the application of the analysis pipeline described above.

Fetal OGM data yielded four maps for the p-arm of chromosome 16, showing label patterns corresponding to a homozygous, paracentric inversion (online supplemental figure S4). The breakpoints were located within the regions 16p13.13 and 16p12.2, both by the automatic SV caller (labelling them as two separate ‘intrachromosomal fusions’, instead of an inversion) and human evaluator (online supplemental table S1, figure 4a,b). The distal breakpoint interval overlapped with SCNN1B exon 13 (transcript ENST00000343070.7), which made the disruption of the gene very likely. Subsequent parental OGM analyses detected the inversion in the heterozygous state in the fetus’ mother and in the homozygous state in the fetus’ father, consistent with his phenotype. For confirmation of the breakpoint, a PCR was designed based on the breakpoint intervals identified by OGM. This yielded a ca 9 kbp PCR amplicon, which was analysed by long-read sequencing. This eventually determined the exact breakpoint within exon 13 (figure 4b,c), confirming the disruptive effect of the inversion and its pathogenicity. The pregnancy was continued; however, further information is unavailable.

Figure 4

Paracentric inversion of the p-arm of chromosome 16. (a) (Left) Ideogram of chromosome 16 (inversion indicated by pink arrow; target regions of FISH probes in red and green). (Middle) Chromosome 16 of the mother and fetus by FISH analysis (RP11-358H10—green and RP11-241M19—orange). (Right) Maternal chromosome 16 by karyotyping with GTG banding (+/+, unremarkable chromosome 16; inv, chromosome 16 believed to carry the inversion). (b) Schematic of the inversion (orientation of the SCNN1B gene indicated by the black arrow). The p-arm is subdivided into three segments for clarity (A–C). The junction of B/C lies within E13 of SCNN1B; hence, the inversion of segment B leads to the disruption of the gene. (c) Breakpoint of segment A/B, with two homologous bases at the breakpoint (coordinates according to GRCh38).

Despite its considerable size of 11 Mbp, the inversion is difficult to identify by GTG banding at a banding level of 550–700 (figure 4a). However, knowledge of the breakpoint allowed its detection by FISH analysis. Consistent with the OGM findings, FISH analysis of the fetal chromosomes revealed two aberrant chromosome 16, while FISH analysis of the maternal chromosomes revealed a normal chromosome 16 as well as the inverted signal distribution of the second, aberrant chromosome 16.

With the exception of FISH analysis, the inversion is outside the detection scope of conventional methods. Therefore, it is important to highlight that the knowledge of the breakpoint also resulted in the development of a customised PCR test, allowing convenient carrier testing of additional family members and discrimination between heterozygous and homozygous carriers (online supplemental figure S4).

Discussion

Until now, OGM has not seen significant adoption in constitutional genetic diagnostics. The reason for this might be that despite a considerable number of reports on its successful application, potential users have not identified a distinctive advantage over SOC methods that justifies its adoption.8–11 28 In contrast, the benefits of the technology, for example in haemato-oncology, are more widely recognised, and its implementation is more advanced.29

Here, we describe the use of OGM to provide information not available by SOC methods, specifically, the locations of breakpoints of SVs, which remain ambiguous without that information. Furthermore, we deem it a low-threshold method to follow-up on uncertain results that could unnecessarily distress the patients (see P5–P7). Based on the sample pool available to us by our own routine genetic testing, we explored a strategy that renders OGM a meaningful and easily applicable tool in constitutional diagnostic practice. This approach focuses on OGM’s strength to complement existing SOC methods with the goal to provide clinically actionable reports, which has been successfully attempted in individual cases.12

Overall, the clinical scenarios that profit from OGM, when using it in the fashion described, are defined by the following criteria: precedent detection of an SV that is suspected but not certain to affect genes with a role in disease development; lack of an informative family history; and unfeasibility of carrier analysis due to, for example, unavailable relatives, egg or sperm donations and suspected late-onset, low-penetrance or recessive, especially X linked recessive, diseases. The detection of such uninterpretable variants usually leads to a dilemma in reporting; especially in the prenatal setting, reporting offers no guidance for the parents and bears the risk of unfavourable consequences for the pregnancy.30 In the prenatal cases described, we were able to provide actionable reports that facilitated the parents’ decision-making and were perceived reassuring for the pregnancy (P1–P3 and P5–P7). In all cases, the pregnancies were continued. The adverse consequences of uninterpretable variants might be less severe in a postnatal setting. However, it might lead to misdiagnosis, incorrect treatment and misdirected carrier testing. Although serious, in our experience, these cases occur rarely.

To date, the majority of available studies focus on OGM as a possible alternative to first-line methods in routine genetic testing, mainly because it is argued that it combines the abilities of multiple detection technologies.9 11 In contrast, we suggest prioritising OGM for a well-defined set of questions rather than as a first-line method. The continuation of the use of SOC methods benefits from the experience and databases associated with them. Additionally, the development of similar repositories for OGM could profit from data of routine samples analysed by several methods. Since our approach focuses on an a priori known region, it also allows to evaluate the resulting OGM data manually, thereby avoiding automated SV calling as a possible source of errors.31 32 In the cases described here, all SVs with sufficient coverage were identified by the SV caller. However, one of two breakpoints of P1 was missed. The CNV caller only detected one of three CNVs, as it is designed for variants >500 kbp (online supplemental table S1). Overall, we deem avoiding automatic calling a necessary precaution until more evidence for the robustness of OGM data and associated variant callers are available.8 11 32

The resolution achieved by OGM was often sufficient for the clinical interpretation of the analysed SVs. When this was not the case, the data provided by OGM enabled the use of more targeted methods (ie, PCR and subsequent sequencing). It remains to be investigated whether whole-genome sequencing with the emerging long-read sequencing technologies may be used for a purpose similar to OGM, without the shortcomings of limited resolution.33

For routine clinical diagnostic purposes, we advocate the use of OGM only for the limited scope outlined above, resulting in a proportionally small number of cases that require OGM. This could facilitate the method’s entry to diagnostic laboratories, especially since it is currently not a reasonable high-throughput technique. Finally, the approach presented here is capable to harness the distinctive advantages of OGM, with appreciable contributions to an enhanced quality of constitutional genetic diagnostics.