Diagnostic yield of pediatric and prenatal exome sequencing in a diverse population

Individuals studied and demographics

A total of 845 patients, comprising 529 pediatric and 316 prenatal probands, were enrolled (Table 1). The pediatric group included more males (290/529, 54.8%) than females (239/529, 45.2%), which was significantly different from a 50:50 distribution (χ2 = 4.92, P = 0.027). The prenatal group also included more male (171/316, 54.1%) than female fetuses (145/316, 45.9%); a difference that was not significant (χ2 = 2.14, P = 0.144). In all, 86/529 (16.3%) of pediatric patients were younger than one year of age and 405/529 (76.6%) were ten years of age or younger at enrollment (Supplementary Table 1), reflecting the early onset of clinical findings associated with neurodevelopmental disorders and the enrollment categories used in this study. In the prenatal cohort, the mean gestational age at enrollment was 23.5 weeks. Overall, 135/316 (42.7%) of pregnancies were terminated and 122 (38.0%) resulted in a living child that survived the neonatal period (Supplementary Table 2). For the remaining pregnancies, 19 (6.4%) resulted in a stillbirth at ≥20 weeks gestation, 7 (2.2%) in a miscarriage at <20 weeks, and 33 (10.5%) in a neonatal death. Of the patients choosing pregnancy termination, 92.6% received ES results after the conclusion of the pregnancy.

Table 1 Participant demographics in the program in prenatal and pediatric genomic sequencing (P3EGS) study.

As proband ages varied at the time of enrollment, we tabulated mean parental ages at the time of the proband’s conception. The median maternal age at the time of proband conception was 28.2 years for pediatric and 33.1 years for prenatal patients. Median paternal age at the time of conception was 32.2 years for pediatric and 35.0 years for prenatal patients (Table 1). Overall, 554/845 (65.6%) pediatric and prenatal patients had at least one parent who self-identified as URM, 155/845 (18.3%) were non-URM (i.e., both parents white/European), and in 136/845 (16.1%) the race/ethnicity was unknown or missing for both parents or one parent while the other self-reported white/European (Table 1; Fig. 1A–D). There were more URM families among the pediatric patients (397/463, 85.7%) compared to the prenatal cohort (157/246, 63.8%). The largest race/ethnicity group among the parents of pediatric cases was Hispanic/Latino (43.1% of mothers, 38.4% of fathers), followed by white/European (18.1% of mothers, 19.1% of fathers). The largest race/ethnicity group among the parents of the prenatal cases was white/European (36.4% of mothers, 38.4% of fathers), followed by Hispanic/Latino (15.5% of mothers, 16.7% of fathers; Table 1).

Fig. 1: Distribution of ancestry in 845 patients enrolled in the Program in Prenatal and Prenatal Genomic Sequencing (P3EGS) study.figure 1

Each chart shows the distribution of ancestry according to the arm of the study (Pediatric and Prenatal) and the sex of the participant. Ancestries depicted are American Indian, Native American (blue), Alaskan Native (Asian (orange), White/European (light blue), Middle Eastern/North African (green), Hispanic/Latino or Latina (dark blue), More than one race/ethnicity (brown), Unknown, none of the above (gray). A Pediatric patients, maternal ancestry. B Pediatric patients, paternal ancestry. C Prenatal patients, maternal ancestry. D Prenatal patients, paternal ancestry.

In the entire cohort, 505/845 patients (59.8%) had public insurance through Medi-Cal or had no insurance coverage. Public insurance was more frequent for pediatric (432/529, 81.7%) compared to prenatal patients (73/316, 23.1%). In addition, 148/529 (28.0%) pediatric patients and 74/316 (23.4%) prenatal patients and mothers were domiciled at an address and zip code that defined a medically underserved area/population (MUA/P) and 144/529 (27.2%) pediatric patients and 78/316 (24.7%) prenatal patients and mothers were domiciled at an address and zip code defined as a health provider shortage area (HPSA). In total, 457/529 (86.4%) pediatric patients and 146/316 (46.2%) prenatal patients met at least one of the US categories.

Diagnostic categories

For the pediatric cohort, the most common diagnostic category was MCA with ID (252/529, 47.6%), followed by MCA without ID (105/529, 19.8%) and ID only (48/529, 9.1%; Table 2). Enrollment in the categories of metabolic disease without ID, epilepsy without ID and neurodegenerative disease/cerebral palsy (CP) with or without ID were relatively low and this may reflect consultation patterns, including a greater emphasis on referrals for genetic testing for patients with ID, in addition to the relative frequencies of these phenotypes and the availabilty of diagnostic panels for genetic testing in addition to ES. The prenatal cohort included 125/316 pregnancies with a single sonographic anomaly (39.6%) and 191/316 with multiple structural anomalies (60.4%, Table 2).

Table 2 Exome sequencing results of P3EGS patients (based on pediatric inclusion criteria and prenatal phenotypes with ultrasound).Exome sequencing

Trio ES was performed in 583 (69.0%) cases, 122 (14.4%) patients were sequenced as duos, most commonly including the mother and child or fetus, and 109 (12.9%) patients were sequenced with a proband first approach (Supplementary Tables 3 and 4). Trio ES was more frequent in prenatal (257/316, 81.3%) compared to pediatric (326/529, 61.6%) patients and there was a higher prevalence of duo sequencing in pediatric (108/529, 20.4%) compared to prenatal (14/316, 4.4%) patients, reflecting a greater availability of fathers for prenatal versus pediatric patients in our study.

Diagnostic yield—pediatric versus prenatal

Overall, 201/845 (23.8%) patients received a positive (defined here as definitive positive or probable positive) diagnosis (Table 3). Of these, 137 (68.2%) had a pathogenic (P) or likely pathogenic (LP) variant in a gene with autosomal dominant (AD) inheritance, 40 (19.9%) had two P or LP variants or one P/LP variant and one variant of unknown significance (VUS) in a gene with autosomal recessive (AR) inheritance, and 24 (11.9%) had a P or LP variant in a gene with X-linked (XL) inheritance. In addition, 94/845 (11.1%) patients had inconclusive results (most had at least one VUS), and 550/845 (65.1%) received a negative result. The overwhelming majority of VUSs in our study (95%) were due to variant uncertainty, as opposed to gene uncertainty. The diagnostic yield was higher in pediatric patients, with 141/529 (26.7%) receiving a positive diagnosis compared to 60/316 (19.0%) in the prenatal group (χ2 = 6.41, P = 0.01). The frequency of inconclusive results was also significantly greater in pediatric (74/529, 14.0%) compared to prenatal patients (20/316, 6.3%; χ2 = 11.74, P = 0.0006). Of interest, the greater diagnostic yield in the pediatric cases was largely attributable to pathogenic variants in genes with AD inheritance (98/529 = 18.5% for pediatric cases versus 39/316 = 12.3% for the prenatal cases), including variants that were de novo, inherited from a parent, or of unknown segregation. The proportion of positive versus inconclusive cases also differed by mode of inheritance. For the pediatric cases, 69.5% of the positive results were for variants with AD inheritance versus 17.7% for AR inheritance, a ratio of 3.9:1. However, among the inconclusive results in pediatric patients, 40.5% were in variants with AD inheritance, while 44.6% were in AR genes, a ratio close to 1:1. This difference in mode of inheritance between positive and inconclusive results was highly statistically significant (χ2 = 19.95, P = 8.0 × 10−6). This trend was not observed among the prenatal group and for positive results, the ratio of AD to AR inheritance was 2.6:1 and among the inconclusive cases, the ratio was 2.0:1.

Table 3 Diagnostic yield and inconclusive rate by mode of inheritance.Diagnostic yield by indication

There was no difference in the diagnostic yield by indication in the pediatric patients (Table 2) and although diagnostic yield for isolated ID was lower (8/48, 16.7%) than for ID with multiple congenital anomalies (71/252, 28.2%), this difference was not significant (χ2 = 2.82, P = 0.09). In the prenatal cases, those with multiple sonographic abnormalities were more likely to have a positive result (45/191, 23.6%) compared to prenatal cases with a single structural anomaly (15/125, 12.0%; χ2 = 6.56, P = 0.01). Interestingly, while the diagnostic yield of pediatric cases was overall higher than of prenatal, in pediatric and prenatal patients with multiple anomalies as a clinical indication for ES, the diagnostic yield was similar; with 100/357 (28.9%) positive pediatric cases compared to 45/191 (23.6%) in the prenatal group (χ2 = 1.22, P = 0.27).

Diagnostic yield by number of family members sequenced

Among the pediatric cases, there was a clear decrease in definitive positive diagnoses in the duo and proband first families (9.3% and 11.8%, respectively) compared to the quad and trio families (26.3% and 22.1% respectively; Supplementary Table 3; Fig. 2A–C). This was primarily observed with patients receiving P or LP results in genes with AD inheritance. In contrast, there was a higher rate of probable positive diagnoses among the duo families (15.7%), compared to the trio (6.7%) or quad families (0%; Supplementary Table 3). The higher rate of probable positive diagnoses was primarily due to variants in genes with AD inheritance when parental segregation of the variant could not be determined. There were fewer definitive positive and probable diagnoses in the ‘proband first’ families, with a yield of 19.7%, although this difference was not significantly different from trios (χ2 = 2.58, P = 0.108). In patients with variants in genes with AR inheritance, there was no overall difference in diagnostic yield by number of parents sequenced for both homozygotes and compound heterozygotes. Among the pediatric families, there was a higher rate of de novo, AD definitive and probable positive results in trios (17.5%) compared to the quad families (5.3%), although this increase was not significant (χ2 = 1.92, P = 0.17). There was a significant excess of positive, inherited AD variants (15.7%) in quads compared to trios (2.5%) (χ2 = 10.34, P = 0.0013).

Fig. 2: Diagnostic yield by sequencing approach in 845 patients enrolled in the Program in Prenatal and Prenatal Genomic Sequencing (P3EGS) study.figure 2

The percentages of definitive positive (orange), probable positive (yellow), inconclusive (green) and negative (brown) results are shown for proband first, duo and trio sequencing approaches. There was no statistically significant difference in diagnostic yield with any sequencing approach. A Diagnostic yield with ‘proband first’ sequencing in pediatric patients. B Diagnostic yield with duo sequencing in pediatric patients. C Diagnostic yield with trio sequencing in pediatric patients. D Diagnostic yield with ‘proband first’ sequencing in prenatal patients. E Diagnostic yield with duo sequencing in prenatal patients. F Diagnostic yield with trio sequencing in prenatal patients.

In the prenatal families, there were no statistically significant differences in diagnostic yield based on the number of parents sequenced (16.7% for quads, 18.7% for trios, 14.3% for duos and 24.2% for proband first; Supplementary Table 4; Fig. 2D–F); however, the number of non-trio cases was small (59 total, or 18.7%), limiting power for comparisons. In the prenatal families, 39/60 (65.0%) of the definitive positive and probable positive results involved genes with AD inheritance, and 34/39 (87.2%) of these variants were de novo. In addition, 15/60 (25%) of definitive positive and probable positive results involved AR genes, while 6/60 (10.0%) involved XL genes.

Diagnostic yield by age and sex of proband and prenatal outcome

Diagnostic yield for pediatric probands did not differ significantly by age of proband (Supplementary Table 1). Comparing positive diagnoses between affected male and female individuals (Supplementary Table 3), we noted a significantly higher diagnostic yield in female probands (77/239, 32.2%) compared to males (64/290, 22.1%) in pediatric families (χ2 = 6.41, P = 0.011). This difference was not observed in the prenatal families (Supplementary Table 4), with a female diagnostic yield of 26/145 (17.9%) and a male diagnostic yield of 34/171 (19.9%). Further examination of the pediatric probands by indication revealed that the sex difference was fully explained by probands with ID, with a diagnostic yield of 33.5% (56/167) for females with ID versus 20.8% (45/216) in males with ID, a statistically significant difference (χ2 = 7.82, P = 0.005; data not shown). There was no different in diagnostic yield in females without ID (28.2%; 20/71) versus males (26.7%;20/75, χ2 = 0.04, P = 0.84; data not shown). Diagnostic yield varied by pregnancy outcome (Supplementary Table 2). The diagnostic yield was significantly higher for pregnancies resulting in neonatal death (27.3%), pregnancy termination (27.4%) and miscarriage (28.6%) compared to pregnancies resulting in stillbirths (10.5%) or living children (8.2%, χ2 = 18.2, df = 4, P = 0.0011).

Diagnostic yield by URM and US status

In the pediatric families, there was a similar diagnostic yield for URM (at least one parent URM) (26.0%) and non-URM (both parents not URM) individuals (27.7%; χ2 = 0.08, P = 0.78; Supplementary Table 5). There was no significant difference in yield based on the number of URM parents (25.1% for two URM parents versus 28.6% for one URM parent). The diagnostic rate was also not significantly different for US families (25.8%) compared to non-US families (31.9%; χ2 = 1.19, P = 0.28) (Supplementary Table 6). Inconclusive results were present in 14.4% of URM individuals compared to 9.2% of non-URM individuals (χ2 = 1.22, P = 0.27), and in 14.9% of US individuals and 8.3% of non-US individuals (χ2 = 2.21, P = 0.14). We also stratified families by URM and US status together (Supplementary Table 6) and diagnostic yield and rate of inconclusive case classifications did not differ across these joint categories. For the prenatal families, the diagnostic yield was also similar between URM (16.0%) and non-URM (15.4%) families (χ2 = 0.13, P = 0.72) and did not differ by the number of URM parents (13.0% for two URM versus 22.2% for one URM parent, Table S5); the same was true for US (17.8%) versus non-US families (20.0%) (χ2 = 0.27, P = 0.60; Table S6). Likewise, the inconclusive rates were 6.7% for URM versus 3.3% for non-URM families (χ2 = 1.34, P = 0.25) and 7.5% for US versus 5.3% for non-US families (χ2 = 0.66, P = 0.42). Joint analysis of URM and US status also revealed no significant differences (Supplementary Table 6).

Parental age effects

Parental age was higher for de novo variants with AD inheritance, with a mean paternal age at conception of 35.3 years for fathers of pediatric patients with de novo variants compared to a mean age of 32.3 years for inherited AD and AR variants (P = 0.05) and a mean age of 32.1 years for patients receiving negative results (Supplementary Table 7). Maternal age was also higher, with a mean maternal age of 30.8 years for mothers of pediatric patients with de novo variants with AD inheritance compared to a mean maternal age of 28.4 years for inherited variants (P = 0.021) and 28.5 years for patients receiving negative results. For prenatal patients, parental ages were also increased for de novo AD variants, but the increase was not statistically significant (Supplementary Table 7).

Distribution of variant types

As anticipated, the type of variant (frameshift, stop, missense, in-frame deletion, and splice-site) correlated with the degree of diagnostic certainty. In the entire group, an analysis of the association of variant type with case classification showed that definitive positive patients had the highest frequency of frameshift variants (27.0%), followed by patients with probable positive results (19.5%) and patients with inconclusive results (9.7%, Supplementary Table 8). This pattern was similar for variants predicting stop-gain/loss, with 29.8% in patients with definitive positive results, 14.3% in patients with probable positive results and 7.5% in patients with inconclusive results. In contrast, missense variants were present in 35.5% of patients with definitive positive results, 55.8% of patients with probable positive results, and 72.0% of patients with inconclusive results. In-frame deletions and splice-site variants were infrequent and showed no clear differences among the case classifications.

Secondary findings

Overall, 712 patients opted to receive secondary findings, including 266/316 prenatal patients (85.0%) and 446/529 pediatric patients (84.3%), as reported previously23. There were 26 secondary findings that were reported, 14 in pediatric patients (2.6%) and 12 in prenatal patients (3.8%, χ2 = 0.88, P = 0.348) (data not shown).

Multivariate analyses

In a multinomial multivariate analysis of case outcome versus sex, prenatal vs pediatric, URM status, US status, maternal age, paternal age, maternal education, household language, insurance, MUA status, HPSA status, and number of family members sequenced, the beta for a diagnostic outcome in pediatric versus prenatal cases was −0.97 (P = 0.0019) for definitive positive, −0.85 (P = 0.061) for probable positive, and −0.85 (P = 0.035) for inconclusive when compared to negative cases. With a Bonferroni threshold of P < 0.0014, paternal age, URM, and US status were not statistically significant, nor was the number of family members sequenced or the remainder of the covariates listed above (Supplementary Table 9).

留言 (0)

沒有登入
gif