Role of next generation sequencing in diagnosis and management of critically ill children with suspected monogenic disorder

Abstract

Next generation sequencing based diagnosis has emerged as a promising tool for evaluating critically ill neonates and children. However, there is limited data on its utility in developing countries. We assessed its diagnostic rate and clinical impact on management of pediatric patients with a suspected genetic disorder requiring critical care. The study was conducted at a single tertiary hospital in Northern India. We analyzed 70 children with an illness requiring intensive care and obtained a precise molecular diagnosis in 32 of 70 probands (45.3%) using diverse sequencing techniques such as clinical exome, whole exome, and whole genome. A significant change in clinical outcome was observed in 13 of 32 (40.6%) diagnosed probands with a change in medication in 11 subjects and redirection to palliative care in two subjects. Additional benefits included specific dietary management (three cases), avoidance of a major procedure (one case) and better reproductive counseling. Dramatic therapeutic responses were observed in three cases with SCN1A, SCN2A and KCNQ2-related epileptic encephalopathy. A delayed turn-around for sequencing results was perceived as a major limiting factor in the study, as rapid and ultra-rapid sequencing was not available. Achieving a precise molecular diagnosis has great utility in managing critically ill patients with suspected genetic disorders in developing countries.

Introduction

Next generation sequencing (NGS) technology has emerged as an essential tool for the evaluation of a heterogeneous cohort of individuals with genetic disorders [1, 2]. It has transformed medical practice by accelerating and simplifying genetic diagnosis by expanding the net of differential diagnosis [3, 4]. With rapidly decreasing costs, this technology has been increasingly used in clinical practice. There are limited studies assessing the utility of NGS in intensive care units especially from low/middle income countries. It is well recognized that these patients are enriched for genetic disease [5]. The rapidly evolving clinical situation in the emotionally charged setting of critical care necessitates timely patient identification, skilled assessment, and sensitive handling of complex genetic issues. We describe our experience of using different NGS based techniques in children with critical illness.

Materials and methods

The study was conducted at a genetic center in affiliation with the intensive care of a tertiary care hospital in Northern India from July 2015 to June 2020. The hospital has 675 in-patient capacity, 35 NICU and 15 PICU beds. Annual NICU/PICU admissions range between 800–1000 patients per year. The subjects comprised neonates and children (upto 18 years), who had a suspected Mendelian disorder and were admitted in the intensive care unit. Genetic consult was sought for 122 patients in study period (2.7%), NGS was advised in 46 patients but could be done in only 22 patients (7 NICU, 15 PICU). The testing was self-funded by families. The prospective cohort included 22 probands (31.4%) where the proband was seen during critical illness and NGS was done. The retrospective cohort comprised of 48 (68.6%) probands, seen in the genetics out-patient, where NGS was done after the period of critical illness. The study was approved by the institution ethics committee (EC/07/18/1384).

Study design

A multidisciplinary working group developed the patient selection criteria with the clinical geneticist as a core coordinator. Standard operating procedures were designed for patient recruitment, informed consent, test performed, report analysis and communication with parents.

We enrolled subjects from July 2015 through June 2020 and included patients retrospectively (after discharge from the NICU/PICU), as well as prospectively (during the NICU/PICU admission).

All patients had at least one of the following features: one or more congenital malformations, dysmorphic features, abnormalities in growth, neurologic impairment (including encephalopathy, seizures or hypotonia), features suggestive of a metabolic condition, multisystem, unexplained, or atypical presentation. Clinical phenotype was reported in HPO terms. Relevant investigations including metabolic tests, radiographs, echocardiogram, or Magnetic Resonance Imaging (MRI) were performed as part of standard care. If a definitive diagnosis was not established by these investigations, the family was counseled for next-generation sequencing tests. New-borns with suspected chromosomal disorders (e.g., trisomy 18) or specific monogenic conditions (e.g., spinal muscular atrophy) or largely non-genetic (e.g., congenital infections) were excluded.

An expert panel of investigators (SB, RP) determined patient eligibility. A second panel of clinicians (SBM, ICV) participated as required for further assessments and discussions. In case of a disagreement among panel members regarding eligibility, majority approval determined the further course. The overall process of enrollment along with inclusion and exclusion criteria is summarized in eSupplementary 1.

Data analysis and interpretation

Next-generation sequencing was performed as panel/clinical exome/whole exome or genome in solo/trio as feasible and based on recommendations of the geneticist. Clinical exome included analysis of exons of all clinically validated genes listed in OMIM. Variant classification based on the standards of American College of Medical Genetics and Genomics (ACMG) was done with a phenotype driven approach [6]. A case was considered diagnosed if a pathogenic or likely pathogenic variant was identified. Variants of Uncertain Significance (VUS) were analyzed by clinical correlation, literature review, segregation analysis and raw data reanalysis on a case-to-case basis. Sanger validation, dosage analysis by MLPA/real time PCR, and mRNA studies were done as required to confirm a definitive diagnosis. All pathogenic and likely pathogenic variants identified in this study have been deposited in ClinVar [7] (Submission Ids: SCV001434393, SCV002097319, SCV002097320, SCV002097321, SCV002097322, SCV002097323, SCV002097324, SCV004042670, SCV004042672, SCV004042841, SCV004042842, SCV004045776, SCV004045777, SCV004045984, SCV004045987, SCV004045988, SCV002583593, SCV002583594, SCV004046662, SCV004046664, SCV004046869, SCV004046870, SCV004046871, SCV004046872, SCV004046873, SCV004046874, SCV004046875, SCV004046876, SCV004046877, SCV004046878, SCV004098990, SCV004098991, SCV004098992, SCV004100881, SCV004099282).

Outcome measures and statistical analysis

The primary outcome measure was diagnostic yield. The secondary outcome measure was the clinical impact of the definite molecular genetic diagnosis. This was assessed as impact on change in medication and/or diet (initiation or discontinuation), need/deferral of a major procedure, palliative therapy and genetic counseling. A single patient could have more than one outcome measure change. The number of families opting for prenatal diagnosis during the next pregnancy and its outcome was also assessed. Descriptive statistics were used for participant characteristics, referral site, interval between symptom onset and genetic testing, turnaround time of sequencing, time to result, change in management and reproductive counseling. Data collection was complete, and no patients were lost to follow up.

Results

Participant demographics and indications for testing

Over 5 years, 75 patients were evaluated. Five patients did not meet the study criteria and were excluded (non-genetic etiology in one, chromosomal in two, specific genetic testing available in two patients). Finally, 70 participants were enrolled for further analysis by NGS.

The study group comprised of seven neonates, 25 infants and 38 children above 1 year of age. The average age at enrollment was 2.47 years (01 day–13 years) and the median age was 1.29 years. Patients were divided into two groups— group I included 40 patients with a critical illness episode requiring NICU admission and group II had 30 infants/children with a post-neonatal critical illness requiring PICU care. In the NICU cohort, 11 neonates had perinatal asphyxia while seven had hypoglycemia. Other presentations included meconium aspiration syndrome, respiratory distress, cardiac involvement, dyselectrolytemia and sepsis. Seven probands had died before NGS could be done, and the test was performed on stored DNA, whereas 12 patients expired later during the study. The patient characteristics are summarized in Table 1.

Table 1.

Demographic data of children enrolled (N = 70).

Criteria	Total 70	Diagnosed 32 (45.7%)	Undiagnosed 38 (54.3%)
Age (at presentation)
Less than 1 month	07	03 (42.9%)	04 (57.1%)
1 month–1 year	25	15 (60%)	10 (40%)
More than 1 year	38	14 (36.8%)	24 (63.2%)
Critical illness
NICU	40	20 (50%)	20 (50%)
PICU	30	12 (40%)	18 (60%)
Gender
Male	46	16 (34.8%)	30 (65.2%)
Female	24	16 (66.7%)	08 (33.3%)
Consanguineous parents
Yes	8	4 (50%)	4 (50%)
No	62	28 (45.2%)	34 (54.8%)
Sibling affected
Yes	13	5 (38.5%)	8 (61.5%)
No	57	27 (47.4%)	30 (52.7%)
Proband alive till completion of study
Yes	51	21 (41.2%)	30 (58.9%)
No	19	11 (57.9%)	8 (42.1%)
Test done after death	7	6 (85.7%)	1 (14.3%)
Proband died later	12	5 (41.7%)	7 (58.3%)

Patient phenotype

The presenting phenotype was neurological in 37 of 70 (52.8%), multi-system in 17 (24.3%), gastrointestinal/hepatic in 5 (7.1%), hematologic in 3 (4.3%) and cardiovascular in 3 (4.3%) probands (Fig. 1a). The remaining five presentations included one each of congenital adrenal hypoplasia, diabetes insipidus, pseudo-Bartter syndrome, renal failure, and surfactant deficiency.

Fig. 1. NGS diagnostic yield as per patient phenotype and dataset analysis.

a Patient phenotype related diagnosis. The outermost circle represents the total number of cases enrolled (N = 70). The middle circle shows that 38 of these cases remain undiagnosed. The inner circle shows that 32 cases were diagnosed by NGS. The presenting phenotype is color coded. The most common phenotype was CNS involvement (blue color) in 37 cases of which 21 remained undiagnosed while 16 were diagnosed by NGS. b Diagnostic yield of NGS. The stacked histograms represent the distribution of diagnosed (32) and undiagnosed (38) cases by different NGS dataset analysis.

Recurrent seizures were present in 37 probands, neuro-metabolic phenotype in 16 and neuro-regression in six. Global developmental delay (GDD) or intellectual disability (ID) was present in 38 probands at the time of critical care admission. Dysmorphism or malformations were documented in 20 patients.

Interval between critical illness and NGS (time to testing)

The study commenced in July 2018. In the prospective arm, 22 patients were recruited from the hospital intensive care units (NICU 7 and PICU 15) in the study window. Seventeen underwent genetic testing during the critical illness, of which a diagnosis was made in 10 patients.

In the retrospective arm, 48 patients were recruited from medical records review. None of them underwent genetic testing during the critical illness. Twenty-two received a confirmed molecular diagnosis by NGS performed on outpatient follow up. There was no difference in diagnostic yield in prospective and retrospective groups.

The time interval between critical illness and NGS showed a wide variation ranging from 1 day to 12 years. There was a gap of more than 1 year between the critical illness and NGS in 25 children. The average turnaround time of NGS in the study was 43 days (range 14–56 days) except in one case with neonatal seizures, where the result was provided within two weeks resulting in early seizure control.

NGS results

Clinical exome (CE) comprising of OMIM genes was done in 51 patients, whole-exome solo in 13, panel testing in six, and whole-genome in one patient. The sequencing was done predominantly as a “solo” NGS, except in two patients where “trio” whole-exome sequencing (WES) of the proband, mother and father was done. The diagnostic yield in the present study was 45.7% with a genetic diagnosis established in 32 of 70 probands (Fig. 1b).

A definite genetic diagnosis was established in four out of eight (50%) probands with consanguineous parents. Amongst 13 families, with a previous history of similarly affected sibling, genetic diagnosis was made in five probands (38.5%). The cohort with recurrent seizures had a diagnostic yield of 45.9% and within this cohort, the yield in the sub-group of epileptic encephalopathy was 55%. The cohorts with CNS, GIT, hematologic or multisystem illness did not show any statistical difference in diagnostic yield, though these have a small sample size. Three infants with a predominant CVS phenotype were also included in the study, but no specific diagnosis was obtained by NGS.

The diagnostic yield in the cohorts by phenotype and by type of NGS are presented in Tables 2 and 3, respectively. Whole exome sequencing captures information about the ~20 K protein coding genes of the human genome, while the clinical exome captures information about 5k genes with known clinical importance (OMIM morbid genes OR Mendeliome). In our study all the causative genes identified on the whole exome sequencing were also present on clinical exome data.

Table 2.

Clinical features of the patient cohort and diagnostic yield (N = 70).

Feature	Total 70	Diagnosed 32 [45.7%]	Undiagnosed 38 [54.3%]
Primary phenotype
CNS	37	16 [43.2%]	21 [56.8%]
GIT/hepatic	5	1 [20%]	4 [80%]
Hematologic	3	2 [66.7%]	1 [33.3%]
CVS	3	0	3 [100%]
Multi-system	17	9 [53%]	8 [47%]
Others (One case each)	5	4 [80%]	1 [20%]
GDD/Intellectual disability
Present	38	18 [47.4%]	20 [52.6%]
Absent	32	14 [43.8%]	18 [56.2%]
Dysmorphism/Malformation
Present	20	7 [35%]	13 [65%]
Absent	50	25 [50%]	25 [50%]
Neuro-regression	6	3 [50%]	3 [50%]
Epileptic encephalopathy	20	11 [55%]	9 [45%]
Neurometabolic	16	5 [31.2%]	11 [68.8%]
Recurrent seizures	37	17 [46%]	20 [54%]

Table 3.

Genotypic and phenotypic distribution by type of NGS technique.

NGS	OMIM#	Gene	Disorder
			Phenotype – Neurological (16)	Phenotype-GIT (1) Haematology (2)	Phenotype - Multi-system (9)	Misc (4) Misc (4)
	219700	CFTR				Cystic Fibrosis
	610188	CEP290			Joubert syndrome type 5
	266100	ALDH7A1	Pyridoxine dependent Epilepsy
	603553	PRF1		Familial hemophagocytic lymphohistiocytosis-2
	237300	CPS1	Hyperammonaemia due to carbamoyl phosphate synthase deficiency
	163950	PTPN11			Noonan syndrome 1
	221750	LHX3			Combined Pituitary Hormone Deficiency type 3
	612164	STXBP1	EIEE-4
	617166	FGF12	EIEE-47
	304800	AVPR2				Nephrogenic DI
	202700	ELANE		Severe congenital neutropenia-1
	616211	WWOX	EIEE-28
	311250	OTC	Ornithine trans- carbamylase deficiency
	300200	NR0B1				CAH with hypogonadotrophic hypogonadism
	607208	SCN1A	EIEE-6
	252160	MOCS2	Molybdenum cofactor def B
	241200	KCNJ1			Antenatal bartter syndrome type 2
	613721	SCN2A	EIEE-11
Neurology panel	615471	FBXL4	Encephalomyopathic mitochondrial DNA depletion syndrome-13
Epileptic panel	607208	SCN1A	EIEE-6
Clinical exome with del/dup	613720	KCNQ2	EIEE7
	219700	CFTR			Cystic Fibrosis
Whole Exome Solo	609015	HADHA			Trifunctional protein deficiency
	272800	HEXA	Tay-Sachs disease
	618285	CACNA1E	EIEE-69
	612949	SLC25A12			EIEE39
	610921	ABCA3				Surfactant metabolism dysfunction pulmonary, 3
	203700	POLG			Mitochondrial DNA depletion syndrome 4A
	245349	PDHX	Lactic acidaemia due to PDX1 deficiency
Trio whole-exome	251880	DGUOK		Mitochondrial DNA depletion syndrome 3
	618744	UGP2			EIEE83
Whole-genome	618580	PIGB	EIEE-80

NGS variant analysis

Ninety-six variants relevant to the phenotype were detected in the NGS analysis. These included 85 single nucleotide variants (SNV), which comprised of 68 missense (14 homozygous and 40 heterozygous), nine nonsense and eight splice–site variants; along with three in-frame indels, and six out of frame indels and one homozygous exonic deletion.

No variants were reported in 11 of 70 cases, “variant of uncertain significance” in 35, likely pathogenic in 11 and pathogenic in 13 cases, resulting in a preliminary diagnosis in 24 out of 70 cases (34.3%). Clinical correlation, literature review, Sanger validation and segregation analysis assisted in nine variants being reclassified as benign/likely benign, and seven as likely pathogenic variants. A separate consent was sought for study of incidental findings during pre-test counseling [8]. Four families had consented for the same but no pathogenic variant in genes unrelated to the phenotype was detected.

Fastq files of NGS were re-examined in eight undiagnosed cases and did not result in any change in classification of the variants in seven cases. However, in one case which had already undergone CE in 2016 followed by trio WES in 2018, reanalysis of the data identified a homozygous c.1A > G UGP2 variant, which disrupts the translational start site in the shorter isoform (NM_001001521), reported as pathogenic in Dec 2019. Thus, the final diagnosis was established in 32 of 70 cases (45.7%), which included 24 diagnosed by preliminary NGS analysis, 7 by reclassification of VUS and one after data reanalysis. Of the 32 diagnosed cases, nine were autosomal dominant, 20 autosomal recessive and three X-linked disorders. Seven novel variants have been detected in this study.

Secondary end points

Impact on clinical management

Establishing a definite diagnosis altered medical management in 11 (34.4%) of 32 patients. These included two cases of SCN1A and one case each of ALDH7A1, AVPR2, CFTR, CPS1, ELANE, KCNQ2, OTC, PTPN11 and SCN2A related genetic disorder (Table 4).

Table 4.

Impact of a definitive diagnosis on management.

Pt	Phenotype	Gene	Diagnosis	Drugs	Diet/Invasive procedure/Specialist referral
2 years F	Onset day 1, recurrent seizures	ALDH7A1: c.1411-1412insG, c.187G > T NM_001182.5	Pyridoxine dependent epilepsy	Pyridoxine supplementation
10 months M	Neonatal onset recurrent vomiting, polyuria	AVPR2: c.185T > C, hemizygous NM_000054.6	Nephrogenic diabetes insipidus	Diuretic, indomethacin
4 months F	Onset 1 month, fever, seizures, resp infection	CFTR: c.1521_1523 delCTT c.1646G > A NM_000492.4	Cystic fibrosis	Pancreatic enzyme supplementation, azithromycin prophylaxis	Pediatric gastroenterologist
8 days M	Onset day 3, lethargy, vomiting, seizures	CPS1: c.236 + 4A > G, homozygous NM_001875.5	Hyperammonaemia carbamoyl phosphate synthase deficiency	Sodium benzoate, arginine	Protein-restricted diet
2.5 years F	Onset 1 year, recurrent mucosal infections	ELANE: c.182C > T heterozygous NM_001972.4	Severe congenital neutropenia	Inj G-CSF three times a week
1 month F	Neonatal seizures day 1, encephalopathy	KCNQ2: c.629G > A heterozygous NM_172107.4	EIEE7	Carbamazepine
5 years M	Onset 2 months, recurrent lethargy, vomiting	OTC: c.386G > A hemizygous NM_00051.6	Ornithine transcarbamylase deficiency	Sodium benzoate	Protein-restricted diet
3 months M	Neonatal, facial dysmorphism, heart disease, juvenile myelomonocytic leukemia	PTPN11: c.218C > T, heterozygous NM_002834.5	Noonan syndrome 1	6-mercaptopurine	Avoidance of Haematopoietic stem cell transplant, Endocrinologist
7 months M	Onset 5 months, Recurrent seizures	SCN1A: c.765_766 delGT heterozygous NM_001165963.4	EIEE6	Valproate, clobazam, potassium bromide	Ketogenic diet
3 years M	Onset 15 months, Febrile seizures, shock	SCN1A: c.1837C > T heterozygous NM_001165963.4	EIEE6	Valproate
5 years F	Neonatal seizures day 3	SCN2A: c. 751G > A heterozygous NM_001040142.2	EIEE11	Carbamazepine

Specialized dietary management was advised in three patients; two with urea cycle defect were recommended a protein-restricted diet and one patient with SCN1A related EIEE6 was advised ketogenic diet. Two families opted for palliative care in view of diagnosis of Tay-Sachs disease and congenital surfactant deficiency disease (Fig. 2). One infant was being considered for haematopoietic stem cell transplant for juvenile myelomonocytic leukemia, which was withheld after molecular diagnosis of Noonan syndrome 1. Knowledge of the diagnosis provided useful and beneficial information about the natural history and prognosis, allowing accurate counseling about recurrence risk in all probands.

Fig. 2. Impact of diagnosis by NGS on outcome of 32 diagnosed cases.

Impact of molecular genetic diagnosis on management. More than one outcome measure change could be present in one case.

Brief description of some cases with exemplary response to genotype differentiated therapy

One-month old girl with features of epileptic encephalopathy from day 1 on multiple AEDs (phenobarbitone, phenytoin, levetiracetam, topiramate) along with biotin, pyridoxine and riboflavin was diagnosed within 14 days. Extensive metabolic testing with urine GCMS, TMS, serum ammonia, lactate and CSF neurotransmitter analysis was non-diagnostic. A heterozygous missense, previously reported variant in KCNQ2, c.629G > A, NM_172107.3 (p.Arg210His) causing EIEE-7 was identified. Inclusion of carbamazepine to the management protocol resulted in dramatic control of seizures, which may be considered as an exemplary precision therapy.

Parental testing confirmed the “de novo” nature of the variant and other AEDs were gradually tapered. At the 18 months follow up, parents reported that the child was seizure-free from the day of starting of carbamazepine, and has normal motor, language, and cognitive development.

A 7-month boy with recurrent prolonged febrile seizures was diagnosed to have SCN1A related EIEE-6 and subsequent guided therapy with valproate, clobazam, potassium bromide and ketogenic diet resulted in partial control of seizures. In view of the genetic variation, follow up and early intervention for speech was advised.

A 5-year girl with seizure onset at day 3 of life, followed by recurrent seizures of tonic-clonic and absence type with mild developmental delay was diagnosed to have SCN2A related EIEE-11. Early-onset seizures in SCN2A are often gain of function mutations which respond well to sodium channel blockers. The child responded to the addition of carbamazepine and is seizure-free at 1 year follow up.

Reproductive outcomes in families

During the limited study period, ten couples had planned pregnancy and requested for prenatal testing by chorionic villus sampling, which includes two couples requesting for a second time, leading to a total of 12 procedures. Two of these were done for autosomal dominant disorders due to CACNA1E related developmental and epileptic encephalopathy (DEE) 69, and FGF12 related DEE 47 pathogenic variations. Both fetuses were unaffected. The remaining ten tests (in eight families) were done for autosomal recessive disorders. Two fetuses in the same family were consecutively affected, and pregnancy was terminated as per the decision of family. The remaining eight pregnancies were unaffected and were continued to term with birth of normal babies. A brief summary of diagnosed cases is presented in eSupplementary 2.

Discussion

Utility of NGS dataset analysis in critical care setting has been evaluated in recent times and the results appear promising [9–20]. We examined NGS as a diagnostic tool in 70 critically ill neonates and children suspected to have a monogenic disorder and achieved a diagnosis in 32 (45.7%) of 70, 50%, (20 of 40) in the NICU cohort versus 40%, (12 of 30) in the PICU cohort, though not statistically significant (p = 0.40). The diagnostic yield of conventional genetic testing without NGS is reported at 26% (34 of 132) in sick neonates [21] compared to a median diagnostic yield of 38.9% in 18 intensive care NGS based genomic studies [3, 9–18, 22–28] (Table 5). In the current study, the impact of a definitive diagnosis on management was observed in 13 of 32 (40.6%) probands with alteration of medication, diet, invasive procedure, and palliative care. This study is unique in that it incorporates a combination of diverse datasets frequently used in clinical practice, thus more closely reflecting the actual utility of NGS by the bedside. Most reported data is from developed nations. The current study is one of the few studies from a developing country context where most genetic testing is not funded and therefore informed, cost-effective testing options are essential.

Table 5.

Comparison of the present study and previous studies of NGS in pediatric critical care.

S.No	Study population	Type of NGS	Overall diagnostic yield	Time to result (days)	First author, year
1	Critically ill infants in NICU	rWGS	11/15 (73%)	12	Soden [3]
2	Critically ill infants <4 months	Rapid GS	20/35 (57%)	23	Willig [9]
3	Newborns and infants in NICU	Targeted WES	8/20 (40%)	14–406	Daoud [10]
4	Critically ill infants in ICU	Targeted rWES	7/23 (30%)	12	van Diemen [11]
5	Infants <100 days in intensive care	ES	102/278 (37%), Critical trio 32/63(51%), Trio 13/39 (33%), Solo 57/176 (32%)	13	Meng [12]
6	Acutely unwell pediatric patients	rWES	21/40 (52%)	16	Stark [13]
7	Infants <4 months in NICU/PICU	rWGS trio	12/37 (32%)	14	Petrikin [14]
8	Acutely ill inpatient infants	rWGS	18/42 (43%)	2–5	Farnaes [15]
9	PICU children	rWGS trio	10/24 (42%)	9	Mestek-Boukhibar [16]
10	PICU children 4 months–18 years	rWGS	17/38 (45%)	14	Sanford [22]
11	Intensively ill, 1 day–15 years in NICU/PICU	WGS trio	40/195 (21%)	21	French [17]
12	Ill newborns in NICU	nGS	0/29 (0%), Inconclusive 5/29 (17%)	–	Ceyhan-Birsoy [18]
13	Critically ill infants <4 months, NICU/PICU CVICU	Total	48/213 (22%)	4.6–11	Kingsmore [23]
		urWGS,	11/24 (46%)	4.6
		rWGS,	18/94 (19%)	11
		rWES	19/95 (20%)	11
14	Exome sequencing in neonates	Diagnostic exome sequencing	25/66 (37.9%)	8	Powis [24]
15	Genetic defects in NICU	Medical exome with CNV analysis	284/2303 (12.3%)	–	Yang [25]
16	Intensive care in <6 months	Rapid Trio-whole exome sequencing	29/50 (58%)	7	Gubbels [26]
17	NICU clinical care	Rapid clinical exome	22/80(28%)	13	D’Gama [27]
18	Critically ill infants	Rapid trio- exome sequencing	6/15 (40%)	16	Wells [28]
19	Critically ill neonates and children, 1 day–18 years, NICU/PICU	NGS (Panel, CE, WES, WGS), solo/trio	32/70 (46%)	43	Present study

In a developing country with restricted resources and predominant out of pocket expenditure, testing phenotype relevant genes remains the first option. However, with decreasing sequencing costs, trends toward exome sequencing are evolving [1, 29, 30]. The incidence of genetic disease in seriously ill infants with disease of unknown etiology, enrolled in the NSIGHT2 trial [23] was 24% (52 of 213 infants). The diagnostic yield was 46% (11of 24) in the gravely ill sub-group where ultrarapid WGS (urWGS) reporting in 4.6 days was used, compared to 20% (37 of 189) in the rWES/rWGS group with TAT of 11 days. Though we did not perform rWES/rWGS, a definitive molecular diagnosis was established in 20 of 40 (50%) probands in NICU cohort in 4–6 weeks. The higher diagnostic yield observed in this study could be due to a small, highly curated cohort where all patients were evaluated by a geneticist prior to NGS testing, with an increased pre-test probability of a genetic diagnosis. Diagnostic yield was augmented by deep phenotyping (case 1), functional testing (case 2) and data reanalysis (case 3) (eSupplementary 2). However, the expertise required for a rapid turnaround time data analysis and genomic precision testing may be only available at very limited tertiary centers in developing nations where genetic literacy is in infancy.

In this study cohort, seven of 11 patients with perinatal asphyxia were confirmed to have a genetic diagnosis. These included CEP290 related Joubert syndrome, STXBP1related EIEE4, combined pituitary hormone deficiency type 3 due to LHX3 variants, CACNA1E related EIEE69, pyruvate dehydrogenase E3 binding protein deficiency due to PDHX, UGP2 EIEE83 and WWOX related EIEE28. Genetic and metabolic causes of neonatal encephalopathy are mistaken for hypoxic ischemic encephalopathy and sub-optimal testing results in missed opportunities for diagnosis. Therefore, the mere presence of perinatal asphyxia should not be a deterrent to genetic testing as implications for therapy and reproductive counseling are significant [31, 32].

Time-to-diagnosis is an essential consideration in critically ill neonates and children [33, 34]. Soden et al. [3]. Reported a 77-month delay between symptom onset and NGS in ambulant patients with neurodevelopmental delay and a time to diagnosis of 11.5 months by WES, which could be reduced by 10 months by rWGS in high acuity patients. In this study, only 17 (24%) of 70 probands underwent NGS during the critical illness with a median turnaround time was 43 days, which is not optimal for therapeutic utilization in acute care. We did not have access to rapid or ultrarapid WES/WGS. Turnaround times (TAT) for results has an impact on clinical utility in the critical illness setting. Of the 36% (9 of 25) patients with delayed diagnosis of more than 1 year, precision therapy could have been initiated earlier in one child with EIEE-11genotype. Rapid or ultrarapid testing was not achieved in this study. NGS testing in this study, as in most developing countries, was outsourced to commercial companies where rapid return of results is a limitation. Trio WES/WGS with a rapid/ultrarapid turnaround of results is a naïve concept in developing nations, as costs are extremely high with predominantly out of pocket expenditure. An inhouse testing facility requires a high sample volume load for cost-effective testing. There are also limitations of expert of data analysis and interpretation of genomic results. Additionally, there is limited appreciation of the benefits of testing in the critical care setting where active management of a sick baby is the prime concern. We also encountered lack of consent by the grieving families and 20% came for genetic consultation for recurrence information during the next pregnancy, possibly as recurrence is a greater perceived fear than the appreciation of rapid testing in a sick child.

Amongst diagnosed patients, results were considered clinically useful in 25% with standard TAT WES (median 136 days) by Daoud et al. [10] and range from 27–95% in rWGS and rWES with TAT of 12–23 days in various studies [9, 11–13, 17, 23, 35] Ultra-rapid WGS with a TAT of 4–5 days reported in NSIGHT1 trial [14] seems to represent an ideal solution; however, the approach may be technically challenging and with significant costs. However, if trio rapid sequencing (WES/WGS) was done for all patients in this study, an early diagnosis would have been established in 30 of 32 patients who received a definitive diagnosis. The two undiagnosed cases include a family requiring mRNA testing to ascertain variant pathogenicity and the second proband where the causative gene, UGP2, was not identified at the time of first reporting [36]. The impact of a definite diagnosis on management was observed in 13 of 32 (40.6%) patients in this study. Of those who consented for testing at the time of intensive care admission, a rapid diagnosis by rWES/WGS could have potentially resulted in genetic guided precision therapy for seizure control in five cases, along with rapid control of neutropenia in one proband. A recent meta-analysis reported a change in clinical management in 6–27% children and recommended that WGS/WES should be considered a first-line genomic test in suspected genetic disease [37]. These results have been replicated over time in several independent studies, systematic reviews and meta-analysis [38–42]. A study from Thailand has also suggested that the use of rapid WES (rWES) as a first tier investigation tool can provide tremendous benefits in diagnosis and management of critically ill patients reporting molecular diagnosis in 25 (46%) and change in management in 24 (44%) of 54 unrelated patients [43]. This information is very relevant while counseling families for rapid genetic testing.

A definitive molecular diagnosis also provides valuable information regarding the mode of inheritance and the consequent risk of recurrence in future pregnancies [44]. A genetic diagnosis facilitates understanding and acceptance in family members and society, provides closure, reduces guilt and self-blame in parents [45, 46]. Additionally, it can provide hope for the future and possible enrollment in clinical trials. We have not included the changes in reproductive risk counseling as a change in management of proband, even though the information has been extremely valuable to the families. We faced a peculiar situation thrice over during this study, where prenatal diagnosis could not be performed due to advanced gestation because of delayed confirmation of genomic results in the proband. These families would have benefitted by rapid NGS testing with its shorter TAT. Most families in the developing world come during pregnancy for direct prenatal diagnosis as physicians are not informed of the need to test the proband prior to fetal testing.

Limitations

Major hurdles to widespread utilization of NGS in this cohort include long turnaround time, possible requirement of testing of parents and limited diagnostic yield. The present study is limited due to its small sample size. A randomized control group could not be used, which would have allowed the comparison of diagnostic yield using conventional genetic techniques other than NGS, time to diagnosis, cost-effectiveness, health and psychosocial impacts for participating families.

Diverse NGS techniques including panel testing, WES as well as WGS have been used in the study, the predominant test utilized being targeted exome focused on OMIM disease-causing genes (Clinical Exome) in 51 of 73 dataset analysis. The turnaround time of nearly 6 weeks of NGS is not optimal in critical care setting but was used due to lack of access to rapid sequencing (TAT: 7–10 days) or ultra-rapid sequencing (TAT: 2–5 days) at the time of this study.

Conclusion

The utility of NGS in critical care illness is being established with increasing awareness and accessibility of the technique in clinical practice. A significant proportion of morbidity and mortality in NICU and PICU is attributed to genetic disorders and an early diagnosis is vital to improve outcomes in this cohort. Our study contributes to establish the role of diverse NGS dataset analysis in critical care illness by achieving a diagnostic rate of 45.7% (32 of 70 probands), with a clinically significant change in outcome in 40.6% (13 of 32 diagnosed probands) by virtue of change in medication, diet, or major change in therapeutic approach. A significant proportion have also been helped by enhanced reproductive options available due to a firm genetic diagnosis. An early diagnosis enabled by rapid or ultra-rapid sequencing will further enhance the clinical outcome in critical care genomics.

Author contributions

All the authors have made substantial contributions to the final manuscript including conception and design; acquisition, analysis or interpretation of data. Each author has participated sufficiently in the work and takes public responsibility for appropriate portions of the content. All the authors agree that they would be accountable for all aspects of the work to ensure that questions related to the integrity or accuracy of any part of the work are appropriately investigated and resolved. SB, ICV and RP made substantial contributions to conception and design. SB, DG, AS, SBM, ICV and RP contributed to acquisition of data. SB, SP, SK and RP contributed to analysis and interpretation of data. SB, SP, ICV and RP contributed in drafting the manuscript and critical revision. All authors approve the final version to be published.

Funding

The study was performed as part of clinical care without research funds.

Data availability

The authors confirm that the data supporting the findings of the study are available within the article and its Supplementary Material. Raw data files will be made available on reasonable request.

Competing interests

The authors declare no competing interests.

Ethical approval

The study was approved by the Institution Ethics Committee (EC/07/18/1384).

Supplementary information

The online version contains supplementary material available at 10.1038/s41431-024-01569-z.