Power and Pitfalls of Computational Methods to Identify New Genes Responsible for Acute Liver Failure of Indeterminate Etiology in Adults

We have read with great interest the recent article by Rakela et al. (1) describing genetic alterations detected by whole exome sequencing in a cohort of 26 adult patients with acute liver failure of indeterminate etiology (ALF-IE). The authors conclude that 12 variants of 11 genes were significantly associated with (ALF-IE), particularly the stop codon of SEPRINB11 identified in 23 of the 26 patients and the missense variant of CYP2D6 detected in 13 patients. However, we have several key concerns regarding the results of the study.

First, Rakela et al. reported that the SERPINB11 (c.268G>T, p.Glu90Ter) variant, recorded as rs4940595 in the database of human single nucleotide polymorphism, was detected in most of the cohort. However, a careful examination of this variant using VarSome (2) and the latest update of the Broad Institute's gnomAD (3), a large database of human polymorphisms, reveals that this SERPINB11 variant has a global allele frequency (AF) as high as 63% for all ethnic subgroups (Tables 1 and 2). Thus, we cannot conclude that the variant is frequently associated with (ALF-IE). According to the pipeline of analysis described in the study (AF threshold of 0.01 in human polymorphism databases), this result can be explained by the fact that authors probably used the old version of gnomAD, in which all the variants and polymorphisms were not aligned with the new version of the genome and not the updated version of gnomAD. This update issue is also true for the second missense variant of CYP2D6, which has neither a high frequency nor an overall frequency for all ethnic subgroups in the updated gnomAD database (Tables 1 and 2).

Table 1.

Extensive information on the variants detected by Rakela et al.

Gene	rs	N	chr	GRCh38	REF	ALT	Canonical RNA transcript	DNA	Protein	Variant type	gnomAD E	gnomAD G
SERPINB11	rs4940595	23	18	63712604	G	T	NM_080475.4	c.268G>T	p.Glu90Ter	Non-sense	—	0.63
	rs4940595	0	18	63712604	G	A	NM_080475.4	c.268G>A	p.Glu90Lys	Missense	—	0.0000279
	rs4940595	0	18	63712604	G	C	NM_080475.4	c.268G>C	p.Glu90Gln	Missense	—	0.00000699
CYP2D6	rs1135840	16	22	42126611	C	G	NM_000106.6	c.1457G>C	p.Ser486Thr	Missense	—	0.577
ANTXRL	rs7091749	8	10	46330066	C	A	NM_001278688.2	c.1878C>A	p.Pro626=	Synonymous	—	0.000007
	rs7091749		10	46330066	C	G	NM_001278688.2	c.1878C>G	p.Pro626=	Synonymous	—	0.194
	rs7091749		10	46330066	C	T	NM_001278688.2	c.1878C>T	p.Pro626=	Synonymous	—	0.0103
MUC6	rs776572312	9	11	1016887	G	A	NM_005961.3	c.5914C>T	p.Pro1972Ser	Missense	0.00002	0.0118
OR6J1	rs1753430	14	14	22634064	A	G	NM_001348233.1	c.748T>C	p.Ser250Pro	Missense	—	0.41
GNAL	rs201898548	3	18	11689671	_	GGCCCT	ENST00000334049.11	c.113_118dupTGGCCC	p.Leu38_Ala39dup	Insertion	0.0842	0.141
	rs201898548		18	11689671	GGCCCT	_	ENST00000334049.11	c.113_118delTGGCCC	p.Leu38_Ala39del	Inframe deletion	0.00104	0.0168
AADACL3	rs3010877	7	1	12719616	C	T	NM_001103170.3	c.310C>T	p.Pro104Ser	Missense	—	0.151
MCL1	rs11580946	3	1	150578851	G	A	NM_021960.5	c.680C>T	p.Ala227Val	Missense	0.00842	0.0086
CYP2D7	rs56404506	11	22	42141186	C	T	NM_001348386.3	c.1196G>A	p.Arg399His	Missense	—	0.281
CYP2D7	rs1800754	15	22	42141587	G	A	NM_001348386.3	c.932C>T	p.Ser311Leu	Missense	—	0.48
	rs1800754		22	42141587	G	T	NM_001348386.3	c.932C>A	p.Ser311Ter	Non-sense	—	—
RRP36	rs200886831	9	6	43021676	_	GCCGGG (6bp)	ENST00000244496.6	c.43_48dupGGGGCC	p.Gly15_Ala16dup	Insertion	0.000075	0.00104
	rs200886831		6	43021676	GCCGGG (6bp)	_	ENST00000244496.6	c.43_48delGGGGCC	p.Gly15_Ala16del	Inframe deletion	0.0419	0.285
	rs200886831		6	43021676	_	GC…GG (12bp)	ENST00000244496.6	c.37_48dupGGGGCCGGGGCC	p.Gly13_Ala16dup	Insertion	—	0.000014
	rs200886831		6	43021676	GC…GG (12bp)	_	ENST00000244496.6	c.37_48del	p.Gly13_Ala16del	Inframe deletion	0.000187	0.000961
KIAA1161	rs4879782	6	9	34372875	G	C	NM_020702.5	c.69C>G	p.Tyr23Ter	Non-sense	—	0.251

ALT, alternative allele; Chr, chromosome; gnomAD, Allele frequency according to gnomAD exome database; gnomAD G, Allele frequency according to gnomAD genome database; GRCh38, position of the variant according to the last annotation of the human genome; N, number of patients with the variants in the cohort described by Rakela et al; REF, reference allele; rs: the identifier (ID) of the variant according to the last version of the human database of single nucleotide polymorphism (dbSNP 151)

Table 2.

AF in GnomAD genome (V3) database of the 2 variants of SERPINB11 and CYP2D6 reported by Rakela et al. in a cohort of adult patients with acute liver failure of indeterminate etiology

Population	SERPINB11: c.268G>T, p.Glu90Ter (rs4940595)				CYP2D6: c.1457G>C, p.Ser486Thr (rs1135840)
	Allele count	Allele number	Homozygotes	AF	Allele count	Allele number	Homozygotes	AF
African	21,033	41,932	5,329	0.502	26,148	40,822	8,881	0.641
Amish	604	900	199	0.671	466	884	128	0.527
Ashkenazi Jewish	2,171	3,322	723	0.654	2,138	3,312	714	0.646
East Asian	1,206	3,124	232	0.386	2,162	3,060	782	0.707
European (Finnish)	6,987	10,438	2,314	0.669	5,217	10,388	1,330	0.502
European (Non-Finnish)	46,721	64,532	16,858	0.724	35,748	64,030	10,432	0.558
Latino	8,124	13,646	2,430	0.595	6,615	13,548	1,734	0.488
South Asian	1,904	3,028	597	0.629	1,663	2,972	489	0.559
Other	1,362	2,150	438	0.633	1,223	2,126	369	0.575
Total	90,112	143,072	29,120	0.629	81,380	141,142	24,859	0.577

AF, allele frequency.

Furthermore, SERPINB11 is a unitary pseudogene in human genome (4). The pseudogene refers to a gene that has lost its functional counterpart during evolution because of a mutagenic event resulting in a premature stop codon. These mutagenic effects are often located in the initial portion of the coding gene, as for SERPINB11. Interestingly, SERPINB11 has the particularity of being polymorphic in humans (4). Hence, the rs4940595 (G/T) variant distinguishes between individuals bearing a functional copy of SERPINB11 with a wild-type c.268G allele and those bearing a nonfunctional copy, with the mutated c.268G>T allele, introducing a stop codon. The extremely high frequency of the mutated allele of SERPINB11 in the general population in different ethnic groups makes the link between this variant and ALF-IE in adults highly unlikely.

Finally, 5 of the 9 other variants located in MUC6, OR6J1, AADACL3, CYP2D7, and KIAA1161 also occur frequently in the gnomAD genome, with an AF up to 40%. For the remaining 4 variants, the results can be considered inconclusive because they correspond to multiallelic polymorphisms which must be referenced precisely, theirs ID being insufficient (Table 1). For instance, rs200886831 of RRP36 detected in 9 patients, refers to 4 different alleles and corresponds to in-frame deletion or insertion of either 6 or 12 bp. One of the 4 alleles has an AF of 45% in gnomAD, whereas the other 3 are extremely rare.

Thus, the conclusion raised by Rakela et al. on the significant association of the 12 variants with ALF-IE in adults is incompatible with careful examination of recent databases. A careful processing of exome and genome data is necessary to find out genes causing ALF-IE in adults.

CONFLICTS OF INTEREST

Guarantor of the article: Micheline Misrahi, MD, PhD.

Specific author contribution: A.H. performed reanalysis and interpretation of published computational data and wrote the letter. M.M. supervised the study, wrote, and revised the manuscript.

Financial support: None to report.

Potential competing interests: None to report.