Expanded classification of hepatitis C virus into 7 genotypes and 67 subtypes: updated criteria and genotype assignment web resource.

UNLABELLED: The 2005 consensus proposal for the classification of hepatitis C virus (HCV) presented an agreed and uniform nomenclature for HCV variants and the criteria for their assignment into genotypes and subtypes. Since its publication, the available dataset of HCV sequences has vastly expanded through advancement in nucleotide sequencing technologies and an increasing focus on the role of HCV genetic variation in disease and treatment outcomes. The current study represents a major update to the previous consensus HCV classification, incorporating additional sequence information derived from over 1,300 (near-)complete genome sequences of HCV available on public databases in May 2013. Analysis resolved several nomenclature conflicts between genotype designations and using consensus criteria created a classification of HCV into seven confirmed genotypes and 67 subtypes. There are 21 additional complete coding region sequences of unassigned subtype. The study additionally describes the development of a Web resource hosted by the International Committee for Taxonomy of Viruses (ICTV) that maintains and regularly updates tables of reference isolates, accession numbers, and annotated alignments (http://talk.ictvonline.org/links/hcv/hcv-classification.htm). The Flaviviridae Study Group urges those who need to check or propose new genotypes or subtypes of HCV to contact the Study Group in advance of publication to avoid nomenclature conflicts appearing in the literature. While the criteria for assigning genotypes and subtypes remain unchanged from previous consensus proposals, changes are proposed in the assignment of provisional subtypes, subtype numbering beyond "w," and the nomenclature of intergenotypic recombinant.
CONCLUSION: This study represents an important reference point for the consensus classification of HCV variants that will be of value to researchers working in clinical and basic science fields.

Soon after the publication of the first nearly complete genome sequence of hepatitis C virus (HCV) in 1989,1 it became apparent that isolates from different individuals or countries showed substantial genetic diversity. After much research and surveying by groups worldwide, this variation was summarized and variants assigned as genotypes and subtypes in a consensus classification and nomenclature system and formal rules were agreed for the assignment and naming of future variants.2 Genotype and subtype assignments required: (1) one or more complete coding region sequence(s); (2) at least three epidemiologically unrelated isolates; (3) a phylogenetic group distinct from previously described sequences; (4) exclusion of intergenotypic or intersubtypic recombination, whether the components were classified or not.

The application of these criteria confirmed the assignment of six distinct genotypes, comprising 18 subtypes. In addition, 58 subtypes were provisionally assigned pending the availability of a complete coding region sequence or additional isolates. This agreement on nomenclature was mirrored by the establishment of several curated databases that organized HCV sequences as they became available and indicated which genotypes and subtypes were confirmed or provisionally assigned (Los Alamos HCV Sequence Database,3 euHCVdb,4 Hepatitis Virus Database: http://s2as02.genes.nig.ac.jp/). Concurrently, a proposal was made to unify the numbering of HCV with reference to the genotype 1a isolate H77 (AF009606).5

Recently, this remarkable agreement and cooperation in HC>V nomenclature has been complicated by several developments. None of the HCV sequence databases are now actively curated and responsibility for naming new genotypes and subtypes has reverted de facto to individual researchers. This, combined with publication delays, has created new contradictions in which isolates assigned to the same subtype (4b: FJ462435, FJ025855, FJ025856, and FJ025854; 6k: DQ278891 and DQ278893; 6u: EU408330, EU408331, and EU408332) belong to different subtypes according to the consensus criteria.2 Another challenge is that the number of complete coding region sequences has increased from 238 in 2005 to more than 1,300. Similarly, the number of variants matching the criteria for assignment as confirmed genotypes/subtypes has expanded from 18 to 67; several recent publications contain figures that are illegible with regard to isolate name and/or accession number,6–10 complicating subsequent comparisons.

Finally, advances in sequencing technology have accelerated the rate at which HCV sequences are generated. Recent articles have reported the partial sequences of 282 isolates from Vietnam11 and 393 isolates from China,10 in each case identifying additional subtypes of genotype 6. Technological advances have also made it easier to obtain HCV complete coding region sequences through both dideoxysequencing and pyrosequencing. The latter technique was recently used to obtain 31 complete coding region sequences belonging to 13 different subtypes.8 More than 225,000 HCV sequences are now available on GenBank and about 30,000 added every year. This volume of sequence information and the diversity of known HCV variants make it increasingly important for researchers to have a single curated resource to refer to for accurate subtype designations, reference genomes and alignments.

This article updates the genotype and subtype assignments2,7 and the nomenclature rules, and describes the establishment of a reference Website hosted by the International Committee for the Taxonomy of Viruses (ICTV) to validate new genotype and subtype assignments, and provide updated reference alignments.

Revision of Confirmed Genotypes and Subtypes

Unique HCV complete or nearly complete coding region sequences available on NCBI Genome (969 sequences, http://www.ncbi.nlm.nih.gov/genome) and the Los Alamos HCV sequence database (1,364 sequences >8,000 nt from http://hcv.lanl.gov/content/index) were aligned within SSEv1.112 using Muscle v3.8.3113 and refined manually. Phylogenetic analysis of sequences containing >95% of the coding region reveals seven major phylogenetic groupings corresponding to genotypes 1-7 (Fig. 1). Within these genotypes, grouping of the constituent subtypes is supported by 100% of bootstrap replications.

Fig. 1

Phylogenetic tree of 129 representative complete coding region sequences. Up to two representatives of each confirmed genotype/subtype were aligned (together with a third extreme variant of subtypes 4g and 6e) and a neighbor joining tree constructed using maximum composite likelihood nucleotide distances between coding regions using MEGA5.83 Sequences were chosen to illustrate the maximum diversity within a subtype. Tips are labeled by accession number and subtype (*unassigned subtype). For genotypes 1, 2, 3, 4, and 6, the lowest common branch shared by all subtypes and supported by 100% of bootstrap replicates (n = 1,000) is indicated by ·.

Based on the consensus criteria,2 confirmed subtypes (indicated by a letter following the genotype) require a complete or nearly complete coding region sequence differing from other sequences by at least 15% of nucleotide positions and sequence information from at least two other isolates in core/E1 (>90% of the sequence corresponding to positions 869 to 1,292 of the H77 reference sequence [accession number AF009606] numbered according to reference5) and NS5B (>90% of positions 8,276 to 8,615) (Table 1). The use of a 15% threshold over the complete coding region is supported by analysis of the large number of potential subtypes now sequenced (Fig. 2). This reveals major and consistently placed gaps in the distribution of pairwise distances between and within subtypes of each genotype as follows: genotype 1: 12.9%-17.0%, genotype 2: 13.1%-17.6%, genotype 3: 12.5%-19.6%, genotype 4: 12.7%-15.3% (except distances of 14% and 14.2% between JX227963 and two subtype 4g sequences), and genotype 6: 9.9%-14.9% (except distances of 13.1%-13.7% between EU246931 and three subtype 6e sequences). Hence, for all genotypes and with remarkably few exceptions, a clear division can be made between isolates that differ by <13% over their complete coding region sequences (members of the same subtype) and those that differ by >15% (different genotypes or subtypes). This analysis includes sequences distinct from any of the confirmed HCV subtypes but not currently represented by three or more independent isolates that remain unclassified subtypes (Table 2). Whether the exceptions noted are due to technical problems or to differing epidemiological histories is unknown.

Table 1

Confirmed HCV Genotypes/Subtypes

Genotype*	Locus/Isolate(s)†	Accession number(s)	Reference(s)
Genotype 1
1a	HPCPLYPRE, HPCCGAA	M62321, M67463	29,30
1b	HPCJCG, HPCHUMR	D90208, M58335	31,32
1c	HPCCGS, AY051292	D14853, AY051292	33
1e	148636	KC248194	9
1g	1804	AM910652	34
1h	EBW443, EBW9	KC248198, KC248199	9
1l	136142, EBW424	KC248193, KC248197	9
Genotype 2
2a	HPCPOLP, JFH-1	D00944, AB047639	35,36
2b	HPCJ8G, JPUT971017	D10988, AB030907	37,38
2c	BEBE1	D50409	39
2d	QC259	JF735114	40
2e	QC64	JF735120	40
2i	D54	DQ155561	41
2j	C1799, QC232	HM777358 JF735113	6,40
2k	VAT96	AB031663	42
2m	QC178, BID-G1314	JF735111, JX227967	40,8
2q	963, 852	FN666428, FN666429	43
2r	QC283	JF735115	40
Genotype 3
3a	HPCEGS, HPCK3A	D17763, D28917	44,45
3b	HPCFG	D49374	46
3g	BID-G1243, QC260	JX227954, JF735123	8,21
3h	QC29	JF735121	21
3i	IND-HCV, BID-G1244	FJ407092, JX227955	8
3k	HPCJK049E1, QC105	D63821, JF735122	47,21
Genotype 4
4a	ED43	Y11604	48
4b	QC264	FJ462435	16
4c	QC381	FJ462436	16
4d	03-18, QC382	DQ418786, FJ462437	49,16
4f	IFBT88, PS6	EF589161, EU392175	50,51
4g	QC193	FJ462432	16
4k	PS3, QC383	EU392173, FJ462438	51,16
4l	QC274	FJ839870	16
4m	QC249	FJ462433	16
4n	QC97	FJ462441	16
4o	QC93	FJ462440	16
4p	QC139	FJ462431	16
4q	QC262	FJ462434	16
4r	QC384	FJ462439	16
4t	QC155	FJ839869	16
4v	CYHCV073, BID-G1248	HQ537009, JX227959	52,8
4w‡	P212, P245	FJ025855, FJ025856	14
Genotype 5
5a	EUH1480, SA13§	Y13184, AF064490	53,54
Genotype 6
6a	EUHK2,6a33	Y12083, AY859526	55,56
6b	Th580	D84262	57
6c	Th846	EF424629	58
6d	VN235	D84263	57
6e	GX004	DQ314805	59
6f	C-0044	DQ835760	60
6g	HPCJK046E2	D63822	47
6h	VN004	D84265	57
6i	Th602	DQ835770	60
6j	Th553	DQ835769	60
6k	VN405	D84264	57
6l	537796	EF424628	58
6m	B4/92	DQ835767	60
6n	KM42, D86/93	DQ278894, DQ835768	17,60
6o	QC227	EF424627	58
6p	QC216	EF424626	58
6q	QC99	EF424625	58
6r	QC245	EU408328	61
6s	QC66	EU408329	61
6t	VT21, D49	EF632071, EU246939	62,19
6u	D83	EU246940	19
6v	NK46, KMN-02	EU158186, EU798760	62,63
6w	GZ52557, D140	DQ278892, EU643834	17,64
6xa‖	DH012, DH028	EU408330, EU408332	18
Genotype 7
7a	QC69	EF108306

Additions and changes from assignments proposed in 2 shown in bold.

Consensus proposed genotype/subtype names. Where multiple sequences of a HCV genotype are available, two sequences have been listed, prioritized by (a) publication date or (b) submission date when unpublished.

Locus (or isolate name if locus is the same as the accession number).

Previously described as 4b.7,14

Sequence obtained from acute phase plasma of a chimpanzee experimentally infected with (human-derived) isolate SA13.

Previously described as 6u.18

Fig. 2

Distribution of p-distances between complete coding region sequences. The frequency of p-distances was calculated within and between genotypes using SSE.12 Intra-genotype pairwise distances were calculated for all available complete coding region sequences except for subtypes 1a, 1b, and 2b where 20 random sequences were used. For p-distances >0.15 (equivalent to a percent difference of 15%), frequencies were scaled to reduce the maximum frequency to less than 300. Distances between genotypes were calculated using one or two representatives of each confirmed and unassigned subtype, with the frequencies scaled as above.

Table 2

Unassigned Complete Coding Region Sequences

Genotype*	Locus/Isolate(s)†	Accession no(s)	Reference
Genotype 1
1_AJ851228	AJ851228	AJ851228	65
1_KC248195	160526	KC248195	9
1_ HQ537007	CYHCV025	HQ537007	52
Genotype 2
2_JF735119	QC331	JF735119	40
2_JF735112	QC182	JF735112	40
2_JF735110	QC114	JF735110	40
2_JF735117	QC297	JF735117	40
2_JF735116	QC289	JF735116	40
2_JF735118	QC302	JF735118	40
Genotype 3
3_JF735124	QC115	JF735124	21
Genotype 4
4_JX227964	BID-G1253	JX227964	8
4_FJ025854‡	P026	FJ025854	14
Genotype 6
6_DQ278891§	KM45,KM41	DQ278891,DQ278893	17
6_JX183550	QC273	JX183550	20
6_JX183552	TV476	JX183552	20
6_JX183549	KM35	JX183549	20
6_JX183551	TV257	JX183551	20
6_JX183553	TV533	JX183553	20
6_JX183554	L349	JX183554	20
6_JX183557	DH027	JX183557	20
6_JX183558	QC271	JX183558	20

Classification of sequences into genotypes but without subtype assignments using the format “genotype_Accession number.”

Locus (or isolate name if locus is the same as the accession number).

Previously described as 4b.14

Previously described as 6k.17

The seven confirmed genotypes (discussed below) comprise 67 confirmed subtypes, 20 provisionally assigned subtypes, and 21 unassigned subtypes. These tables have been posted on the ICTV Website at http://talk.ictvonline.org/links/hcv/hcv-classification.htm and will be updated regularly by the authors with information shared across existing HCV databases (http://hcv.lanl.gov/; http://euhcvdb.ibcp.fr/euHCVdb/), typing tools, and other resources (e.g., http://www.bioafrica.net/rega-genotype/html/subtypinghcv.html; http://comet.retrovirology.lu/; http://hcv.lanl.gov/content/sequence/phyloplace/; http://s2as02.genes.nig.ac.jp/; http://www.viprbrc.org/). Alignments including representatives of these subtypes are available on the ICTV Website and at http://hcv.lanl.gov/content/sequence/NEWALIGN/align.html/.

The process of producing these tables has detected a small number of variants with conflicting assignments. Isolates P026, P212, P245, (FJ025854-6) are described as subtype 4b,14 but these complete coding region sequences show <85% identity to the core/E1 of isolate Z1 (U10235, L16677), provisionally assigned as 4b15 that is more closely related to core/E1 of the complete coding region sequence of isolate QC264 (FJ46243516). P212 and P245 belong to the same, novel subtype for which NS5B sequence is available from a third isolate (P213, GU049362), so this becomes confirmed subtype 4w. Isolate P026 differs from all other genotype 4 sequences by >17.5% but being represented by a single sequence remains currently unassigned (Table 2).

Similarly, isolates KM45 and KM41 (DQ278891,3) have been assigned to subtype 6k,17 but differ by >17% in complete coding region sequence from the subtype 6k isolate VN405 (D84264) and 6.7% from each other, and so remain an unclassified subtype of genotype 6. Two distinct groups of isolates have been assigned to subtype 6u; EU408330-218 and EU246940.19 The latter was submitted first to GenBank and is represented by NS5B sequences from two additional isolates and so is assigned subtype 6u, while EU408330, EU408331, and EU408332 are designated subtype 6xa (see below).

Finally, our analysis of both phylogenetic groupings and sequence distances suggests that a number of isolates20 described in their GenBank accessions as “subtype k-related” (QC273, TV257, TV476, KM35), “subtype l-related” (TV533, L349), “intermediate between subtypes 6m and 6n” (DH027), or “intermediate between subtypes 6j and 6i” (QC271) should be considered as unassigned novel subtypes.

Additional Taxonomic Levels

In making this taxonomic distinction into virus genotypes and subtypes we are aware of the difficulties of imposing a discrete classification scheme on a complex taxonomy. In particular, for genotypes 3 and 6 there are undoubtedly several hierarchies of taxonomic relationships. For example, subtypes 6k and 6l form a clade along with several unassigned genotype 6 isolates.20 A higher-level clade includes these sequences and subtypes 6m and 6n, while a further grouping consists of these subtypes and subtypes 6i and 6j (Fig. 1). These phylogenetic hierarchies are reflected in the discontinuous distribution of p-distances between complete coding region sequences (Fig. 2), which comprises three almost merging distributions (roughly 15% to 20%, 20% to 25%, and 25% to 30%). Three distributions of intersubtype distances were also observed for genotype 3 (20% to 25%, 25% to 27%, and 27% to 30%), two distributions for genotype 2 (18% to 22.5%, 23% to 26.5%), and uniform distributions for genotype 1 (17.7% to 25.4%) and genotype 4 (15.3% to 23.1%). However, the internal divisions defined by the multiple distributions of distances within genotypes 2, 3, and 6 have not been shown to correspond with geographical or epidemiological differences. The higher-level grouping of subtypes 3b, 3g, and 3i does not reflect a common geographical origin distinct from that of 3h and 3k.21 There is also no geographical correlation with the groupings of subtypes 6k, 6l, and various unassigned isolates; for 6m, 6n, and an unassigned isolate; for 6h, 6i, 6j, and an unassigned isolate; for 6a and 6b; for 6f and 6r; or for 6r and 6e.22 Similarly, there are currently no known virological or clinical reasons to recognize these higher-level groupings. Without practical utility, we therefore propose that the observed within-genotype hierarchies are not given any formal recognition in their nomenclature.

Proposed Updates and Changes to Rules for Genotype/Subtype Assignments

Subtype Names

By definition, subtype name assignments would be limited to a maximum of 26 if designated by a single letter suffix (e.g., 2a-2z). We therefore suggest that subtypes are assigned up to the letter “w” and subsequent designations follow the eXtended form xa, xb, … xz, in turn followed by ya, … yz, za, … zz, potentially giving a total of 101 subtypes of each genotype. This avoids potentially ambiguous terms such as “subtype 6x,” which could be interpreted as “genotype 6 of unknown subtype,” or designations such as “subtype 3aa,” which might suggest a relationship with 3a.

Provisional Genotypes

According to the 2005 consensus classification protocol2 new genotypes could be provisionally assigned from a single complete coding region sequence, but partial or complete coding region sequences from additional isolates would be required to confirm these assignments. Since then only one provisional genotype has been identified (7a) represented by a single isolate (QC69, EF108306). Thus, in contrast to subtype assignments, the number of genotypes appears relatively limited and the requirement to sequence multiple isolates now seems over-onerous. We propose that only a single complete coding region sequence is needed to confirm a new genotype assignment; QC69 is therefore confirmed as genotype 7a.

Provisional Subtypes

The 2005 consensus protocol also proposed that provisional subtypes could be assigned on the basis of sequence comparisons in the core/E1 and NS5B regions for at least three independent isolates, requiring in addition a complete coding region sequence before being confirmed. Of the 58 subtypes provisionally assigned in the 2005 article, 38 have now been confirmed (Table 1). However, it is now much easier to obtain complete coding region sequences and very few additional provisional subtypes have been proposed. Instead, some authors have inconsistently labeled unusual isolates with the suffix “?,” “unassigned group I”11,23 or “subtype 1(I).”9 We propose that provisional subtype designations should no longer be provided for variants where complete genome sequences are lacking. The 20 remaining provisionally assigned subtypes will be maintained (Table 3), since they already exist in the literature. Future subtype assignments will only be made (as confirmed assignments) when sequence data from three or more isolates including at least one complete or nearly complete coding region is provided. Where a complete coding region sequence is available but there are fewer than three isolates, we propose that these remain unassigned. In Table 2 these are labeled using the form “Genotype_Accession number,” e.g., 1_AJ851228.

Table 3

Remaining Provisionally Assigned HCV Subtypes

		Accession number(s)*
	Isolate††	Core/E1	NS5B	Reference(s)
Genotype 1
1d	HC1-N15, HC1-N16	L39299, L39302	L38377, L38372	66
1f	FR2	L38350	L38371	66
1i	FR16, QC77	n.a., AY434119	L48495, AY434120	67,68
1j	QC2, QC89	AY434158, AY434128	AY434106, AY434129	67
1k	QC68, QC82	AY434112, AY434122	AY434113, AY434123	67
Genotype 2
2f	JK081, JK139	D49754, D49757	D49769, D49777	47
2g	MED017	n.a.	X93323	69
2h	MED007	n.a.	X93327	69
2l	FR15	n.a.	L48494	68
2n	NL50	L39309	L44602	66
2o	FR4	L38333	L38373	66
2p	NL33	L39300	L44601	66
Genotype 3
3c	NE048	D16612	D14198/D16613	70
3d	NE274	D16620	D14200/D16621	70
3e	NE145	D16618	D16619	70
3f	NE125, PK64	D16614, n.a.	D14203/D16615, L78842	70,71
Genotype 4
4e	CAM600, GB809	L29589, L29629	L29590, L29626	72
4h	GB438, FrSSD35	L29610, n.a.	L29611, AJ291249	72,73
4i	CAR4/1205	L36439	L36437	74
4j	CAR1/501	n.a.	L36438	74

Accession numbers of sequences from the core/E1 and NS5B regions. “n.a.”: not available; “/”: denotes that the core/E1 or NS5B sequences are available from two different accession numbers.

Examples of each provisionally assigned HCV.

Recombinant and Other Forms

One issue that was not addressed in the 2005 consensus protocol2 was the naming of the newly discovered recombinant forms of HCV, their importance being unknown. Nine different recombinant forms of HCV have now been described (Table 4), of which only one (2k/1b) is represented by multiple isolates; no multiple recombinants have been reported (reviewed in reference24). In this context it does not seem necessary to revise the nomenclature generally used in the literature in which “RF” (recombinant form) is followed by the contributory subtypes separated by “/” in the order in which they appear in the complete genome sequence. We suggest that recombinant forms with the same genotypic structure but with different breakpoints or where the component genomic sections are unrelated are numbered consecutively with a numerical suffix (for example, RF2b/1b_1).

Table 4

Recombinant (RF) HCV Complete Coding Region Sequences

RF*	Breakpoint†	Accession	Isolates‡	Reference
RF2k/1b	3186	AY587845	33	75–77
RF2i/6p	3405-3464	DQ155560	1	41
RF2b/1b_1	3456	DQ364460	1	78
RF2/5	3366-3389	AM408911	1	79
RF2b/6w	3429	EU643835	1	64
RF2b/1b_2	3432	AB622121	1	80
RF2b/1a	3429-3440	JF779679	1	81
RF2b/1b_3	3286-3293	AB677530	1	82
RF2b/1b_4	3286-3293	AB677527	1	82

Recombinant forms (RF) for which complete genome sequences are available are named according to the subtypes from which they are derived and in the order in which these appear in the genome.

Breakpoints are numbered with reference to H77 (AF009606).

Number of individuals from whom the RF has been isolated.

Proposals for New Genotype/Subtype Assignments

The ICTV Flaviviridae Study Group is willing to take a coordinating role in the assignment of newly described variants of HCV. We urge researchers who have characterized new HCV variants that potentially qualify as new types or subtypes to contact Donald Smith (D.B.Smith@ed.ac.uk) or any member of the Study group (listed on http://ictvonline.org/subcommittee.asp?committee=25&se=5) in confidence before publication so that naming conflicts can be avoided and appropriate assignments made.

Future Developments

Despite the increasing number and diversity of HCV sequences, the system of classification of variants into genotypes and subtypes has proven surprisingly robust. The seven confirmed genotypes have strong bootstrap support (Fig. 1), and the partition of these genotypes into subtypes that differ over a complete coding region sequence by >15% reflects a natural hiatus in the distribution of sequence distances (Fig. 2). We welcome any comments or suggestions for the proposed classification guidelines. Areas of uncertainty remain with respect to the region of endemicity of genotype 5, represented by a single subtype isolated in Europe, Brazil, North Africa, and South Africa, and genotype 7, isolated from an emigrant from the Congo. We might also anticipate the further discovery of other HCV-like viruses in the genus Hepacivirus,25–28 and variants closer genetically to HCV than the nonprimate hepacivirus that appears to be an endemic infection of horses worldwide.25 As more is learned about the host-specificity and diversity of hepaciviruses, the genotype classification of HCV may be logically incorporated within a unified classification of hepaciviruses at the species and potentially subspecies and subgenus levels.