The histogenesis of regenerative nodules in human liver cirrhosis


  • Wey-Ran Lin,

    1. Centre for Diabetes, Barts and The London School of Medicine and Dentistry, Queen Mary University of London, London, UK
    2. Department of Gastroenterology and Hepatology, Chang Gung University College of Medicine, Taipei, Taiwan
    Search for more papers by this author
  • Siew-Na Lim,

    1. Centre for Neuroscience and Trauma, Barts and The London School of Medicine and Dentistry, Queen Mary University of London, London, UK
    2. Department of Neurology, Chang Gung Memorial Hospital, Chang Gung University College of Medicine, Taipei, Taiwan
    Search for more papers by this author
  • Stuart A. C. McDonald,

    1. Histopathology Unit, London Research Institute, Cancer Research UK, London, UK
    2. Centre for Digestive Diseases, Barts and The London School of Medicine and Dentistry, Queen Mary University of London, London, UK
    Search for more papers by this author
  • Trevor Graham,

    1. Histopathology Unit, London Research Institute, Cancer Research UK, London, UK
    Search for more papers by this author
  • Victoria L. Wright,

    1. Centre for Diabetes, Barts and The London School of Medicine and Dentistry, Queen Mary University of London, London, UK
    Search for more papers by this author
  • Claire L. Peplow,

    1. Centre for Diabetes, Barts and The London School of Medicine and Dentistry, Queen Mary University of London, London, UK
    Search for more papers by this author
  • Adam Humphries,

    1. Histopathology Unit, London Research Institute, Cancer Research UK, London, UK
    Search for more papers by this author
  • Hemant M. Kocher,

    1. Tumour Biology Laboratory, John Vane Science Centre, Barts and The London School of Medicine and Dentistry, Charterhouse Square, London, UK
    Search for more papers by this author
  • Nicholas A. Wright,

    1. Centre for Diabetes, Barts and The London School of Medicine and Dentistry, Queen Mary University of London, London, UK
    2. Histopathology Unit, London Research Institute, Cancer Research UK, London, UK
    Search for more papers by this author
  • Amar P. Dhillon,

    1. Department of Histopathology, University College London Medical School, Royal Free Campus, London, UK
    Search for more papers by this author
  • Malcolm R. Alison

    Corresponding author
    1. Centre for Diabetes, Barts and The London School of Medicine and Dentistry, Queen Mary University of London, London, UK
    • Centre for Diabetes, Blizard Institute of Cell and Molecular Science, Barts and The London School of Medicine and Dentistry, 4 Newark Street, London E1 2AT, UK
    Search for more papers by this author
    • fax: + 44 207 882 2186.

  • Potential conflict of interest: Nothing to report.


Here, we investigate the clonality and cells of origin of regenerative nodules in human liver cirrhosis using mitochondrial DNA (mtDNA) mutations as markers of clonal expansion. Mutated cells are identified phenotypically by deficiency in the entirely mtDNA encoded cytochrome c oxidase (CCO) enzyme by histochemical and immunohistochemical methods. Nodules were classified as either CCO-deficient or CCO-positive, and among 526 nodules from 10 cases, 18% were homogeneously CCO-deficient, whereas only 3% had a mixed phenotype. From frozen sections, hepatocytes were laser-capture microdissected from several sites within individual CCO-deficient nodules. Mutations were identified by polymerase chain reaction sequencing of the entire mtDNA genome. In all cases except for one, the nodules were monoclonal in nature, possessing up to four common mutations in all hepatocytes in a given nodule. Moreover, the identification of identical mutations in hepatic progenitor cells abutting CCO-deficient nodules proves that nodules can have their origins from such cells. Conclusion: These data support a novel pathway for the monoclonal derivation of human cirrhotic regenerative nodules from hepatic progenitor cells. (HEPATOLOGY 2010;51:1017–1026.)

Regenerative nodules (RNs) of hepatic parenchyma surrounded by fibrotic bands are a defining pathological feature in liver cirrhosis, and represent a final common pathway of many chronic liver diseases.1 RNs are generally considered to form through localized proliferation of hepatocytes and their entrapment by deposition of extracellular matrix.2 A number of attempts have been made to establish the clonality of RNs, often by examination of X-chromosome–linked markers in females. For example, in hepatitis C virus–induced liver cirrhosis, approximately half of all RNs have been found to be monoclonal, on the basis of restriction fragment length polymorphism of the X-chromosome–linked phosphoglycerokinase gene and on the random inactivation of the gene by methylation.3 A similar result was obtained by studying the pattern of inactivation of the X-linked human androgen receptor gene (HUMARA), with 45% of RNs being monoclonal.4 However, neither of these studies takes into account the distribution of X-inactivated cells in the liver. Because X-inactivation (Lyonization) occurs relatively early in embryogenesis, the progeny of a single X-inactivated embryonic cell may be clustered together. Groups of cells sharing a common X-inactivation pattern are referred to as “patches”, and in the human colon, such patches may extend across up to 400 crypts.5 Thus in the liver, a RN arising in the middle of a patch would be monophenotypic for an X-linked marker, but could be either monoclonally or polyclonally derived.

Whereas the monoclonality or otherwise of RNs is one issue, a second concerns the cells of origin of RNs. Apart from hepatocytes, hepatic progenitor cells (HPCs) that form ductular reactions (DRs) could also give rise to RNs. It is well recognized that during chronic liver injury, with a reduction in hepatocyte proliferation, a potential stem cell compartment is activated from within the smallest branches of the intrahepatic biliary tree, giving rise to the so-called DRs.6 The extent of this reaction is dependent on the severity of the damage, in both humans7 and mice.8 HPCs that comprise DRs can express markers of both biliary and hepatocyte lineages in rodents9; likewise in human cirrhotic liver, both hepatocytic (HepPar1) and biliary markers (cytokeratin 19 [CK19], neural cell adhesion molecule [NCAM]) are expressed in DRs.10 Moreover, in an elegant three-dimensional study of CK19 staining of human cirrhotic liver, it was shown that small “buds” of hepatocytes in between connective tissue septae appeared to be directly descended from CK19-positive HPCs.11 Whether these nascent hepatocyte buds can evolve into large regenerative nodules is unclear, but this study would suggest that this is entirely possible.

Recently, we described the utility of using mitochondrial DNA (mtDNA) mutations as clonal markers for lineage tracing in normal human liver,12 finding that monoclonal proliferative units appear to extend from the portal rim to the hepatic vein. Such mutations become established in stem cells, and we have identified clonal populations not only in the liver, but also in the small and large intestines and stomach.12–16 Approximately 16.6 kilobases in length, mtDNA is a double-stranded circular DNA molecule that encodes 13 essential proteins involved in mitochondrial oxidative phosphorylation, 22 ribosomal RNAs, and two transfer RNAs (tRNAs). Mitochondria are responsible for generating adenosine triphosphate and cells contain multiple mitochondrial genomes. The mitochondrial genome is prone to mutation not only because it has no protective histones and limited repair mechanisms, but also because it resides in an environment of high oxidase stress.17 Mutations can be found in all mtDNA copies within a cell (homoplasmy) or only a proportion (heteroplasmy). We can observe cells with either a homoplasmic mtDNA mutation or at least a high level of mutation (>80% heteroplasmy) by detecting cytochrome c oxidase (CCO) activity, an entirely mtDNA-encoded enzyme that forms complex IV of the electron transport chain. This stochastic mutation and expansion is a lengthy process, often taking many years,18 so longstanding cells, presumably stem cells, are the only cells that have a sufficient lifespan to acquire mutation levels that result in a detectable biochemical deficiency.

In the human colon, CCO-deficient crypts are rare before the age of 40 years,14 and because cells migrate from the stem cell zone to the luminal surface to be sloughed off in a matter of days, the source of CCO deficiency is undoubtedly the stem cell population at the crypt base. In the liver, although the founder cell of any patch of CCO-deficient cells is also likely to be a long-lived cell, this might not necessarily be a stem cell if, as we know, cell turnover is considerably slower. However, the large cell patches we observe, each with a unique mutation(s), strongly suggests the founder cell is endowed with clonogenicity—a key feature of “stemness”.

In our previous studies12 we found patches of CCO-deficient hepatocytes, and mtDNA sequencing of many individual hepatocytes within a particular patch revealed that these hepatocytes shared a unique mutation indicating a common cell of origin. The fact that the patches invariably abutted the portal rim led us to speculate that each CCO-deficient population was clonally derived from a clonogenic (stem) cell in this location, in line with the “streaming liver” hypothesis.

In this article, we show that most RNs are either uniformly CCO-positive or CCO-deficient, and less than 3% are CCO-mixed nodules, suggesting that most RNs could be monoclonal. Analysis of mtDNA sequences revealed that within a given CCO-deficient RN, all cells harbored the same mutation(s), which proves monoclonality. Furthermore, HPCs within abutting CCO-deficient DRs had the same mutations as the adjacent RN, indicating a common cell origin. As far as we know, this is the first direct evidence that RNs in cirrhosis can be derived from HPCs.


CCO, cytochrome c oxidase; CK19, cytokeratin 19; DR, ductular reaction; HPC, hepatic progenitor cell; IHC, immunohistochemistry; mtDNA, mitochondrial DNA; PBS, phosphate-buffered saline; RN, regenerative nodule; SDH, succinate dehydrogenase.

Patients and Methods


Fourteen formalin-fixed, paraffin-embedded blocks and nine frozen tissue blocks of human cirrhotic liver of varying etiology (hepatitis B, hepatitis C, alcoholic liver disease, Wilson disease, nonalcoholic steatohepatitis, and cryptogenic cirrhosis) were studied. Multicenter ethical approval was obtained according to the requirements of the United Kingdom Human Tissue Act (2006) from the Redbridge and Waltham Forest Local Research Ethical Committee (REC) (Reference 2006/Q0601/33).


The mouse monoclonal anti-Oxphos Complex IV subunit I antibody (anti-CCO1) was purchased from Molecular Probes Invitrogen (Paisley, UK). Anti–cytokeratin 19 (anti-CK19) mouse monoclonal antibody was purchased from Dako (Glostrup, Denmark).


Immunohistochemistry (IHC) staining was performed either on formalin-fixed, paraffin-embedded sections (4 μm thick) or on 15-μm thick frozen sections. Paraffin sections were dewaxed in xylene, and endogenous peroxidase was blocked by incubation in 0.18% hydrogen peroxide in methanol, followed by rehydration through graded alcohols to phosphate-buffered saline (PBS). Sections were then microwaved for 10 minutes in boiling sodium citrate buffer (pH 6.0), cooled in water, and rinsed in PBS. The frozen sections were fixed in cold acetone, incubated in hydrogen peroxide in methanol, and then rinsed in PBS. Sections first were treated with an avidin-biotin blocking kit (Dako, Glostrup, Denmark) to block endogenous avidin and biotin and then were incubated in rabbit serum (1:25; Dako) based on the secondary antibody for blocking nonspecific binding. Primary antibodies were applied for 1 hour at room temperature in a humid chamber. Primary antibody dilutions were: anti-CCO1, at 1:400 and anti-CK19 at 1:100. Sections were washed three times for 5 minutes each in PBS and then incubated for 40 minutes at room temperature with appropriate secondary antibodies conjugated to biotin. After washing, sections were incubated for 40 minutes with a tertiary layer of streptavidin–horseradish peroxidase (1:500; Dako). Peroxidase activity was revealed using 4 mmol/L 3,3-diaminobenzidine as a chromogen in PBS containing 0.2% hydrogen peroxide. Sections were counterstained with Mayer's hematoxylin and then dehydrated through ascending alcohols, cleared in xylene, and mounted in DePeX (BDH, Poole, UK) resinous mounting medium.

Enzyme Histochemistry for Cytochrome c Oxidase/Succinate Dehydrogenase.

Frozen sections were cut at a thickness of 15 μm. Sequential CCO and succinate dehydrogenase (SDH, the presence of which was used to highlight any deficiencies in CCO) histochemistry was performed, as previously described.16 Briefly, sections were air-dried for 30 minutes and then incubated in CCO medium containing 100 mM cytochrome c, 20 mg/mL catalase, and 4 mmol/L diaminobenzidine tetrahydrochloride in 0.2 mol/L phosphate buffer (pH 7.0), all sourced from Sigma-Aldrich (Poole, UK), for 30 minutes at 37°C. Sections were then washed three times in PBS buffer for 5 minutes each and incubated in SDH medium (130 mmol/L sodium succinate, 200 mmol/L phenazine methosulfate, 1 mmol/L sodium azide, and 1.5 mmol/L nitroblue tetrazolium in 0.2 mol/L phosphate buffer, pH 7.0) for 1 hour at 37°C. Sections again were washed in PBS three times for 5 minutes each and dehydrated in ascending ethanols, cleared in Histoclear (Lamb Laboratory Supplies, Eastbourne, UK), and mounted with Permount (Fisher Scientific, Fairlawn, NJ).

Isolation of DNA from Individual Cells.

Frozen sections (15 μm thick) were cut and mounted onto ultraviolet-irradiated P.A.L.M. membrane slides (P.A.L.M. Microlaser Biotechnologies, Bernried, Germany). Cells were laser-capture microdissected using a P.A.L.M. microscope (Zeiss, Bernried, Germany) and collected in sterile, ultraviolet-irradiated 0.5 mL AdhesiveCap (Zeiss) tubes. We took hepatocytes from CCO-deficient nodules, CCO-positive nodules, and cells from surrounding CCO-deficient DRs, the latter being confirmed by the next section being immunostained for CK19. The color boundaries of CCO-deficient and CCO-positive cells (blue and brown, respectively) were easily distinguished. The collection tubes were then centrifuged at 7000g for 10 minutes, the cells were lysed in 14 μL lysis buffer (Picopure, Arcturus, CA) at 65°C for 3 hours and then denatured at 95°C for 10 minutes.

mtDNA Sequencing.

From every cell area laser-captured, we sequenced the entire mitochondrial genome. A two-round amplification method was followed, whereby the first round consisted of amplifying nine fragments spanning the entire genome, and the second round consisted of 36 M13-tailed primer pairs to amplify overlapping segments of the first-round products, as described.14 Polymerase chain reaction products were sequenced by using BigDye version 3.1 Terminator cycle sequencing chemistries on an ABI-Prism 3100 Genetic Analyzer (Applied Biosystems, Foster City, CA) and compared directly with the revised Cambridge reference sequence (rCRS) by using 4Peaks software (Mekentosj BV, Aalsmeer, The Netherlands).


The CCO-positive nodules, CCO-deficient nodules, and CCO-mixed nodules were defined after anti-CCO immunostaining as well-circumscribed nodules with either uniform immunoperoxidase staining, no staining, or a mixture of immunopositive and immunodeficient cells. Ductular phenotype and positive anti–CK19 immunostaining identified the DRs.


IHC Analysis of Cytochrome c Oxidase.

Immunostaining of human cirrhotic liver showed that RNs were composed of either only CCO-positive hepatocytes or CCO-deficient hepatocytes (Fig. 1), whereas RNs composed of both CCO-deficient and CCO-positive hepatocytes (CCO-mixed nodules) were rare (<3%) in our cohort group (Table 1). From our previous experience,12 the presence of wholly CCO-deficient RNs suggests that most RNs are monoclonal in origin. Overall, 10 of the 14 cases of liver cirrhosis had CCO-negative RNs and within these 10 cases, 526 nodules were analyzed. Despite the different etiologies of cirrhosis, most of the RNs showed either homogeneous CCO-immunoreactivity or a complete lack of it, whereas only 2.9% of all RNs were mixed. These nodules varied in size, but all had clear fibrotic boundaries and were surrounded by many DRs. The cells of DRs also had CCO activity; CCO-positive DRs were found mainly around CCO-positive nodules, whereas CCO-deficient DRs could be easily found around CCO-deficient nodules (Fig. 2); however, the finding of CCO-deficient DRs adjacent to CCO-deficient RNs does not in itself indicate the two populations are clonally related, and proof requires the finding of clonal mtDNA mutations as described in Figs. 3–7. Although we found that most DRs were homogeneous in CCO immunoreactivity, some DRs contained clearly demarcated CCO-positive and CCO-deficient cells (Fig. 2C) suggesting that clones can evolve within a DR.

Figure 1.

CCO IHC in cirrhotic liver tissue showing three different types of regenerative nodules. (A) CCO-positive nodule: a nodule contains homogeneous CCO-positive hepatocytes. (B) CCO-deficient nodule: a nodule (denoted by dashed line) is composed of hepatocytes with no hepatocyte immunoreactivity. (C) CCO-mixed nodule: a nodule is composed of both CCO-positive and CCO-deficient hepatocytes.

Table 1. The Analysis of CCO Activity in Cirrhotic RNs with Different Etiologies
EtiologyCase NumberCCO-Deficient Cases*Nodules in CCO-Deficient CasesCCO-Positive Nodules (%)CCO-Deficient Nodules (%)Mixed Nodules (%)
  • *

    Cases with one or more CCO-deficient nodules.

  • The number of nodules was counted from one section of each CCO-deficient case and shown as the total number found for all cases.

  • Alc, alcohol; CCO, cytochrome c oxidase; HBV, hepatitis B virus; HCV, hepatitis C virus; NASH, nonalcoholic steatohepatitis.

HBV77331256 (77.3%)63 (19.1%)12 (3.6%)
Alc416854 (79.4%)13 (19.1%)1 (1.5%)
Alc/HCV218369 (83.1%)13 (15.7%)1 (1.2%)
NASH114435 (79.5%)8 (18.2%)1 (2.3%)
Total1410526414 (78.7%)97 (18.4%)15 (2.9%)
Figure 2.

CCO IHC in cells comprising ductular reactions (DRs) surrounding regenerative nodules. (A) CCO-positive DRs (arrows) abutting a CCO-positive nodule. (B) CCO-deficient DRs (arrowheads) can be seen adjacent to a CCO-deficient nodule. (C) Cells with CCO activity (arrow) and without CCO activity (arrowhead) can be seen in this section of biliary epithelia.

Figure 3.

Regenerative nodules can be monoclonal. (A) An entirely CCO-deficient regenerative nodule (on a laser-capture membrane slide) histochemically stained sequentially for CCO activity (brown) and SDH activity (highlighting CCO-deficiency (blue). (B) Hematoxylin and eosin (H&E) staining illustrating nondysplastic morphology. (C) Three different groups of cells (1, 2, and 3) from the same CCO-deficient nodule and one group of cells (4) from the neighboring CCO-positive nodule were laser-capture microdissected and the entire mitochondrial genome was sequenced. (D) Cytokeratin-19 positivity highlights the surrounding ductular reactions (arrowheads), and red circles denote the areas that had been microdissected. (E) All the CCO-deficient cells contained the same T→C transition mutation at base-pair position 6933 within the MT-CO1 gene, a CCO coding region (black arrow) which was not detected in the neighboring CCO-positive cells (white arrow).

Figure 4.

Closely apposed regenerative nodules can have different cell origins. (A) An entirely CCO-deficient regenerative nodule in the middle (blue, SDH-positive), which is immediately adjacent to another CCO-deficient nodule (left) and to a CCO-positive nodule (right, brown stain). (B) H&E staining highlights the nondysplastic morphology of these regenerative nodules. (C) Two groups of cells (1 and 2) from the central CCO-deficient nodule, one group of cells (3) from the left CCO-deficient nodule and one group of cells (4) from the right CCO-positive nodule were laser capture-microdissected and the entire mitochondrial genome was sequenced. (D) Cytokeratin-19 staining highlights the surrounding ductular reactions (arrowheads), and red circles show the areas that have been microdissected. (E) Cells from areas 1 and 2 from the middle CCO-deficient nodule contained the same T > C transition mutation at position 4313 bps within the MT-T1 gene, a tRNA-Ile coding region (black arrow), but no mutation at locations 7070 bps, 13450 bps and 741 bps (white arrows). Cells from area 3 from the left CCO-deficient nodule contained three different transition mutations: 7070C > T, 13450T > C and 741A > G (black arrowheads), while no mutation was found at position 4313 bps (white arrowhead), the site of mutation in the centre CCO-deficient nodule. No mutation was detected in cells from area 4 from CCO-positive nodule on the right.

Figure 5.

A phenotypically and genotypically mixed nodule. (A) A nodule containing a CCO-deficient area (blue staining), a CCO-positive area (brown staining), and a mixed area. (B) Hematoxylin and eosin (H&E) staining showing nondysplastic morphology. (C) Cell areas 1–3 from the CCO-deficient area, areas 4–5 from the mixed area, and areas 6–7 from the CCO-positive area were laser-capture microdissected and the entire mitochondrial genome was sequenced. (D) Cytokeratin-19 IHC highlights the surrounding ductular reactions (white arrowheads), and the red circles denote the microdissected areas. (E) Cells from areas 1, 2, and 3 from the CCO-deficient area contained the same G→A transition mutation at base-pair position 2690 within the MT-RNR2 gene, a 16S ribosomal RNA coding region (black arrow), whereas cells at 4 and 5 from the CCO-mixed area had a mixture of the same mtDNA mutation (2690G→A) and wild-type DNA (black arrowhead). Cells at 6 and 7 from the CCO-positive area had no mutation (white arrow).

Figure 6.

Mitochondrial DNA genotyping indicates that regenerative nodules and ductular reactions can have a common cell of origin. (A) An entirely CCO-deficient nodule (blue staining). (B) Hematoxylin and eosin (H&E) staining showing nondysplastic morphology. (C) Four groups of cells (1-4) from the same CCO-deficient nodule, three groups of cells (5-7) from the abutting CCO-deficient ductular reactions, which were confirmed by cytokeratin-19 IHC in the next serial section (D, brown staining), and cells (8) from the CCO-positive nodule were laser-capture microdissected and the entire mitochondrial genome was sequenced. (E) Cell areas 1-4 contained the same C insertion mutation between positions 9532 to 9537 within the MT-CO3 gene, a CCO-coding region (black line). Cell areas 5-7 from the abutting CCO-deficient ductular reactions showed the same mutation, whereas cell area 8 from the CCO-positive nodule had no mutation (dash line).

Figure 7.

Mitochondrial DNA genotyping indicates that regenerative nodules can be derived from CK19-positive DRs. (A) An entirely CCO-deficient nodule (blue). (B) Five groups of cells (1-5) from the same CCO-deficient nodule, cells (6) from the adjacent CCO-deficient ductular reactions, confirmed by cytokeratin-19 IHC on the next serial section (C, brown staining), and cells (7) from the CCO-positive nodule (brown) were laser-capture microdissected and the entire mitochondrial genome was sequenced. (D) Cell areas 1-5 all contained four different transition mutations: 2145G→A, 2269G→A, 12362C→T, and 15671A→G (black arrows). (E) Cell area 6 from the abutting CCO-deficient ductular reaction had exactly the same mutations. Heteroplasmy was detected at base-pair locations 2145 and 2269 (arrowheads), whereas the mutations at base-pair locations 12362 and 15671 were homoplasmic (black arrows). (F) Cell area 7 from the CCO-positive nodule had no mutation (white arrows).

RNs Can Be Monoclonal.

Direct mtDNA sequencing of separate hepatocytes from within CCO-deficient RNs demonstrates that each can have a common cell of origin (Figs. 3, 4, 6, and 7), without excluding the possibility, for example, that a monoclonally derived DR could en masse differentiate into a RN. These monoclonal RNs were not dysplastic (Figs. 3B, 4B, and 6B), and indeed were invariably surrounded by DRs (confirmed by serial CK19 IHC; Figs. 3D, 4D, 6D, and 7C). The number of mutations in the mtDNA genome varied from one nodule to the next. Most of the CCO-deficient nodules we analyzed contained only a single mutation in their entire mtDNA genome (Figs. 3E, 4E, and 6E); however, we also found specimens with three mutations (Fig. 4E) and even four mutations (Fig. 7D) within a single CCO-deficient nodule. Because mtDNA mutations most likely require a long time to become established in the population, it is highly probable that these CCO-deficient nodules represent the progeny of a long-lived cell type.

Neighboring RNs Can Have Different Cellular Origins.

Neighboring nodules apparently surrounded by broad connective tissue septae in two-dimensional microscopic sections could in fact be a single entity that merely appears multinodular in histologically sectioned material. To explore this possibility, mtDNA sequencing of three closely opposed nodules was performed (Fig. 4). A CCO-positive nodule had no mtDNA mutations, but of two neighboring CCO-deficient nodules, one possessed a single mutation, while the other had three completely different mtDNA mutations (Fig. 4E). These results illustrate that even nodules located very close to one another are in fact of completely separate cellular inheritance. Furthermore, we could easily find CCO-positive nodules next to CCO-deficient nodules (Figs. 1B, 2B, 3A, 6A, and 7A), again illustrating the multifocal nature of the cirrhotic process.

mtDNA Genotyping of a Phenotypically Mixed Nodule.

Figure 5A illustrates a RN with an area of CCO activity (brown histochemical stain) that sharply contrasts with an area of CCO deficiency (SDH activity, blue stain), with a further area that has a mixture of CCO-positive and CCO-deficient cells. Within the CCO-deficient area, a clonal mtDNA mutation was present, but within the CCO-positive area there was only wild-type mtDNA, whereas both wild-type and mtDNA was present in the sample taken from the mixed area. Therefore, this mixed nodule must be derived from more than one cell, although the nodule could conceivably be in the process of monoclonal conversion if the mtDNA-mutated clone was displacing or being displaced by the wild-type clone.

The Relationship of Regenerative Nodules to Ductular Reactions.

An RN with a single identical mtDNA in all its cells (Fig. 6E), confirming its monoclonal origin, also had an abutting DR (Fig. 6D). Sequencing of cells from CCO-deficient DR showed the same mutation that was present in the RN, demonstrating that the RN and abutting DR have a common inheritance (Fig. 6E). Thus, an origin of the RN from the DR is likely, although it could be argued that the DR could originate from metaplasia of hepatocytes within the DR. Analyzing a further RN with an associated DR (Fig. 7), we again confirmed monoclonality (Fig. 7E), this time with four identical homoplasmic mutations throughout the nodule. Moreover, these same four mutations were also present in the CK19-positive DR, but two of them were heteroplasmic. These observations strongly suggest that the hepatocytes in the RN have an origin from the DR and not vice versa because cells with a totally homoplasmic mutation (the hepatocytes) cannot revert to partially wild-type (the hepatic progenitor cells comprising the DR) characteristics.


In cirrhosis, RNs are often accompanied by a prominent DR, with perhaps the notable exception of primary biliary cirrhosis. DRs represent the progeny of a facultative stem cell compartment thought to be located in the canals of Hering, which extend into the parenchyma in humans.19 In animal models, this stem cell compartment becomes activated in times of regenerative demand when the ability of hepatocytes to enter the cell cycle is compromised. Thus, their presence in cirrhosis is entirely consistent with this theory, and indeed many studies of chronic hepatitis C virus infection20, 21 and hepatitis B virus infection22 have charted an increasing incidence of hepatocyte senescence with increasing fibrosis, with senescence being assessed by expression of p21CIP1 and senescence-associated β-galactosidase. Whether RNs are formed simply by fibrotic dissection of the existing parenchymal cells or whether they arise de novo from individual stem/progenitor cells is unknown.

We have highlighted that RNs can be formed by monoclonal expansion, and not simply by fibrotic dissection of pre-existing liver parenchyma. Furthermore, we demonstrated for the first time that HPCs within the abutting DRs can have the same cell of origin as the monoclonal RN. Whether a monoclonal population of HPCs differentiates en masse into a population of nodular hepatocytes or if only a single or a few HPCs differentiate prior to expansion is unclear.

Previous studies have alluded to the monoclonal nature of many RNs, based on patterns of inactivation of X-linked genes, but as previously mentioned, these studies are confounded by a lack of knowledge concerning the patch size in human liver. In our study, we have used mtDNA mutations as clonal markers for tracing cell lineages. The monoclonal origins of CCO-deficient RNs are based on the minimal odds of finding two random identical mutations occurring at the same position in two separate cells being calculated as 2.48 × 109:1; moreover, we had RNs with three and even four identical mutations in all cells (Figs. 4E and 7D).

Studies of human colonic crypts have suggested that it takes up to 40 years for a colonic crypt cell to become homoplasmic or at least display a high degree of heteroplasmy for a particular mutation,14 and our own studies indicate that the mutant cells appear to be located close to the stem cell niche, at the bottom of the colonic crypt,13 at a ductular location in the pancreas,13 and close to the portal rim in the liver.12 Thus, it seems highly reasonable to suppose that many of the mutations found in the hepatocytes of cirrhotic regenerative nodules have their origins in the potential stem cells located in the canals of Hering. It has been suggested that the finding of monoclonality may help separate purely regenerative nodules from those with malignant potential4; however, no firm conclusions were drawn from the study. Whether the RNs we studied had malignant potential is unknown, but histological examination revealed no dysplastic features.

Extracellular matrix deposition and activation of matrix-producing cells occurs as an early phase of chronic liver injury,23 and it may be that the fibrotic environment is important for establishing the niche for the process of activation and differentiation of HPCs. Differentiation of HPCs to hepatocytes occurs at this fibrotic interface, leading to buds of intraseptal hepatocytes,11 and the RNs we see may represent further evolution of this process. In conclusion, our study has shown that RNs in cirrhosis can be monoclonal, with the cell of origin likely to be a facultative stem cell from the biliary tree.