Molecular pathogenesis of human hepatocellular carcinoma Snorri S.

Thorgeirsson1 &
Joe W. Grisham1,2
Hepatocarcinogenesis is a slow process during which genomic changes progressiel!
alter the hepatocellular phenot!pe to produce cellular intermediates that eole into
hepatocellular carcinoma. "uring the long pre# neoplastic stage, in which the lier is
often the site of chronic hepatitis, cirrhosis, or $oth, hepatoc!te c!cling is accelerated $!
upregulation of mitogenic pathwa!s, in part through epigenetic mechanisms. This leads
to the production of monoclonal populations of a$errant and d!splastic hepatoc!tes that
hae telomere ero# sion and telomerase re#e%pression, sometimes microsatellite
insta$ilit!, and occasionall! structural a$erra# tions in genes and chromosomes.
"eelopment of d!splastic hepatoc!tes in foci and nodules and emergence of
hepatocellular carcinoma are associated with the accumulation of irreersi$le structural
alterations in genes and chromosomes, $ut the genomic $asis of the malignant phenot!pe
is heterogeneous. The malig# nant hepatoc!te phenot!pe ma! $e produced $! the
disruption of a num$er of genes that function in dif# ferent regulator! pathwa!s,
producing seeral molecular ariants of hepatocellular carcinoma. &ew strategies should
ena$le these ariants to $e characteri'ed. 1(a$orator! of )%perimental *arcinogenesis,
*enter for *ancer +esearch, &ational *ancer ,nstitute, &ational ,nstitutes of Health,
-ethesda, Mar!land 2./02, 1S2. 2"epartment of 3atholog! and (a$orator! Medicine,
1niersit! of &orth *arolina at *hapel Hill, *hapel Hill, &orth *arolina 24500, 1S2.
*orrespondence should $e addressed to S.S.T. 6e#mail7 snorri8thorgeirsson9nih.go:.
Hepatocellular carcinoma 6H**: is one of the most fre;uent is# ceral neoplasms
worldwide, with an estimated 5<=,... new cases and almost as man! deaths1,2 in 2....
Much is >nown a$out the deelopment and causes of H**. ,t nearl! alwa!s deelops in
the setting of chronic hepatitis or cirrhosis, conditions in which man! hepatoc!tes are
>illed, inflammator! cells inade the lier and connectie tissue is deposited. These
changes drasticall! alter the matri% and microeniron# ment of the lier?@5. The initial
hepatocellular alterations that pre# cede the appearance of H** include foci of
phenot!picall! altered hepatoc!tes and, su$se;uentl!, d!splastic hepatoc!tes that form
foci and nodules?@5. Some agents that trigger H**s, and account for much of the
mar>ed ariation in its incidence, hae $een identi# fied and their impacts ;uantified=.
The arious causes of H** are perhaps $etter understood than those of an! other maAor
cancer in humans. Burthermore, the main causatie agentsChepatitis - irus 6H-D:,
hepatitis * irus 6H*D: and aflato%in -1 62B-:C which together are responsi$le for
a$out /.E of all H**s in humans=, leae Fmolecular mar>sG on hepatoc!tes that ena$le
the causes of indiidual H**s to $e determined accuratel! in man! instances?,<.
Genomic a$errations hae $een descri$ed in thou# sands of H**s from humans, !ielding
a rich 6although still incom# plete: data$ase?,5. The molecular interactions $etween
hepatoc!tes and the specific etiologic agents of H**, and how these interac# tions disrupt
hepatocellular genes and gene products leading to the deelopment of H**, are $eing
elucidated?,<@/. &eertheless, the molecular pathogenesis of H**Cthat is, the specific
genomic alterations that drie its deelopmentCis not understood. 6The term genomic
alteration is used here to include $oth ;uantitatie changes in gene e%pression occurring
in the a$sence of structural a$normalit!, and ;ualitatie changes in gene e%pression
resulting from a$errations in gene structure.: Here we summari'e data from pu$lished
studies and correlate the results with specific categories of preneoplastic lesions and with
H**. We hae pooled data from man! pu$lished studies, cited in a recent comprehensie
literature reiew? and in more recent pu$lications, in which the categories of the
preneoplastic conditions 6for e%ample, chronic hepatitis and cirrhosis: and the categories
of hepatocellular lesions 6for e%ample, cirrhotic nod# ules or phenot!picall! altered,
adenomatous, d!splastic andHor malignant hepatoc!tes: hae $een descri$ed. We hae
also pooled data from H**s categori'ed $! histological grade according to the
uniersall! applied method of )dmondson and Steiner?. We hae su$Aected the pooled
data to statistical anal!sis $! t!pe of preneoplastic condition or hepatocellular lesion. We
hae included all studies that met these minimal criteria, in order to ealuate the
heterogeneit! of the results of numerous studies from different geographic regions. Iur
anal!sis outlines a general deelopmental se;uence of genomic changes during the
process of hepatocarcinogenesis and indicates molecular pathwa!s that ma! drie
hepatoc!tes to deelop a malignant phenot!pe. 2ccelerated proliferation of hepatoc!tes
and deelopment of monoclonal hepatoc!te pop# ulations occur in all preneoplastic
conditions and altered cell populations, and these changes continue in H**s. Genomic
alterations appear to deelop randoml!, $eginning in preneo# plastic lesions, and their
deelopment escalates in d!splastic hepatoc!tes and H**s. The e%tensie heterogeneit!
of genomic lesions displa!ed $! H**s suggests that H** ma! $e produced $! selection
of $oth genomic and epigenetic alter# ations that compromise more than one regulator!
pathwa!. 2dditional data are needed to full! understand the molecular pathogenesis of
H**, howeer, and we suggest specific studies and new strategies.
Se;uential morphological changes in the lier leading to H** Hepatocarcinogenesis in
humans unfolds during a process that ma! ta>e more than ?. !ears after chronic infection
with H-D or H*D is first diagnosed?,0 6Big. 1:. *irrhosis and H** occur rou# tinel! in a
fraction of patients who deelop chronic infection with H-D or H*D. H**s arise with
increasing fre;uenc! in li# ers that are the site of chronic hepatitis1.,11 and cirrhosis12@
1<, particularl! from d!splastic hepatoc!tes?,14,1/. The tissue lesions that commonl!
precede H** 6chronic hepatitis and cirrhosis containing foci of phenot!picall! altered
and d!splastic hepato# c!tes: proide a framewor> for identif!ing the temporal order with
which genomic alterations deelop during hepatocarcino# genesis 6Big. 1:, the first step in
defining the molecular pathogen# esis of this neoplasm. Genomic changes during the
preneoplastic plastic phase "uring most of the long preneoplastic stage leading to H**,
alterations in gene e%pression are almost entirel! ;uantitatie, occurring $! epigenetic
mechanisms in the a$sence of detected changes in the structures of genes or
chromosomes. )leated e%pression of transforming growth factor#J 6TGB#J:?,10 and
insulin#li>e growth factor#2 6,GB#2:? is responsi$le for acceler# ated hepatoc!te
proliferation. 2ltered e%pression of TGB#J and ,GB#2 reflects the e%pression of seeral
genes that are strongl! upregulated $! epigenetic mechanisms during the preneoplastic
stage 6Big. 2:. 1pregulation of TGB#J and ,GB#2 results from the com$ined actions of
c!to>ines produced $! chronic inflammator! cells that infil# trate the carcinogen#
damaged lier?, iral transactiation?,<@/ and the regeneratie response of the lier to
cell loss?. "!sreg# ulated e%pression of ,GB#2 is associated with altered meth!# lation
and re#imprinting of the ,GB#2 gene?,2., inactiation of the 31 promoter, actiation of
the 3? promoter and produc# tion of large amounts of this growth factor 6ref. ?:. Hepato#
c!tes proliferate repeat# edl!?,21,22 and monoclonal hepatoc!te populations
deelop?,2?@2< 6Big. 2:, along with progressie telomere shortening? 6data not shown:
and re#e%pression of the telomerase en'!me com# ple%?,24@20 6Big. ?:. 2$errant
meth!lation 6h!po# or h!permeth!lation: also alters *pG groups of other genes and
chromosomal segments, $egin# ning in liers that are the sites of chronic hepatitis and
cirrho# sis?,?.@=1 6Big. ?:. )%pression of "&2 meth!ltransferases 6"&MTs:, which
catal!'e the meth!lation and demeth!lation of *pG groups, is increased in a fraction of
liers affected with chronic hepatitis and cirrhosis?,=.,=1. -oth "&MT1 and "&MT?a
are strongl! upregulated in H**s=.,=1. S#adenos!lme# thionine s!nthase and gl!cine &#
meth!ltransferase?, which aug# ment the hepatocellular pool of meth!l groups aaila$le
for meth!lation reactions, also are upregulated in man! H**s. Ither possi$le causes of
epigenetic changes in gene e%pression include changes in chromatin acet!lation, which
has apparentl! not $een e%amined in liers during H** deelopment, and cis# and
transactiation of genes resulting from the actions of iral transactiating molecules,
integrated iral promoters, trans# posed cellular promoters, or some com$ination
thereof?,<@/. Microsatellite insta$ilit! occurs in hepatoc!tes in some chronic hepatitis,
cirrhosis and H**?,?<@=/, $ut a$errations in other t!pes of "&2 repair in H**s hae
not $een reported. Structural alterations 6such as amplifications, mutations, dele# tions
and transpositions: deelop slowl! in a few genes and chro# mosomal loci during the
earl! preneoplastic phase, $ut increase mar>edl! in d!splastic hepatoc!tes and H**s.
Simultaneous e%amination of multiple loci detects allelic deletions in ?.@5.E of liers
with chronic hepatitis or cirrhosis?<,=?,==,=4, in 4.@/.E of d!splastic Big. 2 *hanges in
e%pression of TGB#J and ,GB#2, monoclonalit! and cell c!cling in hepatoc!tes from
liers that were the site of chronic hepatitis 6!ellow: and cirrhosis 6green:, in d!splastic
hepatoc!tes 6$lue: and in H** 6red:. "ata in this and other figures and ta$les are
presented as means K s.d. The data for TGB#J and ,GB#2 indicate the fraction of lesions
that were histochemicall! positie 6L: for each growth factor. ,n $oth instances, the
histochemical findings were sup# ported $! the relatie e%pression of growth#factor
m+&2, ;uantifica# tion of growth#factor protein or $oth. 2steris>s in the $ars for TGB#J
and ,GB#2 for d!splasias indicate insufficient data to ;uantif!, although eidence
suggests that $oth growth factors are highl! oer# e%pressed in d!splastic lesions.
Monoclonalit! was determined from the pattern of integration of H-D "&2 into
hepatoc!te "&2, the ran# dom silencing of a pol!morphic M#lin>ed gene in females,
"&2 finger# print patterns o$tained $! 3*+ amplification of multilocus minisatellite
pro$es or random primers, or com$inations of these fac# tors. Hepatoc!tes containing
proliferating cell nuclear antigen 63*&2: were identified $! immunohistochemistr!.
progress ?=. nature genetics N olume ?1 N august 2..2 -I- *+,M, preneoplasia 61. to
?. !ears: d!splasia 6? to 5 !ears: neoplasia 6O5 !ears: H-D H*D 2B-1 chronic hepatitis
cirrhosis phenot!picall! altered hepatoc!tes d!splastic hepatoc!tes hepatocellular
carcinoma Big. 1 *hronologic se;uence of cellular lesions culminating in the
deelopment of hepatocellular carcinoma in human. . 1. 2. ?. =. 5. <. 4. /. 0. 1..
11. TGB#J 6E L lesions: ,GB#,, 6E L lesions: monoclonalit! 6E L lesions: cell c!cling
63*&2 inde%: 5 1. 15 2. 25 ?. ?5 =. =5 5. 55 3* &2 L ce lls 6E : le si on s af fe ct ed
6E : chronic hepatitis cirrhosis d!splasia H** nodules==,=0 and in almost all
H**s?,5,?<,=?,==. 2mong multiple loci studied simultaneousl!, howeer, the pro#
portion of informatie alleles that are deleted in chronic hepatitis and cirrhosis is much
lower than in either d!s# plastic hepatoc!tes or H**s?,5,?<,=?,==,=4 6Big. ?:. Thus, the
multiplicit! of allelic deletions in affected cell popu# lations is low in chronic hepatitis
$ut rises sharpl! in d!splastic hepatoc!tes, and is highest in H**s. -oth microsatellite
insta$ilit! at identical loci?,5 and identical allelic deletions or gene
mutations?,5,?<,==,=0,5. hae $een descri$ed in cirrhotic and d!splastic nodules and
adAacent H**s, indicating that H**s often arise as clonal outgrowths of cirrhotic
6d!splastic: nodules. Seeral of the structurall! a$er# rant loci detected in preneoplastic
cell populations differ from those found in adAacent H**s?,?<,=4, howeer, suggesting
that man! of the cells har$oring the earl! genomic a$errations do not eole into the
malignant phenot!pe. Genomic changes in H** The earl! epigenetic changes and some
of the earl! structural alterations of genes or loci are not sufficient to induce malignant
phenot!pes in hepatoc!tes, as H**s ultimatel! deelop in onl! a minor fraction of the
liers with chronic hepatitis or cirrhosis. )pigenetic changes in gene e%pression that
continue in H**s appear to act indirectl! $! creating conditions that increase the chance
of generating hepatoc!te populations containing critical com$inations of structurall! and
functionall! a$errant genes. H**s hae recurrent allelic deletions?,5,?<,=?,5.,51 6Big. =:
and regional losses and gains?,5,52@<. 6Big. 5: on seeral chromosomes.
,ndiidual H**s t!picall! contain multiple allelic dele# tions and chromosomal losses
and gains concurrentl!P een small, well#differentiated H**s often contain simultaneous
structural alterations in man! genes and chromosomes<.. The potential causes of
structural a$errations in genes in H**s are numerous and aried. ,n at least some
instances epige# netic changes directl! precede structural alterations in the same genes.
Bor e%ample, oere%pression of the gene c#m!c in H**s is initiall! correlated with
promoter h!pometh!lation and later with gene amplification?. +educed e%pression of
*"Q&22 6also >nown as p1<,&Q=2: is associated predominantl! with promoter
h!permeth!lation?,5,?<,=?,5.,51, $ut loss of het# ero'!gosit! 6(IH:, $iallelic loss and
mutation also are responsi$le for some loss of e%pression of this gene?,5,?5,5.,<1.
H!permeth!lation ma! precede locus deletion, as e%emplified $! loci located on
chromosome 1<;?,?=. 3aracentromeric chromosomal translocations, which are fre;uent
in H**s<2, are also preceded $! e%tensie h!permeth!lation of these chromosomal
regions?<. 2dditionall!, $oth H-D and 2B- produce structural changes in the genomic
"&2?,<@/, which ma! induce mutations, $rea>s and rearrangements. Molecular prod#
ucts of H-D and H*D ma! impair the function of en'!mes inoled in "&2 repair?,<@
/. Microsatellite insta$ilit! reflects impaired "&2 mismatch repair and leads to
frameshift muta# tions and deletions<?, and ma! presage the e%pression of a Fmuta# tor
phenot!peG $! affected cells<=. *omplete erosion of telomeres, which ma! e%ist in highl!
replicated preneoplastic and neoplastic hepatoc!tes, results in Fstic>!G chromosome ends
that are analogous to dou$le#strand $rea>s and produces mitotic non#disAunction and
chromosome disruption<5. I%idatie and nitrosatie damage to the genomic "&2 of
hepatoc!tes occurs in the chronicall! inflamed lier?,<< and can lead to mutations in
genes if not repaired<4. I%idatie damage also mutates mito# chondrial "&2 during
chronic hepatitis, cirrhosis and H**</. Se;uential deelopment of genomic structural
a$errations 2s indicated $! their relatie a$undance in preneoplastic lesions and H**s,
structural genomic changes 6eidenced $! allelic deletions and segmental gains and
losses: deelop slowl! during chronic hepatitis and cirrhosis 6Big. <:. 2lthough a few
allelic deletions hae $een detected in chronic hepatitis and cirrho# sis?<,=?,==,51,
chromosome segmental gains and losses hae not?,5,54, perhaps reflecting the relatie
insensitiit! of compara# tie genomic h!$ridi'ation 6*GH: for detecting small
segmental chromosomal lesions. Bormation of $oth allelic deletions==,=0 and segmental
gains or losses?,54 appears to escalate sharpl! in d!splastic hepatoc!tes, and multiple
alleles or loci are affected simultaneousl!. &either the fraction of informatie alleles that
are deleted nor the num$er of segmental gains and losses differ Big. = 2utosome arms
that contain allelic deletions in more than ?.E of reported H**. (IH was detected $!
methods emplo!ing restric# tion fragment length or microsatellite pol!morphisms. Ither
auto# some arms contain allelic deletions in less than 2.E of H**. Big. ? *hanges in
telomerase e%pression, a$errant gene meth!lation, microsatellite insta$ilit! and fraction
of alleles deleted in hepato# c!tes from liers the site of chronic hepatitis 6!ellow: and
cirrhosis 6green:, d!splastic hepatoc!tes 6$lue: and H** 6red:. Telomerase was detected
$! conentional T+23 assa!s 6usuall! with internal stan# dard: or $! semi;uantitatie
assa!s. These assa!s were supported $! +T@3*+ of telomerase or hT)+T m+&2.
Microsatellite insta$ilit! indi# cates the fraction of lesions that showed at least one
unsta$le locus. "eleted alleles refers to the fraction of lesions that showed the loss of one
allele at one or more loci. progress nature genetics N olume ?1 N august 2..2 ?=1 chronic
hepatitis cirrhosis d!splasia H** . 1. 2. ?. =. 5. <. 4. /. 0. telomerase a$errant
meth!lation microsatellite insta$ilit! alleles deleted po si ti e le si on s or lo ci 6E : 5 15
1. 2. 25 ?. ?5 =. =5 in fo rm at i e al le le s de le te d 6E : . 1. 2. ?. =. 5. <. 4. 1p
1; =; 5; <; /p 0p 1?; 1<p 1<; 14p chromosome arm H ** 6E : w ith (I H
significantl! $etween d!splastic lesions and H**s 6Big. <:. &ew structural a$errations
deelop at some loci as H**s progress from small, well#differentiated tumors to large,
poorl! differenti# ated tumors?,5,?<,51,5=,55 6Big. 4:. 3roduction of new structural
a$errations in a clone of malignant cells leads to clonal dier# gence and neoclonal
eolution of cells that now contain a more dierse mi%ture of genomic a$errations.
Heterogeneit! of genomic changes in H** 2lthough nearl! all H**s contain multiple
genes and loci with structural alterations, alterations in indiidual genes 6Ta$le
1:?,5,?=,==,<.,</@4?, allelic deletions, and losses or gains of specific chromosomal loci
rarel! affect more than half of H**s anal!'ed 6Bigs = and 5:. This genomic
heterogeneit! ma! result in part from the diergent characteristics of the H**s that hae
$een studied, which include tumors that are small, well# differentiated and locali'ed as
well as those that are large, poorl! differentiated and metastatic. -ut een when the data
are segregated according to indicators of H** progression, including histological grade
6Big. 4: and si'e or clinical stage 6data not shown:, there remain high coefficients of
ariation among results from different studies. Heterogeneit! of genomic a$errations ma!
also reflect the actions of different causatie agents. 2 nota$le e%ample of an a$erration
in gene structure related to a specific cause of H**s is the high fre;uenc! of mutations
affecting Ser 2=0 of p5? in the tumors of patients chronicall! e%posed to 2B-?,/,4=. 2
few stud# ies5=,5< 6$ut not all?,5,55: hae also suggested that losses and gains of
chromosome regions ma! differ in H**s caused $! H-D as compared to H*D infection.
Heterogeneit! at the molecular leel ma! also result from the modification of gene
function $! different mechanisms, including altered meth!lation, allelic deletion and
mutation, among others. *urrent eidence suggests that the regulator! pathwa!s
controlling proliferatie c!cling, replicatie lifespan 6Fimmortalit!G:, death, dif#
ferentiation and genomic integrit! ma! each $e a$normal in some H**s. Together these
regula# tor! pathwa!s contain thousands of separate molecules whose a$errant e%pression
or structure ma! su$ert precise control of cellular phenot!pe. The multiple regulator!
circuits that control cellular phenot!pe form comple%, redundant networ>s in which
indiidual pathwa!s interact with one another. Some molecules, such as p5?, occup!
critical positions at FnodesG interconnecting seeral regulator! pathwa!s. 2n a$erration in
a single gene whose protein product occupies such a node can impair the con# trol of
more than one cellular process and disrupt seeral regulator! pathwa!s. The mar>ed
ariation in e%pression of specific genetic a$errations in H**s documented in numerous
independent studies, coupled with the limited FpenetrationG of an! single genomic
a$erration among all H**s e%amined 6Ta$le 1P Bigs = and 5:, suggests an alternate
possi$ilit! to e%plain molecular heterogene# it!. ,t is possi$le that a num$er of different
com$inations of a$er# rant genes and regulator! pathwa!s can generate the neoplastic
phenot!pe in hepatoc!tes, and that this results in seeral ariants or su$sets of tumor
haing different growth properties. This has important implications for accurate
molecular diagnosis and prognosis. The design of targeted molecular therapies ma! need
to $e tailored to the specific molecular phenot!pe of the particu# lar H**. Studies
concurrentl! e%amining multiple genes in the same H**s will $e re;uired to ealuate
this possi$ilit! thor# oughl!. -ut data alread! aaila$le support the notion that a$er#
rations in different genes and molecular pathwa!s are responsi$le for the clinical
properties of different H**s. Bor e%ample, H**s with mutant R#catenin are clinicall!
less aggressie than those without this mutation $ut har$oring multiple chromosomal
a$errations?,5,42,45. This difference is pro$a$l! lin>ed to a particular mechanism of
H** deelop# ment. 2dditional studies are needed to identif! other molecular ariants of
H** and to correlate molecular status with clinical course. *om$inations of genomic
a$errations detected in different clonal preneoplastic lesions and in separate H**s
located in the same lier differ from one another. This reflects $oth the random creation
of indiidual genomic a$errations and the clonal popu# lations that derie from them.
Heterogeneit! of a$er# rant genomic com$inations in malignant clones supports the idea
that dierse com$inations of genomic a$errations can sufficientl! d!sregulate Big. <
Se;uential deelopment of structural genomic a$erra# tions, as reflected $! allelic
deletions 6red: and chromosome regional losses and gains 6green: in hepatoc!tes in liers
the site of chronic hepatitis and cirrhosis, d!splastic hepatoc!tes and H**. Big. 5
2utosome arms that show regional losses 6red: and gains 6green: in more than 2.E of
reported H**, as detected $! comparatie genomic h!$ridi'ation. ,n gen# eral, regional
losses $! *GH correspond well with (IH 6Big. =:. progress ?=2 nature genetics N
olume ?1 N august 2..2 . 1. 2. ?. =. 5. <. 4. 1p =; 5; <; /p 0p 1?; 1<p 1<; 14p 1p
<p /; 14; chromosome arm H ** 6E : w ith lo ss es or ga ins . 5 1. 15 2. 25 ?. ?5 =.
=5 5. 55 <. chronic hepatitis cirrhosis d!splasia H** ? < 0 ch ro m os om e a$ er ra tio
ns 6* G H: 12 &S allelic deletions *GH in fo rm at i e al le le s de le te d 6E :
hepatoc!tes that the! $ecome malignant. Burther# more, the concurrent presence of
genomicall! dis# tinct malignant clones?,4< suggests that the entire lier undergoes
neoplastic deelopment as a FfieldG44 with multiple hepatoc!tes progressing to
malignanc! simultaneousl! $ut $! different Fgenomic routesG and at different rates.
1nidentified genes ma! $e inoled in H** +ecurring allelic losses or gains hae $een
detected on 1= chromosome arms in more than ?.E of all H**s in which allelic
deletions?,5,?<,=?,=0,51 6Big. =: and loss or gain of chromosomal regions?,5,52@<. 6Big.
5: hae $een anal!'ed. ,n theor!, the deleted chromosome regions ma! contain tumor#
suppressor genes. ,ndeed, some fre;uentl! deleted chromo# some regions contain genes
that fre;uentl! undergo allelic or chromosome regional loss in H** 6Bigs. = and 5:, such
as the tumor#suppressor genes p5? and +$, which are often inactiated $! deletion or
mutation 6Ta$le 2:. The gene *"Q&22, which is inactiated $! mechanisms including
meth!lation, deletion and mutation, is also located in a fre;uentl! deleted chromosomal
region?,5,?<,5.,<1 6Ta$le 1:. (i>ewise, some of the areas of chromo# some regional gain
6Big. 5: contain genes encoding such mole# cules as c#m!c, which is oere%pressed in
most H**s? and is amplified in ?.E 6refs ?,5P Ta$le 1:. &ew genes that ma! $e inoled
in the deelopment of H** hae $een isolated from a fre;uentl! deleted area of
chromosome /p?,4/, $ut specific H**# related genes hae not $een identified in most
chromosomal regions that fre;uentl! show genomic loss or gain in H**. Whether the
anon!mous genes located in these chromosome regions are causall! related to H**
deelopment, or are merel! F$!standersG whose alteration merel! reflects widespread
chromo# some insta$ilit! in H**s, is not clear. That the most fre;uent losses and gains in
H**s inole onl! a limited num$er of chromosome arms 6Bigs. = and 5: suggests that
each affected region contains one or more genes whose loss or altered function or
e%pression con# tri$utes to the deelopment and progression of H**. ,t further implies
that either the deelopment or persistence of these changes is not random. When the
a$ilit! of a hepatoc!te to con# trol genomic integrit! is impaired, genetic structural
alterations are li>el! to occur stochasticall! in the face of sustained high rates of cell
proliferation, coupled with diminished capacit! to repair new a$errations. The notion that
genetic changes are randoml! generated during hepatocarcinogenesis is supported $! the
high fre;uenc! of a$errations in house>eeping genes?, man! of which are not directl!
inoled in the deelopment of a malignant phe# not!pe, and $! the progressiel!
increasing num$er and ariet! of loci affected as hepatocarcinogenesis continues. 2lso,
of se# eral chromosome arms that contain fre;uentl! deleted regions, in a significant
fraction of affected H**s, the entire chromo# some arm is lost?,5. Bre;uent loss of entire
chromosome arms suggests the action of a process without specificit! for particular
genes. H!permeth!lation of paracentromeric chromosome regions?< ma! $e the
mechanism causing the nonspecific loss of chromosome arms. The most fre;uent
com$inations of gene a$errations detected in H**s ma! result from the darwinian
selection of alleles proiding surial and growth adantages from a randoml! generated
mi%ture of altered genes. To sole this conundrum and to full! understand the molecular
pathogenesis of H**, it will $e necessar! to identif! the genes contained in the affected
chromosome arms and ascertain the specificit! of their relationship to the deelopment of
H**. Bocusing future studies Most preious genomic studies hae inestigated full!
deeloped H**sP much less is >nown a$out the genetic $ases of preneoplas# tic lesions
6foci and nodules of phenot!picall! altered and d!s# plastic hepatoc!tes: and
hepatocellular adenomas 6H*2s:, the $enign neoplastic counterpart of H**. The
molecular patho# genesis of H** cannot $e understood without more detailed stud! of
the genomic alterations characteri'ing its earl! deelop# ment and the genomic
differences $etween $enign and malig# nant hepatocellular tumors. The pro%imate
cellular precursors of H**Cfoci of phenot!picall! altered hepatoc!tes and d!splastic
hepatoc!tesCmerit especiall! intense stud!. (aser#capture microscop! 6(*M: ena$les
these indiidual lesions to $e selec# tiel! recoered as cellular clones for genomic
anal!sis40. +ecent studies of preneoplastic lesions recoered $! (*M and similar
techni;ues hae !ielded proocatie results, suggesting that a$er# rations of particular
genes and chromosomes ma! presage the emergence of H**s==,=4,=0,54, $ut additional
studies are needed. Big. 4 2ccumulation of selected structural genomic a$errations in
H** with progressie deterioration of tumor cell dif# ferentiation. Well differentiated
6green:, moderatel! differentiated 6$lue: and poorl! differentiated 6red: H**