ENGLISH ABSTRACT: BACKGROUND
Lymphomas arise in cells of the immune system. They vary widely in cytomorphology,
immunophenotype and clinical course as well as response to treatment - all of which are
factors that determine prognosis. The substantial geographical differences that exist for
Hodgkin and other lymphoproliferative disorders have been previously reported from the
Lymphoma Reclassification Project.
OBJECTIVE
This investigation was in two parts. In the first we tested the hypothesis that, using
comparable treatment regimens, outcome from a private academic centre in the Western
Cape region of South Africa would be similar to that from the first world. To this end a series
of 512 individuals were analysed. In the second tissue samples from these patients where the
most common subtype is diffuse large B-cell lymphoma (n=93) randomly receiving either
standard combination chemotherapy in the form of the CHOP regimen or the identical
program with rituximab which is an anti-CD20 monoclonal antibody were further
investigated in cooperation with the University of Nebraska Medical Centre using an
immunohistochemistry based method, also known as tissue microarray.
PATIENTS AND METHODS
In the first cohort consecutive and comprehensive records of patients diagnosed with
lymphoma aged 14 years and older seen between October 1998 in July 2006, were scrutinized.
After exclusion for a variety of reasons 398 cases suitable for further definitive study
remained.
In the second group tissue samples from a total of 149 biopsy proven de novo diffuse large Bcell
variants could be further evaluated. After additional refinement a total of 93 remained
that met all the entry criteria for the study in which 48 received chemotherapy alone and the
remaining 45 chemo-immunotherapy. Demographic features were well matched. The initial stratification of these 93 cases, based on phenotyping with immunohistochemistry
employing a tissue microarray method, yielded two populations depending on the criteria
used. Each of these primary subdivisions was further evaluated for expression of BCL2 and
LMO2 both of which are recognised to predict response. Finally, for each variable, clinical
characteristics and survival outcome were compared between chemotherapy as a single
modality or the same regimen combined with rituximab.
STATISTICAL ANALYSIS
Overall and event-free survival were calculated as the years from diagnosis to death, loss to
follow-up, first progression or relapse respectively. The Kaplan Meier method was used to
estimate distribution and the log rank test to compare survival between groups. Patient
characteristics for each cohort specified were tested with Chi-squared or Fisher's exact test for
small samples.
RESULTS
In the first part of this study all 398 cases were retrospectively analysed and showed a similar
treatment outcome with regards to overall and event-free survival at 36 months when
compared to first world reference centers. Adverse factors identified were similar to published
experience comprising constitutional symptoms, prior treatment with chemotherapy,
intermediate or high-risk scores as defined by the International Prognostic Index, histologic
grading and certain anatomical sites of primary tumour. In contrast gender, staging by Rye or
Rai classification, retroviral infection and prior treatment with radiotherapy were without
effect.
In the second part tissue samples from 93 de novo DLBCL, subtyped using an investigational
immunohistochemical based Tissue Micro Array (TMA) were contrasted to the approximately
corresponding categories as defined either by Hans and associates using a three marker panel
into germinal or non-germinal centre subtypes or by Choi and colleagues with two additional antibodies into germinal centre or activated B-cells. Not surprisingly when compared to
DNA-based gene expression profiling, as a reference point, concordance was different being
86 as opposed to 93% for the respective algorithms. The rationale for this exercise was to
determine relative cost-effectiveness in an affordable but accurate technology that may be
applicable to under resourced areas of the globe. Additionally the prognostic marker
expression for BCL2 and LMO2 was investigated with regards to treatment outcome.
Using the investigational tissue microarray approach the addition of rituximab to standard
chemotherapy group did not show any significant improvement on 5 years overall (63% vs
59%, p 0.68) or event-free survival (42% vs 39%, p 0.94). Similarly no differences were
evident in subtype analysis. Interestingly however, when segregated on the Choi criteria,
cytotoxic drugs alone showed a non-significant trend in improved survival (74% vs 55%, p
0.32) as well as event-free survival (44% vs 40%, p 0.42) for the germinal centre as opposed
to the activated B-cell subtype. Nevertheless not even a small difference could be
demonstrated in the presence of rituximab.
According to Choi, both regimens (chemotherapy with or without the addition of rituximab)
revealed similar results to the Hans algorithm on 5 years OS as well as 3 year EFS when
comparing GCB versus non-GCB subgroups.
BCL2 and LMO2 marker expression of the respective immunohistochemical (IHC) subtype,
despite small sample size, revealed the following. Analysis by Choi criteria on survival for
BCL2, no matter for which subsets (GCB or ABC) or treatment modality (chemotherapy with
or without the addition of rituximab) showed no difference in 5 years OS or EFS.
However, in contrast, a significant difference for better EFS (p=0.0015) in the BCL2 positive
group of the ABC subgroups subtypes treated with rituximab containing chemotherapy. These
results must be interpreted with care due to the very low sample size and the follow up time of
only 9 months. For LMO2 similar results on survival outcome were seen thus showing no
difference in 5 years OS or EFS – regardless of subtype or treatment modality. Also here, this was contrasted by better EFS (p=0.039) in the LMO2 positive group of ABC subtypes when
treated with the rituximab containing regimen. Again the reservation about small numbers
and the 14 months observation period apply.
DISCUSSION AND CONCLUSION
Most of the studies examining the proportions of subtypes in large series of DLBCL patients
have been predominantly carried out on Western populations. While 50% of patients in North
America or Europe express GCB phenotypes, this is only 31% in their Asian counterparts. Thus
far, no study has been published on lymphoma subtypes in South African populations using a
Tissue Micro Array (TMA) method.
Despite certain limitations of this study, due to a variety of reasons such as loss of analysable
data, variability of representation of the South African population as a whole or low sample
size leading to a potential source of error, these results confirm that lymphoproliferative
disorders are heterogeneous. Neverless this group can be treated successfully if exact staging,
classification and risk assessment defines the holistic management plan – no matter if in a
developed or under resourced setting.
With the introduction of sophisticated methods such as gene expression profiling (GEP),
diffuse large B-cell lymphoma (DLBCL) can be classified into the prognostically favourable
germinal center B-cell–like (GCB) and the more aggressive activated B-cell–like (ABC)
subtypes.
Against the background of resource restriction we therefore used an immunohistochemical
(IHC) stain method to analyse formalin-fixed, paraffin-embedded tissue samples. Here the
algorithm by Choi et al., originally described by Hans et al., showing a high concordance rate,
was comparable to the GEP classification. The use of the IHC based TMA methodology was
shown to be a simple, cost effective and a robust alternative to GEP which is currently
regarded as the gold standard for the classification in lymphomas. It provides a useful
prognostic tool in stratifying DLBCL or other entities in future, even when frozen tissue
samples are not available for GEP analysis. With the current budgetary limitations in public hospitals chemotherapy protocols for
lymphoproliferative disorders exclude agents such as rituximab. Local therapeutic drug
committees consider the approximately 15% overall survival benefit seen at 5 years for DLBCL
when rituximab is added to combination chemotherapy as too marginal for justifying the
arising additional expenses. Accordingly, demonstration that a specific molecular subtype
accounts for superior outcome when using these regimens is needed and would provide
convincing evidence for the use of this monoclonal antibody in a resource constrained setting.
AFRIKAANSE OPSOMMING: AGTERGROND
Limfome ontstaan vanuit selle van die immuunsisteem. Hulle toon ‘n groot verskil met
betrekking tot sitomorfologie, immunofenotipe, kliniese verloop, asook respons op
behandeling – almal faktore wat prognose bepaal. Die Limfoom Herklassifikasieprojek het
getoon dat daar substansiële geografiese verskille bestaan vir Hodgkin se limfoom en ander
limfoproliferatiewe toestande. Huidige navorsing oor hierdie maligniteite skep die indruk dat
hulle skaars is in Afrika, met die inligting wat wel beskikbaar is hoofsaaklik gebaseer op
gevallestudies deur klinici en patoloë.
DOEL
Hierdie studie bestaan uit twee dele. In die eerste toets ons die hipotese dat die uitkomste vir
die behandeling van limfoom in ‘n privaat akademiese instansie geleë in die Wes-Kaap, Suid-
Afrika, dieselfde is as wat in die res van die wêreld gesien word indien soortgelyke
behandeling toedgedien word.`n Reeks van 512 pasiënte is in die analise gebruik. Vir die
tweede hipotese het`n kohort van die mees algemene subtipe, naamlik diffuse groot B-sellimfoom
(DLBCL) (n = 93), lukraak òf die standaardkombinasie-chemoterapie in die vorm
van CHOP òf `n identiese program met rituximab, `n anti-CD20 monoklonale teenliggaam,
gekry. Die teenliggaam is in samewerking met die Universiteit van Nebraska se Mediese
Sentrum getoets met behulp van `n immunohistochemies-gebaseerde metode, ook bekend as
weefsel mikro-skikking (tissue microarray or TMA).
PASIËNTE EN METODE
Vir die eerste kohort is die opeenvolgende en volledige rekords van pasiënte wat ouer as 14
was en tussen Oktober 1998 en Julie 2006 in `n privaat-gebaseerde akademiese instansie
gediagnoseer is, noukeurig ondersoek. Na sekere uitsluitingskriteria toegepas is, was daar 398
gevalle wat geskik was vir verdere analise. In die tweede deel van die navorsing kon 149
pasiënte wat nuut met diffuse groot B-sel-limfoom (DLBCL) gediagnoseer is, verder
geëvalueer word. Drie-en-neëntig van hierdie 149 pasiënte het aan die kriteria vir insluiting voldoen; 48 het slegs die standaard chemoterapie (CHOP) ontvang en die oorblywende 45 het
chemo-immunoterapie (chemoterapie plus rituximab (R-CHOP) ontvang. Die demografiese
eienskappe van beide groepe het goed vergelyk.
Vanweë die kriteria wat gebruik is, het die aanvanklike stratifikasie van 93 gevalle, wat deur
middel van die TMA-metode op fenotipering met immunohistochemie gebaseer was, twee
populasies gelewer. Hierdie primêre subdivisies is verder geëvalueer vir uitdrukking van
BCL2 en LMO2, beide erkende voorspellers van respons. Laastens is kliniese eienskappe en
oorlewingsuitkoms ná behandeling met chemoterapie as `n enkel modaliteit of dieselfde
chemoterapie met rituximab, vir elke veranderlike met mekaar vergelyk.
STATISTIESE ANALISE
Algehele en insident-vrye oorlewing is bereken as die jare vanaf diagnose tot sterfte,
verdwyning van pasiënte tydens opvolg, eerste progressie of terugval van die toestand. Die
Kaplan Meier-metode is gebruik om distribusie te bepaal en die log rank-toets om oorlewing
tussen die groepe te vergelyk. Pasiëntkenmerke vir elke gespesifiseerde kohort is met die Chikwadraattoets
of Fisher se eksakte toets in die geval van kleiner groepe getoets.
RESULTATE
In die eerste gedeelte van die studie is al 398 gevalle terugwerkend geanaliseer en daar is
gevind dat die behandelingsuitkoms met betrekking tot totale oorlewing en insident-vrye
oorlewing teen 36 maande vergelykbaar was met uitkomste in eerstewêreldsentrums.
Nadelige faktore met betrekking tot oorlewing was soortgelyk aan reeds gepubliseerde data en
het die volgende ingesluit: gestelsimptome, voorafgaande behandeling met chemoterapie,
intermediêre of hoë-risiko tellings soos deur die Internasionale Prognostiese Indeks bepaal,
histologiese gradering en sekere anatomiese setels van primêre tumor. In teenstelling met
internasionale ondervinding het geslag, steiering volgens Rye- of Rai-klassifikasie, retrovirale
status en vorige behandeling met radioterapie geen invloed gehad nie. In die tweede gedeelte is weefsel van 93 nuut-gediagnoseerde DLBCL-pasiënte wat deur
middel van die TMA-metode subtipeer is, vergelyk met naastenby ooreenstemmende
kategorieë soos deur Hans et al. met `n drie-merkerpaneel in kiem- en nie-kiemsentrumsubtipes,
of deur Choi et al. met twee bykomende teenliggame in kiemsentrum of
geaktiveerde B-selle gedefinieer. Met die DNA-gebaseerde geen-uitdrukkingsprofiel as `n
verwysingspunt, was die ooreenstemming tussen die onderskeie algoritmes na verwagting
verskillend, met 86% teenoor 93%. Die onderliggende rasionaal was om die relatiewe
lonendheid te bepaal van `n bekostigbare maar akkurate tegnologie, wat moontlik in gebiede
met onvoldoende hulpbronne toegepas kon word Verder is die prognostiese merkeruitdrukking
vir BCL2 en LMO2 ook met die oog op die uitkomste van behandeling ondersoek.
In vergelyking met standaard chemoterapie, het die gebruik van die TMA-tegniek met
toevoeging van rituximab geen noemenswaardige verbetering in die algehele 5-jaar oorlewing
(63% vs 59%, p = 0.68) óf insident-vrye oorlewing (42% vs 39%, p = 0.94) getoon nie.
Insgelyks was daar geen duidelike verskille in subtipe-analise nie. Dit is egter interessant dat,
met die toepassing van die Choi-kriteria, sitotoksiese middels op hul eie, in teenstelling met
die geaktiveerde B-sel-subtipe, nie ‘n statisties belangrike neiging tot beter oorlewing (74% vs
55%, p = 0.32) of insident-vrye oorlewing (44% vs 40%, p = 0.42) vir die kiemsentrumsubtipe
(GCB) getoon het nie. Dit was selfs nie eers moontlik om ‘n klein verskil in die
teenwoordigheid van rituximab te demonstreer nie.
Volgens Choi, het chemoterapie (met óf sonder die toevoeging van rituximab) by vergelyking
van kiemsentrumsubtipes met nie-kiemsentrumsubtipes soortgelyke resultate gelewer as die
Hans algoritme na 5-jaar oorlewing, sowel as 3-jaar insident-vrye oorlewing. Ten spyte van ’n klein steekproef het verdere subtipe-analise van BCL2- en LMO2- merkeruitdrukking van die onderskeie immunohistochemiese (IHC) subtipes die volgende
aan die lig gebring: Analise met gebruik van die Choi-kriteria vir subtipe-analise vir BCL2-
uitdrukking het, ongeag die subtipe (GCB of ABC) en die behandelingsmodaliteit
(chemoterapie met of sonder toevoeging van rituximab), geen verskil getoon in 5-jaar
oorlewing en insident-vrye oorlewing nie. Daar was egter `n noemenswaardige verskil ten
opsigte van beter insident-vrye oorlewing (p = 0.0015) in die BCL2-positiewe groep van die
ABC-subtipes wat met rituximab-bevattende chemoterapie behandel is. Hierdie resultate
moet egter met sorg geïnterpreteer word weens die klein steekproef en kort opvolgperiode van
slegs nege maande.
Soortgelyke resultate met betrekking tot die uitkoms vir oorlewing is vir LMO2 gelewer,
naamlik geen verskil ten opsigte van 5-jaar oorlewing of insident-vrye oorlewing ongeag die
subtipe of behandelingsmodaliteit. Hier ook was daar egter beter insident-vrye oorlewing (p
= 0.039) in die LMO2-positiewe groep van die ABC-subtipe wanneer rituximab-bevattende
chemoterapie toegedien is. Weereens moet die resultate met sorg interpreteer word weens
die klein steekproef en die kort opvolgperiode van 14 maande.
BESPREKING EN GEVOLGTREKKING
Die meeste studies wat reeds die verhoudings van subtipes in groot getalle DLBCL-pasiënte
ondersoek het, is onder westerse populasies uitgevoer en daar bestaan onsekerheid oor of
dieselfde verhoudings in Afrika van toepassing is. Terwyl 50% van alle DLBCL-pasiënte in
Noord-Amerika of Europa die GCB-fenotipe uitdruk, kom dit in slegs 31% van hierdie
pasiënte in Asië tot uitdrukking. Geen studie is tot dusver met behulp van die TMA-metode
onder limfoomsubtipes in populasies in Afrika onderneem nie, veral nie onder pasiënte met
verswakte immuunstelsels nie.
Ten spyte van sekere beperkings van hierdie studie, waarvoor daar verskeie redes is, soos die
verlies van analiseerbare data, die wisselende verteenwoordiging van die Suid-Afrikaanse bevolking as ‘n geheel en die klein steekproef, wat vergissing in die hand kan werk, bevestig
die resultate dat limfoproliferatiewe toestande heterogeen is. Desnieteenstaande kan hierdie
groep met sukses behandel word indien ‘n holistiese bestuurplan presies volgens stadiums,
klassifikasie en risiko-assessering uitgevoer word – ongeag of dit in `n ontwikkelde gebied of
een met min hulpbronne plaasvind.
Met die beskikbaarheid van ingewikkelde metodes soos geen-ekspressie profielskepping
(GEP), kan DLBCL in prognosties-voordelige kiemsentrum B-sel-agtige (GCB) en die meer
aggressiewe geaktiveerde B-sel-subtipes geklassifiseer word. Teen die agtergrond van
beperkte hulpbronne, is immunohistochemiese (IHC) kleuringsmetodes gebruik om -
weefselmonsters wat in formalien gefikseer en in paraffien ingebed was, te analiseer. Hier was
die algoritme wat oorspronklik deur Hans et al. beskryf is en deur Choi et al. verder ontwikkel
is, en hoë ooreenstemming toon, vergelykbaar met die GEP-klassifikasie.
Die gebruik van die IHC-gebaseerde metodologie is as ‘n eenvoudige, effektief lonende en
kragtige alternatief tot GEP, wat tans as die goudstandaard vir klassifikasie van limfoom
beskou word, bewys. Dit verskaf ‘n nuttige prognose-hulpmiddel vir die stratifisering van
DLBCL of ander entiteite in die toekoms, selfs wanneer gevriesde weefselmonsters nie vir
GEP-analise beskikbaar is nie.
Onder die huidige begrotingsbeperkings in staatshospitale is middels soos rituximab by die
protokolle vir die behandeling van limfoproliferatiewe toestande uitgesluit. Plaaslike
terapeutiese middel-komitees beskou die nagenoeg 15% algehele oorlewingsvoordeel teen vyf
jaar, wat vir DLBCL moontlik is met die toevoeging van rituximab by kombinasiechemoterapie,
as te randstandig om die bykomende onkoste te regverdig. Daarvolgens is dit
noodsaaklik om te demonstreer dat `n spesifieke molekulêre subtipe tot ’n beter uitkoms lei
wanneer hierdie metode gebruik word en dat dit oortuigende bewyse sal lewer vir die gebruik
van hierdie monoklonale teenliggaam in `n omgewing met beperkte hulpbronne.
