(CANCER RESEARCH (SUPPL.) 52, 54S6s-5464s. October 1, 1992]

Pathological Classification of Non-Hodgkin's Lymphoma for EpidemiologicalStudies1

Dennis D. Weisenburger2

Department of Pathology and Microbiology, University of Nebraska Medical Center, Omaha, Nebraska 68198-3135

Abstract

Non-Hodgkin's lymphoma (NHL) consists of a heterogeneous group

of disorders which have been difficult to study by epidemiológica!meansin the past. However, recent advances in knowledge of the biologyof NHL and improvements in its classification will greatly improvethe quality of epidemiológica!studies in the future. Use of the WorkingFormulation and the current International Classification of Diseasesfor Oncology, along with paraffin immunohistochemistry, allow thedelineation of NHL subgroups with possible etiological significancebased on the biology of the disease. The collaboration of epidemiologistswith expert pathologists in the design, performance, and evaluationof epidemiológica! studies of NHL is essential for such studies to bemeaningful.

Introduction

NHL3 has long been recognized as a heterogeneous group of

disorders based on morphological appearance, clinical presentation, and response to therapy. In recent years, the use ofimmunological and molecular biological techniques has led toimportant advances in our knowledge of lymphocyte differentiation and has provided the basis for a better understanding ofthe cellular origin and pathogenesis of NHL. Currently, thevarious types of NHL are thought to represent neoplastic cellsarrested at various stages in the normal differentiation scheme,although the key events in malignant transformation may actually occur in cells at an earlier stage of differentiation (1, 2).

A number of difficulties have been associated with the performance and review of epidemiological studies of NHL andrelated hematopoietic cancers (3). Foremost has been the constantly evolving nomenclature and confusing array of complexclassification systems for NHL. These systems have evolvedand improved the classification of NHL over the years as newscientific information has become available and been incorporated. However, epidemiologists and nosologists have traditionally relied on the ICD developed by the WHO to classify NHL.Unfortunately, the ICD has not kept up with the rapid changesin the state of the art of NHL classification over the years, andthe ICD system has not been particularly useful for epidemiological studies (Table 1). However, in 1976, the WHO published the ICD-O-1 as a supplement to the ICD, Ninth Revision. At that time, six different classifications for NHL (WHO,Rappaport, Lukes/Collins, Dorfman, British, and Kiel) were inuse, and the ICD-O-1 tried to incorporate all of the varioushistological terms of these classifications, with no preferencegiven to any one classification scheme (4). However, no attemptwas made to translate the terminology of one classificationinto another, and some of the outdated terminology (Table

1Presented at the National Cancer Research Workshop, "The Emerging Epidemic of Non-Hodgkin's Lymphoma: Current Knowledge Regarding EtiologicalFactors." October 22-23. 1991, Bethesda, MD.

2 To whom requests for reprints should be addressed.3 The abbreviations used are: NHL, non-Hodgkin's lymphoma; ICD, Interna

tional Classification of Diseases; ICD-O, International Classification of Diseasesfor Oncology; ICD-O-1, the first ICD for Oncology; ICD-O-2, the second ICDfor Oncology; ICD-10, International Classification of Diseases, Tenth Revision;WF, Working Formulation; ML, malignant lymphoma; WHO, World HealthOrganization.

1) was retained. Thus, difficulty with the use of the ICD system for epidemiological studies was compounded rather thanimproved.

Use of the Working Formulation for HistologicalClassification

In response to increasing confusion and controversy regarding the clinical usefulness of the six NHL classifications, theNational Cancer Institute of the United States sponsored aninternational study in the late 1970s to determine whether oneclassification was more satisfactory than another (5). However,no one scheme was found to be superior to another and, instead,a WF for clinical usage was developed. The WF was not proposed as a new classification, but as a means of translationamong the various systems. The establishment of ten majorhistological types and three clinical grades in the WF was primarily based upon morphology and differences in survival, although other clinical parameters such as age and potential forcure were also considered. However, major deficiencies of theWF include the lack of a provision for classification based onimmunological type (i.e., B-cell, T-cell), the separation of someentities that are closely related biologically and the groupingtogether of others that are unrelated, and the lack of inclusionof new entities (i.e., mantle zone and intermediate lymphocyticlymphoma). Shortly after publication of the WF in 1982, temporary ICD-O codes were assigned to each of the WF histological types, and the utility of the new system for epidemiologicalstudies was evaluated (6).

The lymphoma section in the ICD-10 is based on the WF,and the corresponding ICD-O-2 provides morphology codes forall of the major NHL classification schemes in current use (7).The ICD-O-2 system has been field tested by the Surveillance,Epidemiology, and End Results (SEER) Program and will beformally adopted by all cancer registries in the United States in1992. The WF groups and ICD-O codes, as well as the corre

sponding Rappaport and Kiel terms and codes, are shown inTables 2 and 3. Table 4 lists a number of special types of NHLnot included in the WF, as well as codes for unclassifiable NHL.Cases of NHL are classified as having either a follicular (nodular) or diffuse growth pattern, and all lymphomas with anunspecified pattern are grouped in the diffuse category. Composite lymphomas, i.e., those with both a follicular and diffusepattern, should be categorized as follicular for epidemiologicalstudies, and the lowest grade morphology category should beused.

The ICD-O system also provides a section for coding bytopography, thus allowing the distinction to be made betweennodal and extranodal lymphomas. The majority of lymphomas(75%) arise in lymph nodes (topography code C77 ) orlymphatic tissues such as the tonsils, Waldeyer's ring, spleen, orthymus, and these are considered "nodal" lymphomas. The rest

of the lymphomas (25%) arise in extranodal sites such as thestomach, intestine, breast, or brain. When deciding whether alymphoma is nodal or extranodal, it is the primary site of thetumor that is to be considered. This distinction is important

5456s

on June 3, 2020. © 1992 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

PATHOLOGY OF NHL FOR EPIDEM1OLOGICAL STUDIES

Table 1 Eighth International Classification of lymphatic neoplasms

ICD code Histológica! type

200.0 Reticulum cell sarcoma200.1 Lymphosarcoma201.0 Hodgkin's disease

202.0 Giant follicular lymphoma202.1 Mycosis fungoides202.2 Other primary' lymphoid neoplasms (chloroma. compound

lymphoma, leukosarcoma, and malignant neoplasms nototherwise classifiable)

202.9 Other forms of lymphoma (benign lymphoma, benign orunspecified neoplasm of bone marrow, benign lymphoidpolyp, ocular lymphoma)

since extranodal lymphomas in certain organs may be linked tospecific etiologies.

Use of Paraffin-reactive Antibodies for Immunological

Classification

The ICD-O-2 system also provides codes for the immunolog-ical classification of NHL. These codes are given as the sixthdigit of the code for each case. Each NHL may be coded asT-cell type (code 5), B-cell type (code 6), or as immunologicaltype not determined (code 9). Nearly all cases of NHL can becategorized as having either a B-cell or a T-cell phenotype ifdetailed immunological and molecular biological studies areperformed on frozen tumor tissue. However, frozen tumor tissue is usually not available for use in epidemiológica! studies.

Recently, the development of reliable and specific monoclonal antibodies for the immunophenotyping of NHL in paraffin sections has obviated the need for frozen tissue (8-15). Alist of commonly used paraffin reactive antibodies and theirsources is given in Table 5. These antibodies are best used as apanel, since none is entirely specific, and patterns of reactivityshould be utilized for immunophenotypic interpretation. Onlythe diffuse NHLs need to be phenotyped, since follicular NHLis always of B-cell type. In a recent case-control study of 201NHLs in eastern Nebraska (16), using only the antibodies L26and UCHL-1 (9, 12), the authors found that 80% of the lymphomas were of B-cell type, 10% were of T-cell type, 5% were

of indeterminant phenotype due to inconclusive staining, andtissue was unavailable for study in the other 5% of cases. Segalet al. (15) have recently reported on the reliability and costeffectiveness of an antibody panel including L26, UCHL-1, andLeu 22 in NHL. In that study, 96% (76 of 79) of the lymphomaswere marked definitively using paraffin sections. These findingsindicate that paraffin immunophenotyping of NHL is highlyreliable and should be performed in future epidemiologicalstudies of NHL.

The immunophenotypes of the various histológica! categoriesof NHL in the WF, based on frozen tissue data from 1160 casesin the Nebraska Lymphoma Study Group Registry, are given inTable 6. All of the lymphomas in WF categories A to D and J,and the vast majority in categories E and G, are of B-cell origin,whereas categories F and H consist of a mixture of B- and T-cellcases. T-cell lymphomas predominate in category I, but thetotal number of cases is small. Overall, 88% of the 1160 casesof NHL were of B-cell type, and only 12% were of T-cell type.These findings should be largely reproducible using the paraffin-reactive antibodies listed in Table 5. As such, the WFappears to be a good system for categorization of the B-celllymphomas, but it is less than satisfactory for the T-cell lymphomas. With immunophenotyping, however, the special codesfor peripheral T-cell lymphoma (Table 4) could be used andmay result in the delineation of etiological entities.

Use of Expert Pathologists as Part of the Team

A major problem for epidemiological studies is the inaccuracy of NHL diagnoses by local pathologists ( 17-20), which canbe further compounded by a high rate of nonconcurrence (23%)in abstracted diagnoses (21). In addition, death certificates typically under-report deaths due to NHL and are likely to be apoor source of more specific information with regard to classification (22). A number of comparative studies (17-20) haveshown that the contributing pathologists' diagnosis of NHL is

confirmed by expert pathologists in 86 to 93% of the cases, butcomplete agreement with regard to pattern and cytology occursin only 49 to 56% of the cases. Major diagnostic discrepancies

Table 2 Working Formulation with related terms in the Rappaport and Kiel classifications and ICD-O code numbers

Working Formulation ICD-O code Rappaport classification Kiel classification

Low gradeA. ML, small lymphocytic 9670/3

ML, plasmacytoid lymphocytic 9671/3ML, consistent with chronic lymphocytic leukemia 9823/3

B. ML, follicular small cleaved cell 9695/3

C. ML, follicular mixed small cleaved and large cell 9691/3

Intermediate gradeD. ML, follicular large cell 9698/3E. ML, diffuse small cleaved cell 9672/3

F. ML. diffuse mixed small and large cell 9675/3

G. ML, diffuse large cell 9680/3cleaved 9681/3noncleaved 9682/3

High gradeH. ML, large cell, immunoblastic (diffuse) 9684/3I. ML, lymphoblastic (diffuse) 9685/3

J. ML, small noncleaved cell, Burkitt's 9687/3

(usually diffuse)ML, small noncleaved cell (usually diffuse) 9686/3

ML, lymphocytic, well differentiated 9670/3ML, lymphocytic, plasmacytoid 9671/3

ML, nodular lymphocytic, well differentiated9693/3. intermediate (mantle cell) 9694/3.poorly differentiated 9696/3

ML, nodular mixed lymphocytic-histiocytic9691/3

ML. nodular histiocytic 9698/3ML, diffuse lymphocytic, poorly differentiated

9672/3, intermediate (mantle cell) 9673/3ML, diffuse mixed lymphocytic-histiocytic

9675/3ML, diffuse, histiocytic 9680/3

ML, lymphocytic 9670/3ML. lymphoplasmacytic 9671/3ML, consistent with chronic

lymphocytic leukemia 9823/3ML. centroblastic/centrocytic

9692/3

ML, centroblastic/centrocytic9692/3

ML, centroblastic, follicular 9697/3ML, centrocytic 9674/3

ML, diffuse centroblastic-centrocytic9676/3

ML, diffuse centroblastic 9683/3

ML, diffuse, histiocytic 9680/3 ML. immunoblastic 9684/3ML. lymphoblastic, convoluted/nonconvoluted ML, lymphoblastic 9685/3

9685/3ML. undifferentiated, Burkitt's 9687/3 Burkitt's lymphoma 9687/3

ML, undifferentiated, non-Burkitt's 9686/3

5457s

on June 3, 2020. © 1992 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

PATHOLOGY OF NHL FOR EP1DEMIOLOG1CAL STUDIES

Table 3 Interrelationships of the Rappaport and Kiel classifications of non-Hodgkin's lymphoma (1CD-O codes) and the Working Formulation groups

A to J denote Working Formulation groups (see Table 2).

Lymphocytic (smallcell)NOS"or

diffuseA.

96709671E.

9672967396749592Nodular

orfollicularB.

9696969396949695Histiocytic

(largecell)NOS

ordiffuseG.

968096819682

9683H.

96849593Nodular

orfollicularD.

96989697Mixed

lymphocytic and histiocytic(small and largecell)NOS

ordiffuseF.

96759676Nodular

orfollicularC.

96919692Other

malignantlymphomasNOS

ordiffuseI.

9685J.

96869687

' NOS, growth pattern not otherwise specified.

Table 4 Special and unclassified non-Hodgkin 's lymphomas and ICD-O code numbers

9592/39593/39594/3

9677/39700/39701/39702/3

9703/39704/3

95909591

Other non-Hodgkin 's lymphomas

LymphosarcomaReticulum cell sarcomaMicroglioma (brain)

Malignant lymphomatous polyposisMycosis fungoidesSezary's diseasePeripheral T-cell lymphoma, not otherwise

specifiedPeripheral T-cell lymphoma, T-zoneLymphoepithelioid lymphoma

9705/3 Peripheral T-cell lymphoma, angioimmunoblastic9706/3 Peripheral T-cell lymphoma, pleomorphic small cell9707/3 Peripheral T-cell lymphoma, pleomorphic medium and

large cell9711/3 Monocytoid B-cell lymphoma9712/3 Angioendotheliomatosis9713/3 Angiocentric T-cell lymphoma9714/3 Anaplastic large cell (Ki-1) lymphoma

Unclassified non-Hodgkin 's lymphomas

Malignant lymphoma, not otherwise specifiedMalignant lymphoma, non-Hodgkin's, not otherwise

specified

9595 Malignant lymphoma, diffuse, not otherwise specified9690 Malignant lymphoma, nodular (follicular), not otherwise

specified9709 Cutaneous lymphoma, not otherwise specified

Table 5 Paraffin antibodies which are useful for immunophenotypingnon-Hodgkin's lymphoma

AntibodyL26

(CD20)MB2LN2 (CD74)UCHL-1 (CD45RO)T3 (CD3)Leu 22 (CD43)SourceDako

BiotestICNDakoDakoBecton DickinsonB-cell

T-cell±

+

between contributing pathologists and expert pathologists occurred in 21% of the cases in one study (19), and a diagnosisother than NHL is made by the experts in about 5% of cases(19, 20). In the only published study utilizing the WF, Dick etal. (20) found good concordance (80%) between the diagnosesof the contributing pathologists and the experts for follicularlymphoma and diffuse small lymphocytic lymphoma. However,concordance with regard to the other diffuse lymphomas wasonly 41%. Dick et al. (20) also found that contributing pathologists often underdiagnose follicular lymphoma, resulting in afalse increase in diffuse lymphoma, and often include cases ofchronic lymphocytic leukemia with small lymphocytic lymphoma. Other studies (18, 19) have also found a low rate of concordance (45 to 55%) for diffuse lymphoma, but they have alsofound a lower rate for follicular lymphoma (52 to 58%).

The above findings indicate that expert pathologists areneeded in epidemiological studies of NHL to provide accurateand consistent histological classification of the cases and, thus,decrease the inordinately high rate of diagnosis misclassifica-tion which would tend to obscure any relationships betweenexposure and disease. Expert pathologists are also needed tointerpret the immunohistochemical stains for B- and T-cellphenotyping and to correlate these results with the histologicalfindings. In addition, expert pathologists often have a good

working relationship with the contributing pathologists andclinicians involved in the care of patients in epidemiologicalstudies, and they can facilitate the performance of such studiesat the regional and community levels. Finally, expert pathologists have the knowledge and expertise to advise epidemiologists with regard to the design and evaluation of such studies,and they should be included from the start in the planning ofsuch projects.

Understanding the Biology of Non-Hodgkin's Lymphoma

In order to use a histological classification of NHL for epidemiological purposes, one must first understand the biology ofNHL and how the various histological types relate to one another in the normal differentiation scheme (1, 2). Pluripotentstem cells in the bone marrow give rise to both immature B-cellsand immature T-cells in the adult human. These cells thenmigrate via the blood to the peripheral lymphoid organs (lymphnodes, spleen, thymus) where they differentiate into mature,functional cells of the immune system. Cancers of the immunesystem, i.e., NHL or lymphoid leukemia, arise when geneticevents occur within immature lymphoid cells in the bone marrow or more mature cells in the peripheral organs (1, 23). Theseevents result in chromosomal aberrations which turn key geneson or off and cause the cells to become malignant, proliferate inan uncontrolled manner, and spread to other sites. The causesof the cytogenetic events which result in lymphoid neoplasia arelargely unknown, some being random events and others due toenvironmental genotoxins such as organic chemicals or ionizing radiation. Lymphoid malignancies of the bone marrow andperipheral blood are called leukemia, either acute or chronic,whereas those arising in the peripheral organs are called NHL.As such, the distinction between lymphoid leukemia and NHLis largely distributional with overlapping entities, such that a

5458s

on June 3, 2020. © 1992 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

PATHOLOGY OF NHL FOR EPIDEMIOLOGICAL STUDIES

Table 6 Immunophenotypes of 1160 non-Hodgkin's lymphomas according

to the Working FormulationGerminal Center

Working Formulation%ofcases

B-cell T-cell

Low gradeA. ML, small lymphocytic 14 100 0B. ML, follicular small cleaved cell 5 100 0C ML, follicular mixed cell 12 100 0

Intermediate gradeD. ML. follicular large cell 15 100 0E. ML, diffuse small cleaved cell 6 92 8F. ML, diffuse mixed cell 6 36 64G. ML, diffuse large cell 22 94 6

High gradeH. ML, large cell, immunoblastic 15 73 27I. ML, lymphoblastic 2 17 83J. ML, small noncleaved cell 3 100 0

Total 100 88 12

particular lymphoid neoplasm may manifest both lymphoma-tous and leukemic features during the course of the disease. Forexample, lymphoid leukemias can sometimes spread to the peripheral lymphoid organs and present clinically like NHL, andthe cells of NHL sometimes spread to the blood and manifest asleukemia. The tissue correlate of chronic lymphocytic leukemiais small lymphocytic NHL, and that of acute lymphoblasticleukemia is lymphoblastic NHL. The pathological classification of lymphoid neoplasia is usually based upon the clinicalnature of the disease (i.e., acute, chronic, low grade) and thepredominating cell type within the neoplastic proliferation,even though the key genetic events may occur in a cell at anearlier stage of differentiation.

Although the precise sequence of B-cell differentiation is unknown, considerable progress has recently been made in ourunderstanding of the development and function of the normalhumoral immune system (2). Currently, the various types ofB-cell neoplasia are thought to represent cells arrested at various stages in the normal differentiation scheme. Simplified diagrams showing the various stages of the normal humoral immune response and their corresponding B-cell neoplasms aregiven in Fig. 1 to 3. Fig. 1 shows the overall scheme of B-celldifferentiation, whereas Fig. 2 shows the cells of the germinal

Fig. 2. Sequence of normal B-cell differentiation within the germinal center.Each malignancy represents a successive stage at which neoplastic differentiationis arrested. SNC, small noncleaved cell lymphoma; LNC. large noncleaved celllymphoma; LC, large cleaved cell lymphoma; SC, small cleaved cell lymphoma.[Reprinted from D. D. Weisenburger et al., 1987, p. 541, with permission (35)).

IB-PC

Fig. 3. Sequence of normal B-cell differentiation in the primary or secondaryplasma cell reaction. Each malignancy represents a successive stage at whichneoplastic differentiation is arrested. LNCFCC, large noncleaved follicular centercell lymphoma: SL, small lymphocytic lymphoma; IB-PC, plasmacytoid immunoblastic lymphoma; PC, plasma cell myeloma. [Reprinted from D. D. Weisenburger et al., 1990, p. 103, with permission (2)].

center reaction and Fig. 3 shows the cells of the plasma cellreaction which arise from mature (or memory) B-cells. Thereader is encouraged to refer to a recent review of this topic (2)for the rationale behind and a more detailed description of thisprocess.

Basically, the various B-cell neoplasms appear to recapitulatethe normal stages of B-cell differentiation. The lymphocyticneoplasms consist predominantly of nondividing (resting) cells,whereas the blastic neoplasms consist of cells in the prolifera-tive phase of the cell cycle. Partial blocks in differentiation arenot uncommon and result in the development of "mixed cell"

lymphomas consisting of cells at closely related stages of differentiation. However, some of the categories of B-cell NHL inthe WF are not pure and consist of tumors corresponding tocells at different and sometimes distant stages of differentiation. However, there is currently no simple or practical means

CORTICAL FOLLICLE

Manti« /on«

Fig. 1. A simplified diagram showing thevarious stages of the normal humoral immuneresponse. Each malignancy represents a successive stage at which neoplastic differentiation is arrested. AUL, acute undifferentiatedleukemia: cALL, common acute lymphoblasticleukemia; i'/./.. chronic lymphocytic leuke

mia; SL, small lymphocytic lymphoma; IL,intermediate lymphocytic lymphoma; FCC,follicular center cell lymphoma; SL-PC,plasmacytoid-small lymphocytic lymphoma.¡Reprinted from D. D. Weisenburger et al.,1987. p. 781, with permission (34)).

PRIMARY IMMUNE,^ESPONSE SECONDARY IMMUNE RESPONSE

Stem C«l<AUL)

PRIMARY IMMUNE RESPONSE

O(A)V

—¿�<"•'

5459s

on June 3, 2020. © 1992 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

PATHOLOGY OF NHL FOR EPIDEMIOLOGICAL STUDIES

by which to separate these latter tumors into more pure subgroups. The reader is again encouraged to refer to the above-mentioned review (2) for a more detailed description of thevarious types of B-cell neoplasia and their interrelationships. Abrief summary of the various NHL cell types, their ICD-O-2codes, and their cells of origin and relationships is given inTable 7, which can be correlated pictorially with the cells in Fig.1 to 3.

In a similar manner, pluripotent stem cells in the bone marrow give rise to immature or pre-T-cells which then migrate viathe blood to the thymus (1). In the thymus, the cells migratefrom the cortex to the medulla during the process of differentiation, and they then leave the thymus as mature 1 -cells and

migrate to the peripheral lymphoid organs. The lymphomas ofT-cell lineage can be divided into two categories based on theirorigin from immature T-cells or mature T-cells. The lymphoidneoplasms of immature T-cells are acute lymphoblastic leukemia and lymphoblastic NHL (WF category I), whereas all of theNHLs arising from mature T-cells can be grouped together asperipheral T-cell NHL (WF categories E to H) or subdividedinto special subtypes (Table 4). However, our understanding ofthe biology of the T-cell system is less complete then that of theB-cell system, and the sequence and morphology of matureT-cell activation and differentiation have not been well characterized. However, mycosis fungoides and Sézary'ssyndrome

correspond to mature T-cells which home to and reside in theskin and, as such, represent a special subset of peripheral T-celllymphoma.

Use of the Working Formulation (ICD-O-2) forEpidemiológica! Studies

Some epidemiologists have suggested that hematopoieticcancers should be subdivided as finely as possible in epidemiológica!studies in order to maximize the prospect that homogeneous etiological entities are studied, and they state that thevarious types of hematopoietic cancer should only be groupedtogether for investigative purposes when there is some indication that the cancers may share a common etiology. Unfortunately, NHL cannot be subdivided into standard etiological

categories at the present time, since the etiologies and patho-genesis of NHL are not well understood. Furthermore, sincethe various types of NHL are related to one another and they allarise from a common hematopoietic stem cell, it is valid togroup all of the various types of NHL together for epidemic-logical study. This will ensure that diseases of similar etiology,even those which are misclassified, will be included in the studyand that useful information will not be lost. However, this maydilute important data to the point that small increases in riskassociated with certain subtypes may be missed. On the otherhand, it is likely that any significant increases in risk that areidentified will actually underestimate the true risk due to dilution of the data. However, specific disease entities should alsobe analyzed separately in order to distinguish important differences in target cell specificity for various exposures which maybe etiologically important. Obviously, NHL could be simplysubgrouped by immunophenotype (i.e., B-cell and T-cell types).A variety of other possible etiological groups of NHL and related disorders is given in Table 8. The standard WF groups, thecodes for which are given in the second column of Table 2 underICD-O code, and some more specific ICD-O categories for thevarious entities within each group are also given in Table 8. Ingeneral, use of the standard WF groups will tend to dilute thedata more than use of the more specific categories since some ofthe WF groups are heterogeneous (Table 7). However, eitherapproach should be useful for most types of epidemiologicalstudy. Other group combinations may also be appropriate depending on the nature of the study, and consultation with anexpert pathologist should always be obtained prior to any aggregate analysis.

Latency Period of Non-Hodgkin's Lymphoma

The latency period for the development of NHL following anenvironmental exposure is largely unknown. One valid sourceof information on the latency period of NHL can be foundin the literature on NHL developing after the treatment ofHodgkin's disease with chemotherapy and/or radiotherapy

(24, 25). In such studies, the median latency period for NHL isabout 5 to 6 yr (24, 25), not unlike that for secondary acute

Table 7 Histological types of B-cell non-Hodgkin's lymphoma and their origins and interrelationships

Working Formulation ICD-O-2 codes Cellular origin and relationships

A. ML, small lymphocytic

ML, plasmacytoid lymphocytic

B. ML, follicular mantle cell, intermediatelymphocytic

B-D. ML, follicular center cell

E. ML. diffuse small cleaved cell

ML, diffuse mantle cell, intermediatelymphocytic

F. ML, diffuse mixed cell

G. ML, diffuse large cell, cleaved and noncleaved

H. ML, large cell, immunoblastic. plasmacytoid

I. ML. lymphoblastic

J. ML, small noncleaved cell

9670/3

9671/3

9693/3, 9694/3

9695/3, 9696/3, 9692/3, 9691/3,9692/3, 9698/3, 9697/3

9672/3

9673/3, 9674/3

9675/3, 9676/3

9780/3, 9681/3, 9682/3, 9683/3

9684/3, 9680/3

9685/3

9686/3, 9687/3

Majority correspond to immature virgin It-cells of the diffusecortex; closely related to chronic lymphocytic leukemia; minoritycorrespond to memory B-cells

Mature B-cells of the primary or secondary (memory) response withplasma cell differentiation

Corresponds to immature virgin B-cells of the primary follicle andmantle zone; closely related to small lymphocytic

Corresponds to cells of the follicular germinal center; closely relatedto each other and to their diffuse follicular center cell counterparts

Diffuse counterpart of follicular small cleaved cell NHL: closelyrelated to other follicular center cell NHL

Diffuse counterpart of follicular mantle cell NHL; closely related tosmall lymphocytic

Diffuse counterpart of follicular mixed-cell NHL; closely related toother follicular center cell NHL

About 40 to 50% are diffuse counterparts of follicular large-cellNHL; remainder correspond to transformed cells at earlier orlater stages

About 30% are closely related to diffuse large-cell NHL of follicularcenter cell origin; remainder likely correspond to transformedcells of the primary or secondary plasma cell reaction

Majority correspond to pre-B- or early bone marrow B-cells andare closely related to common and other types of acuteB-lymphoblastic leukemia

Rarely follicular; majority correspond to cells of the early germinalcenter reaction and are closely related to other follicular centercell NHL; minority may correspond to an earlier B-cell

5460s

on June 3, 2020. © 1992 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

PATHOLOGY OF NHL FOR EPIDEMIOLOGICAL STUDIES

Table 8 Possible etiologic groups of non-Hodgkin's lymphoma and related

disorders

B-cell type (88%)

1. Small lymphocytic lymphoma (A": or 9670/3) + chronic lymphocytic

leukemia (9823/3)2. Small lymphocytic lymphoma (A; or 9670/3 + 9693/3) + intermediate

lymphocytic lymphoma (E; or 9694/3 + 9673/3)3. Follicular lymphoma (B + C + D; or 9696/3 + 9691/3 + 9698/3)4. Follicular center cell lymphoma (B-l-C+D + F + G + J;or 9696/3

+ 9691/3 + 9698/3 + 9675/3 + 9680/3 + 9686/3 + 9687/3)5. Diffuse large cell lymphoma (F + G + H: or 9675/3 + 9680/3)6. Blastic lymphoma (G + H + J: or 9680/3 + 9682/3 + 9686/3 + 9687/3)

T-cell type (12%)

1. Lymphoblastic lymphoma (I; or 9685/3) + acute lymphoblastic leukemia"(9821/3)

2. Peripheral T-cell lymphoma (E + F + G + H; or 9672/3 + 9675/3+ 9680/3)

3. Mycosis fungoides/Sézary'sdisease (9700/3 + 9701/3)

" A to J denote Working Formulation groups (see Table 2).

TIME -

Fig. 4. Idealized latency curves for non-Hodgkin's lymphoma associated with

short-lerm, high-dose (A) and long-term, low-dose (fi) carcinogenic exposures.

leukemia (26, 27). In these studies (24, 25), the latency of NHLin some cases was as short as 2 yr, but it may also be as long as>15 yr (28). In contrast, the average latency period for thedevelopment of malignant lymphoma (including Hodgkin's dis

ease) among young (<25 yr) atomic bomb survivors who received a radiation dose of > 100 rads was about 9 yr, with alonger latency for those receiving smaller doses (29). Interestingly, the average latency was 14 yr in older survivors whoreceived >100 rads (29), suggesting that the latency period maybe dependent on age. In contrast, the median latency period foracute leukemia associated with chronic, low-dose exposure tobenzene is 15 to 20 yr (27, 30), and the latency in similarsituations would be expected to be similar for NHL.

An idealized graph of the above data showing latency curvesfor NHL due to short-term, high-dose (Curve A) and long-term,low-dose (Curve B) carcinogenic exposures is shown in Fig. 4.Based on these data, short-term, high-dose exposures would beexpected to result in a short latency period, whereas long-term,low-dose exposures would be expected to result in a long latency period. This relationship of dose and exposure time tolatency has also been demonstrated in numerous carcinogenesisstudies in animals. For example, Melnick et al. (31, 32) haveperformed animal studies in which exposure to 1,3-butadienewas stopped after limited periods of time to assess the effects ofexposure level and duration of exposure on the outcome ofbutadiene-induced NHL. In these studies, NHL was induced inmice exposed to 625 ppm of butadiene for only 13 wk. Theincidence of lymphoma in mice exposed to 625 ppm of butadiene for 26 wk was approximately 2 times that in mice exposedfor only 13 wk. However, when the exposure was reduced byone half to 312 ppm and exposure duration was extended to 52

wk, the incidence of NHL was reduced by 90%. Thus, themultiple of the exposure concentration times the exposure duration (cumulative dose) did not predict the incidence of lymphoma in mice. Peto (33) has also suggested that the cumulativedose of a carcinogen may be less predictive of an increased riskof cancer in humans than the intensity of the exposure. Thus,knowledge of the nature and timing of carcinogenic exposuresshould be considered in the design of epidemiológica! studies.For example, in the setting of an environmental carcinogen, theaccepted belief that risk ratios should be higher for long-termworkers than for short-term workers may not always be true,particularly if the short-term workers received the highest exposures. Also, short-term workers with high exposures arelikely to have a shorter latency period than long-term workerswith lower exposures. These considerations are important forthe design and interpretation of epidemiological studies, sincesuch studies may arbitrarily eliminate short-term workers fromthe study or attribute increased risks in such workers to priorexposures.

Conclusions

In summary, current knowledge and understanding of thebiology and classification of NHL are important for the designand evaluation of epidemiological studies. Many of the problems of the past can be eliminated by the use of the WF andICD-O-2 systems for the classification of NHL; the use ofimmunological markers; and the collaboration of epidemiologists and expert pathologists in the design, performance, andevaluation of such studies. Future studies which incorporate theconcepts put forth herein will be more likely to identify endogenous and environmental factors which play a role in the etiology and pathogenesis of NHL.

References

1. Magrath, I. Lymphocyte ontogeny: a conceptual basis for understandingneoplasia of the immune system. In: I. T. Magrath (ed.). The Non-Hodgkin'sLytnphomas. pp. 29-48. Baltimore. MD: Williams & Wilkins. 1990.

2. Weisenburger, D. D.. Harrington. D. S., and Armitage. J. O. B-cell neoplasia:a conceptual understanding based on the normal humoral immune response.In: P. P. Rosen and R. E. Fechner (cds.). Pathology Annual. Part I, Vol. 25.pp. 99-115. Norwalk, CT: Appleton and Lange, 1990.

3. Tsongas. T. A. Occupational factors in the epidemiology of chemically induced lymphoid and hematopoielic cancers. In: R. D. Irons (ed.). Toxicologyof the Blood and Bone Marrow, pp. 149-177. New York, NY: Raven Press,1985.

4. WHO. International Classification of Diseases for Oncology, Ed. 1, pp.40-43. Geneva, Switzerland: WHO, 1976.

5. The Non-Hodgkin's Lymphoma Pathologic Classification Project. NationalCancer Institute sponsored study of classification of non-Hodgkin's lympho-

mas: summary and description of a Working Formulation for clinical usage.Cancer (Phila.), 49: 2112-2135, 1982.

6. Percy, C., O'Conor, G., Gloeckler Reis, L., and Jaffe, E. S. Non-Hodgkin's

lymphomas: application of the International Classification of Diseases forOncology (ICD-O) to the Working Formulation. Cancer (Phila.), 54: 1435-1438, 1984.

7. WHO. International Classification of Diseases for Oncology, Ed. 2, pp.45-47. Geneva, Switzerland: WHO, 1990.

8. Norton, A. J.. Ramsay. A. D., Smith. S. H., Beverley, P. C. L.. and Isaacson,P. G. Monoclonal antibody (UCHL1) that recognizes normal and neoplasticT-cells in routinely fixed tissues. J. Clin. Pathol., 39: 399-405. 1986.

9. Linder, J., Ye, Y., Harrington, D. S., Armitage, J. O., and Weisenburger, D.D. Monoclonal antibodies marking T lymphocytes in paraffin-embeddedtissue. Am. J. Pathol., 127: 1-8, 1987.

10. Strickler, J. G., Weiss. L. M., Copcnhaven, C. D., Bindl, J.. McDaid. R..Buck. D.. and Warnke, R. S. Monoclonal antibodies reactive in routinelyprocessed tissue sections of malignant lymphoma, with emphasis on T-celllymphomas. Hum. Pathol., 18: 808-814. 1987.

11. Poppcma, S., Hollema, H.. Visser, L., and Vos, H. Monoclonal antibodies(MT1, MT2. MBI, MB2. MB3) reactive with leukocyte subsets in paraffin-embedded tissue sections. Am. J. Pathol., 127: 418-429, 1987.

12. Linder, J., Ye, Y., Armitage. J. O.. and Weisenburger, D. D. Monoclonalantibodies marking B-cell non-Hodgkin's lymphoma in paraffin-embeddedtissue. Mod. Pathol., /: 29-34, 1988.

5461s

on June 3, 2020. © 1992 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

DISCUSSION

13. Andrade. R. E., Wick, M. R., Frizzerra, G., and Gajl-Peczalska, K. J. Im-munophenotyping of hematopoietic malignancies in paraffin sections. Hum.Palhol., 19: 394-402, 1988.

14. Mason, D. Y.. Krissansen. G. W., Davey, F. R., Crumpton, M. J., and Gatter,K. C. Antisera against epitopes resistant to denaturation on T3 (CD3) antigencan detect reactive and neoplastic T-cells in paraffin embedded tissue biopsysamples. J. Clin. Pathol., 41: 121-127, 1988.

15. Segal. G. H., Stoler, M. H., Fishleder, A. J., and Tubbs, R. R. Reliable andcost-effective paraffin section immunohistology of lymphoproliferative disorders. Am. J. Surg. Pathol., 15: 1030-1041, 1991.

16. Zahm, S. H., Weisenburger, D. D.. Babbitt, P. A., Saal, R. C., Vaught, J. B.,Cantor, K. P.. and Blair, A. A case-control study of non-Hodgkin's lym-phoma and the herbicide 2,4-dichlorophenoxyacetic acid (2,4-D) in EasternNebraska. Epidemiology, /: 349-356. 1990.

17. Kim, H., Zelman. R. J., Fox, M. A., el al. Pathology Panel for LymphomaClinical Studies: a comprehensive analysis of cases accumulated since inception. J. Nati. Cancer Inst., 68: 43-67, 1982.

18. Velez-Garcia. !•'..Durant, J.. Gams, R., and Bartolucci, A. Results of a

uniform histopathology review system of lymphoma cases. A ten-year studyfrom the Southeastern Cancer Study Group. Cancer (Phila.). 52: 675-679.1983.

19. Bird, C. C, Lauder, I.. Kellett, H. S., Chorlton, I., Barnes, N.. Darwin, C,Cartwright, R. A., and Boyko, R. Yorkshire Regional Lymphoma Histopathology Panel: analysis of five years' experience. J. Pathol., 143: 249-258,

1984.20. Dick. F.. Van Lier. S., Banks. P.. Frizzerra, G., Witrak. G.. Gibson. R..

Everett, G., Schuman, L.. Isaacson. P.. O'Conor. G., Cantor. K.. Blattner,W., and Blair, A. Use of the Working Formulation for non-Hodgkin's lym

phoma in epidemiologie studies: agreement between reported diagnoses anda panel of experienced pathologists. J. Nati. Cancer Inst., 78: 1137-1143,1987.

21. Dick, F.. Van Lier. S. F., McKean, K., Everetl. G. D., and Blair. A. Non-concurrence of abstracted diagnoses of non-Hodgkin's lymphoma. J. Nati.

Cancer Inst., 78: 675-678. 1987.22. Percy, C., Stanek, E., and Gloeckler, L. Accuracy of cancer death certificates

and its effect on cancer mortality statistics. Am. J. Public Health, 71: 242-250, 1981.

23. Haluska, F. G., Tsujimoto, Y., and Croce, C. M. The molecular genetics ofnon-Hodgkin's lymphoma. In: I. T. Magrath (ed.). The Non-Hodgkin's Lym-phomas. pp. 96-108. Baltimore, MD: Williams & Wilkins, 1990.

24. Jacquillat, C. Khayat, D.. Desprez-Curely, J. P.. Weil. M., Brocheriou, C.,

Auclerc, G., Chamseddine, N., and Bernard, J. Non-Hodgkin's lymphomaoccurring after Hodgkin's disease. Four new cases and a review of the literature. Cancer (Phila.), 53: 459-462, 1984.

25. Bennett, M. H., MacLennan. K. A.. Vaughan Hudson, G., and VaughanHudson, B. Non-Hodgkin's lymphoma arising in patients treated forHodgkin's disease in the BLNI: a 20-year experience. Ann. Oncol. 2 (supplì:83-92, 1991.

26. Tucker, M. A., Coleman, C. N., Cox, R. S., Varghese, A., and Rosenberg, S.A. Risk of second cancers after treatment for Hodgkin's disease. N. Engl. J.Med., J«:76-81, 1988.

27. Rosner, F., and Grunwald, H. W. Chemicals and leukemia. In: E. S. Henderson and T. A. Lister (eds.). Leukemia, pp. 271-287. Philadelphia, PA: W.B. Saunders. 1990.

28. Coleman, C. N., Kaplan, H. S., Cox, R., Varghese, A., Bulterfield. P., andRosenberg, S. A. Leukemias, non-Hodgkin's lymphomas, and solid tumors inpatients treated for Hodgkin's disease. Cancer Surv., /: 733-744, 1982.

29. Nishiyama, H., Anderson, R. F., Ishimaru, T., Ishida, K., li, Y., and Okabe.N. The incidence of malignant lymphoma and multiple myeloma in Hiroshima and Nagasaki atomic bomb survivors. 1945-1965. Cancer (Phila.),32: 1301-1309, 1973.

30. Rinsky, R. A., Smith, A. B., Hornung. R., Filloon, T. G., Young, R. J., Okun,A. H., and Landrigen, P. J. Benzene and leukemia. An epidemiologie riskassessment. N. Engl. J. Med.. 316: 1044-1050, 1987.

31. Melnick, R. L., Huff, J.. Chou. B. J., and Miller, R. A. Carcinogenicity of1.3-butadiene in C57BL/6 x C3HF, mice at low exposure concentrations.Cancer Res., 50: 6592-6599, 1990.

32. Melnick, R. L., Huff, J. E.. Rayero«,J. H., Chou, B. J., and Miller, R. A.Inhalation toxicology and Carcinogenicity of 1,3-butadiene in B6C3F! micefollowing 65 weeks of exposure. Environ. Health Perspect., 86: 27-36, 1990.

33. Peto, J. Problems in dose-response and risk assessment. In: D. G. Hoel, R. A.Merrill, and F. P. Perera (eds.). Risk Quantitation and Regulatory Policy.Banbury Report 19, pp. 89-97. Cold Spring Harbor. NY: Cold Spring Harbor Laboratory. 1985.

34. Weisenburger. D. D.. Linder. J.. Daley, D. T.. and Armitage. J. O. Intermediate lymphocytic lymphoma: an immunohistologic study with comparisonto other lymphocytic lymphomas. Hum. Pathol., /*: 781-790, 1987.

35. Weisenburger, D. D., Harrington, D. S., and Armitage, J. O. B-cell neoplasiarecapitulates the normal humoral immune response. In: R. P. Gale and D. W.Golde (eds.). Recent Advances in Leukemia and Lymphoma, pp. 535-543.New York, NY: Alan R. Liss, 1987.

Discussion

Mrs. Percy: I am an Editor of the ICD-O-2' and I want to

bring you up to date on a few things. First of all, the SEERProgram and most of the major registries in the United Statesfollowing the SEER guidelines have been using the equivalentof ICD-O-2, i.e., the field trial edition since the beginning of1988. Even before that, when Dr. O'Conor and myself and

others here assigned ICD-O numerics to specific morphologicalgroups in the working formulation, our published paper permitted retrieval of comparable data back to about 1984.

I believe mortality from NHL is definitely underreportedbecause of AIDS; when AIDS is associated with NHL, myimpression is that since 1987, when they introduced the newcodes for AIDS, most of those cases were coded with an underlying cause of death as AIDS. Unless you do multiple causeanalysis, you don't get the NHL.

Dr. Zahm: I wondered if anyone has looked at the category oftumors of ill-defined, unknown primary site to see if that'schanged over time. I'm wondering if some of the cases may be

the extranodal NHL cases; in particular, they may be movingfrom that category.

Dr. Cartwright: Yes. In the study I referred to where wereanalyzed the 1960 through 1967 cases, we also included the

1The abbreviations used are: ICD-O-2. International Classification of Diseasesfor Oncology, second edition: SEER, Surveillance. Epidemiology, and End Results;NHL, non-Hodgkin's lymphoma: AIDS, acquired immune deficiency syndrome;HD, Hodgkin's disease; CLL, chronic lymphocytic leukemias.

category of undifferentiated carcinomas and found—with mid-1970s to mid-1980s techniques—10% of those in fact werelymphomas of one sort or another.

Dr. O'Conor: This would appear then to be a significant group

of tumors where misclassification could have some effect ontrends.

Dr. Cartwright: Yes.Participant: Dr. Cartwright, I'd like to ask you what other

disease categories were examined? You said that you looked ateye diseases and some other diseases.

Dr. Cartwright: Yes. The main category of error was HD.When we looked at the so-called HD, we did indeed find 8-10%who were NHL.

Dr. O'Conor: I recall that in the studies that Dr. Correa and

I did in Connecticut and California, we had a similar experience. Approximately 10% of cases that had been diagnosed asHD we reclassified as NHL.

Dr. Banks: May I ask what relative numbers we're talking

about? When we talk about undifferentiated carcinoma, 10% ofthat category being recruited, what is the absolute number ofsuch reported cases in relation to the absolute number in percentage of NHL? Is this significant?

Dr. Cartwright: No. The numbers are low.Dr. Banks: I think this is very important because, for the

nonepidemiologists in the group this morning, having thesocioeconomic and educational levels correlated with riskfor NHL immediately makes one suspect that maybe they

5462s

on June 3, 2020. © 1992 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

1992;52:5456s-5461s. Cancer Res   Dennis D. Weisenburger  Epidemiological StudiesPathological Classification of Non-Hodgkin's Lymphoma for

  Updated version

  http://cancerres.aacrjournals.org/content/52/19_Supplement/5456s

Access the most recent version of this article at:

   

   

   

  E-mail alerts related to this article or journal.Sign up to receive free email-alerts

  Subscriptions

Reprints and

  .pubs@aacr.orgDepartment at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

  Permissions

  Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/52/19_Supplement/5456sTo request permission to re-use all or part of this article, use this link

on June 3, 2020. © 1992 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from