research paper
Bone marrow histological patterns can predict survival of patients with grade 1 or 2 follicular lymphoma: a study from the Groupe d‘Etude des Lymphomes Folliculaires
Danielle Canioni,1 Pauline Brice,2 Eric Lepage,3 Myrna Chababi,1 Ve´ronique Meignin,3 Bruno Salles,4 Luc Xerri,5 Pierre-Yves Pe´aud,6 Philippe Rousselot,3 Michel Peuchmaur,7 Philippe SolalCe´ligny8 and Nicole Brousse1 for The Groupe d’Etude des Lymphomes Folliculaires9 1
Pathology Department, Hoˆpital Necker-Enfants Malades, Paris, 2Hematology Institute, Hoˆpital
Saint-Louis, Paris, 3Biostatistics and Medical Informatics Department, Hoˆpital Henri-Mondor, Cre´teil, 4Medicine Institute, Hoˆpital de Chaˆlon, Chalon/Saoˆne, 5Pathology Department, Institut Paoli-Calmette, Marseille, 6Hematology Department, Hoˆpital de Valence, Valence, 7 Pathology Department, Hoˆpital Robert-Debre´, Paris, 8Hematology Department, Centre Jean Bernard, Le Mans, and 9Groupe d’Etude des Lymphomes Folliculaires (GELF), Hoˆpital, SaintLouis, Paris, France
Received 30 January 2004; accepted for publication 26 April 2004 Correspondence: N. Brousse, Hoˆpital NeckerEnfants Malades, 149, rue de Se`vres, 75743 Paris
Summary The influence of bone marrow biopsy (BMB) histology on prognosis and management of follicular lymphomas (FL) remains controversial. A total of 390 patients with grade 1 or 2 FL were prospectively included in the multicentric Groupe d’Etude des Lymphomes Folliculaires trial and their BMB reviewed in order (i) to quantify the ratio of lymphomatous foci (LFo) area to that of BMB size (LFo/BMB), (ii) to determine the BMB patterns for a practical grading of marrow infiltration, (iii) to assess the intra- and interobserver reproducibility of this grading and (iv) to analyse this grading on event-free (EFS) and overall survival (OS), using univariate and multivariate analyses. A total of 267 patients (68%) had BMB involvement, with inter- and intra-observer reproducibility for classifying the patterns of involvement of 91 and 96%, respectively. Uni- and multivariate analyses demonstrated the adverse influence of (i) a ratio of LFo/BMB ‡0Æ1, i.e. three or four nodules/ medullary space or ‡1 nodule + foci of diffuse involvement on EFS (P ¼ 0Æ03) and (ii) two different histological patterns in the same BMB on EFS (P ¼ 0Æ004) and OS (P ¼ 0Æ001). This latter finding was only significant in patients with a high tumour burden and remained significant in multivariate analysis. These results indicate that BMB histology can predict survival of FL patients with a high tumour burden, and may help in defining their treatment. Keywords: bone marrow pathology, follicular lymphoma, prognosis, statistical analysis.
Cedex 15, France. E-mail:
[email protected]
Follicular lymphomas (FL) are classified as ‘mature B-cell neoplasms’ in the Revised European-American Classification of Lymphoid Neoplasms (REAL) and the WHO (World Health Organization) classification (Harris et al, 1994, 1999), and as low-grade B-cell lymphomas in the Working Formulation for Clinical Usage (Non-Hodgkin’s Lymphoma Pathologic Classification Project, 1982) and in the International Lymphoma Study Group Classification (Chan et al, 1994). They represent lymphomas with an indolent course and patients with FL survive for a median of 7–10 years. However, FL spread early during their evolution and have usually reached a disseminated stage at presentation (stages III/IV) doi:10.1111/j.1365-2141.2004.05046.x
and at each progression, around 15–20% of FL progress to lymphomas of higher histological grade and with a poor prognosis (Bastion et al, 1991; Martin et al, 1995; Harris et al, 1999). Although numerous putative prognostic factors have been extensively investigated, such as age, stage B-symptoms, extranodal sites, lactate dehydrogenase (LDH) level, tumour bulk and cytology (Bastion et al, 1991; Romaguera et al, 1991; Hermans et al, 1995), their respective prognostic values for treatment options still remain controversial. In this respect, a bone marrow biopsy (BMB) is routinely performed as part of the initial staging process. Lymphomatous bone marrow (BM) involvement is found in 40–70% of cases (McKenna et al,
ª 2004 Blackwell Publishing Ltd, British Journal of Haematology, 126, 364–371
Histological Patterns of BM Involvement Predict Survival in FL 1975, 1992; Foucar et al, 1982) at diagnosis and has been considered by some, but not all authors as a poor prognostic factor (Romaguera et al, 1991; Hermans et al, 1995; Izumi et al, 1996). The objectives of this histological study were, first, to evaluate, by means of a semi-quantitative method, the growth patterns and the severity of BM involvement in FL; secondly, to determine a simple descriptive grading system and to assess its reproducibility, and thirdly, to determine the influence of the different patterns on patients outcome after therapy. With this aim, the initial BMB from patients with FL included in the Groupe d’Etude des Lymphomes Folliculaires (GELF) 86 prospective multicentre trial, conducted by this cooperative group in France and Belgium, were re-examined.
Table I. Grid of histological findings applied to the bone marrow biopsy of follicular lymphoma patients. Parameters Size of BMB (excluding cortical bone) Number of marrow spaces
Marrow FL infiltration
Growth pattern of FL
Materials and methods Eligibility criteria Eligibility criteria for inclusion in this study were: (i) patients enrolled in the GELF protocol between 1986 and 1995 (GELF 86) (Lepage et al, 1990; Solal-Ce´ligny et al, 1993, 1998; Brice et al, 1997); (ii) patient BMB available for review.
The GELF 86 protocol This GELF protocol was an open, phase 3 trial including 550 FL patients enrolled from 40 centres. All previously untreated patients were eligible for the study if they had grade 1 disease according to the WHO classification [15 Cbs per hpf) was not included in the GELF protocol (Lepage et al, 1990; SolalCe´ligny et al, 1993, 1998; Brice et al, 1997). In all these cases, the diagnosis of FL was confirmed by a panel review. Patients under 70 years of age and classified as having Ann Arbor stages II, III or IV disease were eligible. At study entry, the tumour burden was evaluated and patients were classified into two groups according to the absence (group 1) or presence (group 2) of any of the following criteria: nodal or extranodal tumour mass with a diameter >7 cm; involvement of three or more nodal sites, each with a diameter >3 cm; systemic B-symptoms; marked splenomegaly; serous effusion; local risk of compression (epidural, ureteral, etc.) and leukaemia or cytopenia.
Histological analysis of BMB The first histological examination of BMB for all patients included in the GELF protocol was performed at their original medical centre. Most BMB were embedded in paraffin with only 7% embedded in epoxy resin. Different slides were stained with haematoxylin–eosin–safran, Giemsa and Gordon–Sweet stains. For the semi-quantitative study, all the BMB were reviewed and graded by the same observer according to a
normal Marrow reticulin fibres In LFo Normal; >normal In myeloid cells Normal; >normal *The area of the lymphomatous focus (LFo) was measured with a reticle (objective ·4) and the ratio of the area of an LFo to that of the BMB excluding cortical bone and trabecular bone was calculated. PT, paratrabecular.
highly specific grid of 10 items including the different patterns of FL involvement, the size of lymphomatous cells, the number of lymphomatous foci (LFo), the morphology (quantity, maturity) of myeloid cells and the presence or absence of myelofibrosis (Table I). BM involvement was quantified at low magnification (objective ·4), with a semi-quantitative method using a reticle that measured the ratio of the area of the LFo to that of the BMB size excluding cortical bone and trabecular bone (LFo/BMB). LFo could be measured in most BMB (95%). For the remaining cases, lymphomatous involvement could not be measured either because of technical problems or of the predominance of interstitial involvement, making this semi-quantitative method unreliable. In most biopsies, LFo areas did not differ widely between different sections of the same BMB. In rare cases (3Æ6%), the lymphomatous involvement was only observed on some BMB sections; in these cases
ª 2004 Blackwell Publishing Ltd, British Journal of Haematology, 126, 364–371
365
D. Canioni et al it was always mild (LFo/BMB 60 years, B-symptoms, >1 extranodal site, LDH level, compressive syndrome, lymphoma cell type, tumour bulk and cytopenia (Cox, 1972). Terms for the interaction of each of these covariates with treatment were also included in order to assess whether the treatment effect was consistent over a range of different covariate values. Estimates of the treatment effect are expressed as hazard ratios (HR), 95% confidence 366
intervals (CI) and associated P-values. The nominal significance level for the end points was 5% (two-sided test).
Results Inclusion of patients The clinical data and follow-up were available for 390 of the 406 BMB (96%) of patients included in the GELF 86 protocol that were reviewed semi-quantitatively for this study. These 390 patients constitute our study group and, according to the previously described parameters, they were assigned to the low (group 1, n ¼ 129) or high (group 2, n ¼ 261) tumourburden groups. Their initial clinical characteristics are summarized in Table II.
BMB histological characteristics The FL involvement was found in 267 of the 390 BMB reviewed (68%) (Fig 1; Table III). Among the 123 patients without BM involvement, 12% presented with one clearly demarcated perivascular lymphoid nodule, composed of small round lymphoid cells mixed with plasma cells and histiocytes, which was diagnosed as a ‘non-specific’ lymphoid nodule. The size of the BMB (excluding cortical bone) was usually large enough to be reliable for study because, in 271 cases (69Æ5%), it was >1 cm. Intramedullary spaces were also sufficiently numerous to allow good histological study (‡10 in 286 cases, 73%). Lymphomatous involvement was found in
Table II. Clinical characteristics of patients with follicular lymphoma and response to treatment (n ¼ 390). Characteristics Age (years) Median Range Sex Male Female Performance status 1 Elevated serum LDH Serum albumin (1 Area of LFo or LFo / BMB ratio (n ¼ 259) Normal Normal Normal Plasma cells (n ¼ 369) Normal >Normal
267 (68) 123 (32)
*Not available for all patients. LFo, lymphomatous foci.
122 (50) 122 (50) 53 (20) 124 (48) 82 (32) 242 (91) 22 (8) 3 (1) 112 (30) 196 (52) 69 (18) 360 (99) 3 (1) 267 (72) 103 (28) 349 (95) 20 (5)
foci in 244 of the 267 involved BMB (91%). This focal pattern consisted of one isolated, well-delineated focus within each medullary space in 50% of the cases, two foci in 28%, three foci in 19% and four foci in 3%. When there was >1 focus, the LFo pattern was not the same for all the patients [e.g. two paratrabecular (PT) or one PT associated with one nodular non-PT]. Isolated PT involvement was a growth pattern frequently observed in FL (36%), whereas an isolated nodular non-PT pattern was observed in only 4% of involved BMB and a diffuse pattern in 3%. Nevertheless, a combination of several different growth patterns of involvement was the most frequent feature encountered in our series (56%) and, among them, the association of PT and nodular non-PT patterns was the most common (41%). This combination of PT and nodular non-PT patterns was sometimes associated with one or several other growth patterns (interstitial or diffuse). The ratios of the LFo to BMB areas were separated into three levels following the grading criteria given in Table I. In most cases, BMB involvement lymphoma infiltrate consisted of small cells with angular ‘cleaved’ nuclei (91%); in rare cases, a mixture of small and large cells (8%) or only large cells (1%) were seen. In three cases, clusters of epithelioid histiocytes, mimicking a granuloma, were seen in the BMB. The involvement patterns, similar in groups 1 and 2, were tested as prognostic factors comparing one pattern of involvement (mostly isolated PT) with two or more. Interstitial and perisinusal involvements were not observed as a single pattern. Analysis of reticulin staining showed a slight fibrosis in LFo and no fibrosis in the myeloid lineage. The other main histological features of FL patients’ BMB included in this study are summarized in Table III.
ª 2004 Blackwell Publishing Ltd, British Journal of Haematology, 126, 364–371
367
D. Canioni et al Table IV. Comparative description of the two methods used for the evaluation of bone marrow biopsy involvement.
Grade
Semi-quantitative method
1 2
n < 0Æ01 0Æ01 £ n < 0Æ1
3
n ‡ 0Æ1
Simple descriptive grades £1 small nodule/MS One large nodule/MS or two small nodules/MS Three or four nodules/MS or ‡1 nodule + foci of diffuse or interstitial involvement
MS, medullary space.
The panel of the GELF 94 protocol, which reviewed 100 involved BMB following the new grades described in Table IV, found an interobserver reproductibility of 91%. Only one of the nine discordant cases produced a difference of more than one grading. The haematopathologist who twice reviewed 100 involved BMB of the GELF 86 found an intra-observer reproducibility between the ratio Lfo/BMB and the grades in 96% of cases. In the four cases for which grades were not in agreement with the ratio, the difference was never more than 1 grade. Finally, the pathologist who twice reviewed these same 100 involved BMB by using the new grading, found an interobserver reproducibility of 80% at the first review. This result improved at the second review, reaching 85%. There were 15 cases in which the haematopathologist and the pathologist disagreed regarding the grades, but the difference was never more than one grade. The intra-observer reproducibility was 82%, with a difference of two grades in only four cases.
Factors influencing survival Clinical factors. The beneficial effect of interferon-a on group 2 patients in terms of EFS and OS has been reported (SolalCe´ligny et al, 1993, 1998). However, for group 1 patients, no difference was found between the three treatment arms in terms of freedom from treatment failure and OS (Brice et al, 1997). The 8-year OS rate for the whole population was 54 ± 5% (median survival at 100 months). BM involvement versus normal BM did not influence OS (56 ± 6% versus 53 ± 4%, P ¼ 0Æ3). B-symptoms (P < 0Æ001) and >1 extranodal site (P < 0Æ0001) significantly affected OS. The presence of one criterion of high tumour burden, as defined above, was a major adverse prognostic criterion, with a 8-year OS rate of 70 ± 8% (median survival not reached) for group 1 and 50 ± 6% (median survival: 92 months) for group 2 (P ¼ 0Æ003). BM morphological factors. Analysis of the LFo/BMB ratio, the number of LFo and the different patterns of involvement demonstrated the significant influence on EFS of an LFo/BMB ratio ‡0Æ1 [three or four nodules/medullary space (MS) or 1 368
nodule + foci of diffuse or interstitial involvement ¼ new grade 3] (P ¼ 0Æ03) as well as two or more different patterns of involvement in the same BMB (P ¼ 0Æ004). The histological characteristics of myeloid lineages or plasma cells did not contribute significantly in predicting the clinical evolution of FL. Considering the entire population, the presence of two or more different patterns of BM involvement in the same BMB was the only BMB histological parameter that adversely affected OS (8-year OS rate: 47 ± 8% versus 68 ± 6%) (P ¼ 0Æ001). No difference in OS was found when normal BMB (8-year OS rate: 56 ± 7%) and only one pattern of BM involvement in the same BMB (5-year OS rate: 53 ± 6%) were considered. The same analyses were conducted for the two tumourburden subgroups. The distributions of patients according to absence of BM involvement, one pattern of involvement and ‡2 patterns of involvement were similar in both groups (39%, 27% and 34% for group 1, and 28%, 31% and 41% for group 2, respectively). Univariate analysis demonstrated the absence of influence on OS of two or more patterns of involvement in the same BMB in group 1 (low tumour burden) (Fig 2A), while this parameter significantly influenced survival in group 2 (Fig 2B). In this latter subgroup, introduction of this parameter into a multivariate analysis with other prognostic factors for FL identified two independent prognostic factors for survival: >1 extranodal site [HR: 1Æ8 (95%CI, 1Æ4–2Æ2); P ¼ 0Æ007] and ‡2 patterns of involvement in the same BMB [HR: 1Æ5 (95%CI, 1Æ1–1Æ9); P ¼ 0Æ04].
Discussion Advanced stage FL, although considered to be low-grade B-cell lymphomas, are usually not curable but have a long survival. No consensus has been reached on prognostic factors and optimal treatment strategy (Kaplan & Meier, 1958; Cox, 1972; Romaguera et al, 1991). A BMB is routinely performed to stage patients with FL and the frequency of BM involvement is higher in low-grade lymphomas (40–70%) than in intermediate- and high-grade lymphomas (30–35%) (McKenna et al, 1975, 1992; Foucar et al, 1982). The prognostic value of BMB histological findings in FL is controversial. In some reports, the BM status was an important parameter for initial staging and in the follow-up of patients who have advanced disease, to evaluate their responses to therapy (Foucar et al, 1982; Pond et al, 1989; Buhr et al, 2002). In another low grade B-cell lymphoma, chronic lymphocytic leukaemia, the BM pattern was considered to be predictive of the clinical evolution by the French Cooperative Group on Chronic Lymphocytic Leukemia (1990). Nonetheless, Izumi et al (1996) reported only a slight correlation between BM involvement and both duration of remission and OS. Other reports did not consider the BMB findings as prognostic elements for patients with advanced clinical stages of FL and thus do not contribute to therapeutic decisions (Bennett et al, 1986; McKenna, 1992). Further
ª 2004 Blackwell Publishing Ltd, British Journal of Haematology, 126, 364–371
Histological Patterns of BM Involvement Predict Survival in FL
Fig 2. Overall survival according to the histological pattern of BMB involvement in (A) low tumour burden patients (group 1), P ¼ 0Æ7, and (B) high tumour burden patients (group 2), P ¼ 0Æ0001. (— —) no involvement [n ¼ 50 (A) and 73 (B)]; (••) 1 pattern of involvement [n ¼ 35 (A) and 82 (B)]; (ÆÆ) ‡2 patterns of involvement [n ¼ 44 (A) and 106 (B)].
assessment of the potential value of BMB examination thus appears necessary. We reviewed 390 initial BMB from FL patients included in the GELF 86 protocol and analysed the histological patterns of BMB, using a semi-quantitative method, to evaluate the impact of these patterns and the severity of BM involvement on patient outcome. We demonstrated that the pattern of BM lymphomatous involvement is an important, statistically significant and independent predictive factor. In contrast to reports that focal paratrabecular (PT) involvement is the most common involvement pattern in FL (Dick et al, 1974; McKenna et al, 1975; McKenna, 1992; Hermans et al, 1995), we found more commonly a combination of several patterns in our large series, similar to Torlakovic et al (2002). PT and nodular non-PT patterns were observed either alone or associated with one or several other patterns. In most of our patients, BM involvement consisted of small-sized cells. The GELF 86 trial was divided into two therapeutic groups as a function of the tumour burden. This stratification was subsequently justified by the difference in terms of OS when high and low burdens were evaluated (median survival of
8 years and >10 years, respectively). Two parameters were found to be significantly associated with a shorter EFS: an LFo/ BMB ratio of ‡0Æ1, i.e. three or four nodules/MS or ‡1 nodule + foci of diffuse or interstitial involvement and two (or more) different growth patterns in the same BMB. OS was significantly shortened only in the high tumour burden (group 2) when two (or more) different types of lymphomatous involvement were present in the same BMB. Thus, it can be concluded that in patients with low tumour burden, BMB has little effect on other clinical parameters. In patients with a high tumour burden, the BMB findings, e.g. the pattern of involvement, has a major impact regardless of other parameters. In the only two other large studies addressing BM involvement in FL patients (275 cases and 52 cases), the authors did not assess the extent and the pattern of the involvement and did not define the tumour burden (Bennett et al, 1986; Pond et al, 1989). This could explain why Bennett et al (1986), contrary to our significant findings on survival, concluded that in low-grade lymphomas, the OS of patients with involved BMB was not worse than that for stage III patients. The other BMB histological parameters evaluated, such as the quantity and the maturity of myeloid lineages, the
ª 2004 Blackwell Publishing Ltd, British Journal of Haematology, 126, 364–371
369
D. Canioni et al quantities of eosinophils and plasma cells, and the presence of myelofibrosis, did not contribute significantly to predict the response to therapy and OS of FL patients. The second main point of this study demonstrated that BMB lymphomatous involvement could be evaluated by a simple descriptive grading that would be reliable for any surgical pathologist. This simplified histological grading of BM involvement was reproducible with several independent haematopathologists and by a trained pathologist. In conclusion, this semi-quantitative morphological study of 390 initial BMB of FL patients demonstrated that BMB histological patterns in high-tumour burden FL could predict EFS and OS. The presence or absence of BM involvement was not sufficient criteria to correlate with the prognosis. The histological pattern of BM involvement, such as an LFo/BMB ratio (‡0Æ1 or three or four nodules/MS or ‡1 nodule + foci of diffuse or interstitial involvement, and two or more different growth patterns in the same BMB) is one of the criteria that should be used in the choice of initial treatment of patients with high tumour-burden FL. Our study is the first attempt to evaluate the BM in a such wide series of patients treated following the same protocols; this series is also exceptional because of the long duration of follow-up.
Acknowledgments We would like to thank Pr James O. Armitage from the University of Nebraska Medical Center for reading our paper and his advice. We would like to acknowledge Fre´de´ric Charlotte, Bettina Fabiani and Eric Labouyrie for their help in establishing the new simple grading system. We are indebted to Catherine Balmale and Nicolas Riou for their help with data entry and analysis; Saadia Houga and Martine Grimal for secretarial help, Gise`le Le Gall and Martine Lanniel for technical assistance and to the following GELF members for contributing cases to the study: Drs M.F. d’Agay, C. Allard, R. Angonin, J. d’Anjou, B. Audhuy, B. Aufauvre, J.C. Barats, E. Baumelou, J. Benevent, F. Berger, A.M. Bertrand, M. Blanc, A.C. Baglin, R. Bordahandy, D. Bordessoule, A. Bosly, J. Bosq, R. Bouabdallah, J. Brie`re, J.C. Brouet, P.A. Bryon, J.P. Carbillet, S. Castaigne, T. Caulet, A.M. Chesneau, B. Christian, J.P. Clauvel, B. Coiffier, F. Cosnard, M. Delos, T. De Revel, M. Divine´, C. Doyen, P. Dubigeon, H. Duplay, G. Dupont, E. Dupuy, C. Duval, B. Epardeau, J.P. Fermand, A. Ferrant, M. Ffrench, N. Froment, J. Gabarre, F. Gaillard, F. Galateau, C. Garnier, P. Gaulard, C. Gisselbrecht, J. Hamels, J.L. Harousseau, M. Hayat, A. Herrera, C. Hopfner, N. Horschowski, R. Jeandel, Y. Kerneis, J.F. Knopf, E. Labouyrie, G. Laurent, C. Lavignac, B. Leger-Ravet, F. Lejeune, G. Lepeu, M. Leporrier, J.C. Mandard, G. Marit, J.P. Marolleau, C. Marty-Double, G. Merignargues, J.L. Michaux, J.M. Micle´a, N. Milpied, P. Mineur, M. Moriceau, H. Noe¨l, E. Oksenhendler, P. Oriol, N.J.P. Padilla, M. Patey, A. Petit, B. Pignon, G. Pinon-Netter, J. Quillard, J.F. Rame´e, M. Raphae¨l, M.C. Raymond-Gelle, 370
O. Reman, M. Renoux, F. Reyes, R. Riou, J.F. Rossi, F. Rumilly, G. Salles, A. Schmitz, H. Schvartz, J.F. Schwed, C. Sebban, J. Simony-Lafontaine, B. Taine, G. Tertian, A. Thyss, A.M. Touchais, P. Travade, and J. Troncy.
References Bastion, Y., Berger, F., Bryon, P.A., Felman P., French M. & Coiffier B. (1991) Follicular lymphomas: assessment of prognostic factors in 127 patients followed for 10 years. Annals of Oncology, 2(Suppl. 2), 123–129. Bennett, J.M., Cain, K.C., Glick, J.H., Johnson G.J., Ezdinli E. & O’Connell M.J. (1986) The significance of bone marrow involvement in non-Hodgkin’s lymphoma: the Eastern Cooperative Oncology Group Experience. Journal of Clinical Oncology, 4, 1462– 1469. Brice, P., Bastion, Y., Lepage, E., Brousse N., Haioun C., Moreau P., Straetmans N., Tilly H., Tabah T. & Solal-Ce´ligny P. (1997) Comparison in low-tumor-burden follicular lymphoma between an initial no-treatment policy, prednimustine or interferon-alpha: a randomized study from the Groupe d’Etude des Lymphomes Folliculaires. Journal of Clinical Oncology, 15, 1110–1117. Buhr, T., Langer, F., Schlue, J., von Wasielewski, R., Lehmann, U., Braumann, D. & Kreipe, H. (2002) Reliability of lymphoma classification in bone marrow trephines. British Journal of Haematology, 118, 470–476. Chan, J.K., Banks, P.M., Cleary, M.L., Delsol G., de Wolf-Peeters C., Falini B., Gatter K.C., Grogan T.M., Harris N.L., Isaacson P.G. Jaffe E.S., Knowles D.M., Mason D.Y., Muller-Hermelink H.K., Pileri S.A., Piris M.A., Ralfkiaer E., Stein H. & Warnke R.A. (1994) A proposal for classification of lymphoid neoplasms (by the International Lymphoma Study Group). Histopathology, 25, 517–536. Cox, D.R. (1972) Regression models and life tables. Journal of the Royal Statistical Society (B), 34, 187–202. Dick, F., Bloomfield, C.D. & Brunning R. D. (1974) Incidence, cytology and histopathology of non-Hodgkin’s lymphomas in the bonemarrow. Cancer, 33, 1382–1398. Foucar, K., Mc Kenna, R.W., Frizzera, G., & Brunning, R.D. (1982) Bone marrow and blood involvement by lymphoma in relationship to the Lukes-Collins classification. Cancer, 49, 888–897. French Cooperative Group on Chronic Lymphocytic Leukemia (1990) Natural history of Stade. A chronic lymphocytic leukemia untreated patients. British Journal of Haematology, 76, 45–47. Harris, N.L., Jaffe, E.S., Stein, H., Banks P.M., Chan J.K., Cleary M.L., Delsol G., de Wolf-Peeters C., Falini B. & Gatter K.C. (1994) A revised European-American classification of lymphoid neoplasms: a proposal from the International Lymphoma Study Group. Blood, 84, 1361–1392. Harris, N.L., Jaffe, E.S., Diebold J., Flandrin G., Muller-Hermelink H.K., Vardiman J., Lister T.A. & Bloomfield C.D. (1999) World Health Organization Classification of Neoplastic Diseases of the Hematopoietic and Lymphoid Tissues. Report of the Clinical Advisory Committee Meeting – Airlie House, Virginia, November 1997. Journal of Clinical Oncology, 17, 3835–3849. Hermans, J., Krol, A.D., van Groningen, K., Kluin P.M., KluinNelemans J.C., Kramer M.H., Noordijk E.M., Ong F. & Wijermans P.W. (1995) International prognostic index for aggressive nonHodgkin’s lymphoma is valid for all malignancy grades. Blood, 86, 1460–1463.
ª 2004 Blackwell Publishing Ltd, British Journal of Haematology, 126, 364–371
Histological Patterns of BM Involvement Predict Survival in FL Izumi, T., Imagawa, S., Muroi, K., Shimizu R., Takahashi H., Hoshino Y., Tomizuka H., Kawakami T., Toya K., Kurata H., Tsunoda J., Furukawa J., Suzuki T., Komatsu N., Ohsaka A., Amemiya Y., Takagi S., Yoshida M., Sasaki R., Hatake K. & Miura Y. (1996) Retrospective analysis on 21 patients with follicular lymphoma. Rinsho Ketsueki, 37, 7–13. Kaplan, E.S., Meier, P. (1958) Nonparametric estimation from incomplete observations. Journal of American Statistic Association, 53, 457–481. Lepage, E., Sebban, C. & Gisselbrecht, C. (1990) Treatment of lowgrade non Hodgkin’s lymphoma: a randomized prospective study with an assessment of doxorubicin in a controlled trial. Hematology Oncology, 8, 31–39. Martin, A.R., Weisenburger, D.D., Chan, W.C., Ruby E.I., Anderson J.R., Vose J.M., Bierman P.J., Bast M.A., Daley D.T. & Armitage J.O. (1995) Prognostic value of cellular proliferation and histologic grade in follicular lymphoma. Blood, 85, 3671–3678. McKenna, R.W. (1992) The bone marrow manifestations of Hodgkin’s disease, the non-Hodgkin lymphomas, and lymphoma-like disorders. In: Neoplastic Hematopathology (ed. by Knowles DM), pp. 1135–1180. Baltimore, MD. McKenna, R.W., Bloomfield, C.D. & Brunning, R.D. (1975) Nodular lymphoma: bone marrow and blood manifestations. Cancer, 36, 428–440. Non-Hodgkin’s Lymphoma Pathologic Classification Project (1982) National Cancer Institute sponsored study of classification of
non-Hodgkin’s lymphomas: summary and description of a working formulation for clinical usage. Non-Hodgkin’s Lymphoma Pathologic Classification Project. Cancer, 49, 2112–2135. Pond, G.D., Castellino, R.A., Horning, S. & Hoppe R.T. (1989) NonHodgkin’s lymphoma: influence of lymphography, CT, and bonemarrow biopsy on staging and management. Radiology, 170, 159– 164. Romaguera, J.E., McLaughlin, P. & North, L. (1991) Multivariate analysis of prognostic factors in stage IV follicular low-grade lymphoma: a risk model. Journal of Clinical Oncology, 9, 762–769. Solal-Ce´ligny, P., Lepage, E., Brousse, N., Reyes, F., Haioun C., Leporrier M., Peuchmaur M., Bosly A., Parlier Y., Brice P., Coiffier B. & Gisselbrecht C. for The Groupe d’Etude des Lymphomes de l’Adulte (1993) Recombinant interferon alpha-2b combined with a regimen containing doxorubicin in patients with advanced follicular lymphoma. The New England Journal of Medicine, 329, 1608–1614. Solal-Ce´ligny, P., Lepage, E., Brousse, N., Tendler, Cl., Brice, P., Haioun, C., Gabarre, J., Pignon, B., Tertian, G., Bouabdallah, Rossi, J.F., Doyen, C. & Coiffier, B. (1998) Doxorubicin-containing regimen with or without interferon alpha-2b for advanced follicular lymphomas: final analysis of survival and toxicity in the Groupe d’Etude des Lymphomes Folliculaires 86 trial. Journal of Clinical Oncology, 16, 2332–2338. Torlakovic, E., Torlakovic, G. & Brunning, R.D. (2002) Follicular pattern of bone marrow involvement by follicular lymphoma. American Journal of Clinical Pathology, 118, 780–786.
ª 2004 Blackwell Publishing Ltd, British Journal of Haematology, 126, 364–371
371