Original article. E. Maartense 1 *, S. le Cessie 2, H. C. Kluin-Nelemans 3, P. M. Kluin 4, S. Snijder 5, P. W. Wijermans 6 & E. M.

Transcription

1 Original article Annals of Oncology 13: , 2002 DOI: /annonc/mdf198 Age-related differences among patients with follicular lymphoma and the importance of prognostic scoring systems: analysis from a population-based non-hodgkin s lymphoma registry E. Maartense 1 *, S. le Cessie 2, H. C. Kluin-Nelemans 3, P. M. Kluin 4, S. Snijder 5, P. W. Wijermans 6 & E. M. Noordijk 7 1 Department of Internal Medicine, Reinier de Graaf Gasthuis; 2 Department of Medical Statistics, Leiden University Medical Center, Leiden; Departments of 3 Haematology and 4 Pathology, Academic Hospital Groningen, Groningen; 5 Comprehensive Cancer Center West, Leiden; 6 Department of Haematology, Leyenburg Hospital, The Hague; 7 Department of Clinical Oncology, Leiden University Medical Center, Leiden, The Netherlands Received 2 January 2002; accepted 14 January 2002 Background: The influence of age on the outcome of follicular non-hodgkin s lymphoma (FL) was studied in a population-based non-hodgkin s lymphoma registry. Patients and methods: This study comprised 214 follicular lymphoma patients. Grade I/II was considered separately from grade III FL. The data were analyzed with respect to three age groups: <60, and 70 years. Results: The overall survival rate decreased in the older age groups. Grade III patients showed a statistically significant decrease in overall survival in comparison with grade I/II patients (P = 0.03). Cause-specific survival analysis showed that in the older age groups, there was an increasing influence of concomitant disease on the death rate, especially among grade III FL patients >70 years of age. The survival curve in grade III FL patients was shown to reach a plateau. The prognostic scoring system, according to the Italian Lymphoma Intergroup, fitted better to grade I/II patients, while the International Prognostic Index showed better discrimination amongst grade III patients. Conclusions: Separate grading for follicular lymphoma is useful. An age >70 years has a negative impact on outcome, but the contribution of concomitant disease herein is important. Different prognostic scoring systems should be applied to the different grades of FL. Key words: cause-specific survival, elderly patients, follicular lymphoma, prognostic score Introduction Over the past 30 years, the treatment of follicular non- Hodgkin s lymphoma (FL) has not resulted in a major breakthrough. Whereas FL patients as a whole experience a median survival of 7 10 years, several subgroups can be discerned, each predicting a better or worse survival. Among these, the age of the patient and pathology grading of the FL are probably the most important predictive factors. With regard to the latter, the recognition of three histological grades, based upon the number of centroblasts, is common practice, although there is no consensus on the method of grading. As a compromise, the World Health Organization (WHO) Classification suggested the use of at least two grades by proposing the *Correspondence to: Dr E. Maartense, Department of Internal Medicine, Reinier de Graaf Gasthuis, Reinier de Graafweg 3-11, 2625 AD Delft, The Netherlands. Tel: ; Fax: ; Emaarten@knmg.nl grouping of grades 1 and 2 [1, 2]. In this classification, grade 1 and 2 FL behave similarly in natural history and treatment response, thereby differing from grade 3. However, the Non- Hodgkin s Lymphoma Classification Project (NHLCP) did not distinguish between the three grades in its reports on treatment results of the non-hodgkin s lymphomas [3 5], as was done in the development of a prognostic model for this lymphoma [6, 7]. Others reported separately on grade 1 and 2 FL with respect to prognostic factors and treatment results [8, 9], and noticed the similarity between grade 3 FL and diffuse large B-cell lymphoma with regard to outcome [10 12]. Age-related differences in the outcome of non-hodgkin s lymphomas (NHL) are well known [4, 13]. In a previous analysis on patients in a population-based NHL registry the denominator elderly was given to patients >70 and >65 years of age, for indolent and aggressive NHL, respectively [14]. To determine more precisely the fate of patients with FL, and the impact of grade in an unselected group of patients, we 2002 European Society for Medical Oncology

2 1276 analyzed the data with respect to the age groups <60, and 70 years from the Comprehensive Cancer Center West (CCCW) NHL registry. In addition, we paid special attention to the relative and cause-specific survival (CSS) and to the prognostic scoring systems according to the International Prognostic Index (IPI) [15] and Italian Lymphoma Intergroup (ILI) [7]. Patients and methods A thorough analysis according to age in this population-based NHL registry has been reported previously [13, 14]. In summary, all new cases with NHL in the region covered by the CCCW (The Netherlands; covering 1.6 million inhabitants) were prospectively registered from June 1981 to December After exclusion of patients with primary cutaneous T-cell lymphoma, classical chronic lymphocytic leukemia, acute lymphoblastic leukemia, multiple myeloma, plasmocytoma or a post-mortem diagnosis only of NHL, this amounted to 1168 patients. The histological diagnosis had to be confirmed by a panel of hematopathologists. The subclassification of FL was performed, in accordance with the NHL Pathologic Classification Project [16], into the following categories: predominantly small, mixed small and large, and predominantly large cell. According to this system, in the CCCW registry the cut-off points were <20%, 20 50% and >50% centroblasts, respectively. The REAL classification recommended the terminology grade 1, grade 2 and grade 3, but did not recommend any specific criteria [17]. However, the WHO recommendation for the grading of FL is based on the exact counting of the number of centroblasts in 10 neoplasic follicles, expressed per 40 high-power microscopic field [2]. Because this was not performed in the CCCW registry and to avoid any confusion we decided to use the terminology grade I/II FL and grade III FL for FL with <50% and >50% centroblasts, respectively. Staging was based on the Ann Arbor system [18]. Among grade I/II patients the three age groups showed adequate staging for 86, 87 and 65% of patients, respectively. Among grade III patients this was true for 89, 80 and 36% of patients, respectively. Inadequate staging was due mainly to the omission of a bone marrow biopsy. Although consensus on treatment was reached within the NHL Study Group of CCCW, the choice of treatment was the responsibility of the local physician. Until 1989, the general guidelines for FL were radiotherapy for localized lymphomas, cyclophosphamide, vincristine, prednisone (CVP) chemotherapy or chlorambucil for stage III IV grade I/II FL, cyclophosphamide, doxorubicin, vincristine, prednisone (CHOP) chemotherapy for stage II IV grade III FL, followed by radiotherapy for bulky tumors or persistent lesions. Treatment outcome was related to complete remission rate, overall survival (OS), relative survival and CSS. Overall survival was calculated from diagnosis until death (all causes, event) or last follow-up (censored; July 2000). The OS per age group was compared to the expected survival for a cohort of the general Dutch population (data obtained from Statistics Netherlands) with the same age and sex distribution. Relative survival probabilities, defined as observed survival probability divided by the expected survival probability, were calculated using Hakulinen s relative survival software [19]. The causes of death were divided among NHL, toxicity of treatment, other malignancies, concomitant diseases and unknown. Cause-specific survival was determined for the four known categories and was censored by death not due to NHL or treatment-related toxicity. As prognostic factors for survival, we considered, along with age, the number of extranodal sites, serum lactate dehydrogenase (LDH), performance status (Karnofsky index) and stage to determine the IPI [15], as well as gender and erythrocyte sedimentation rate (ESR). Because an ageadjusted ILI has not been validated, we omitted an age-adjusted analysis of the IPI. The prognostic model according to ILI identified age, gender, number of extranodal sites, B-symptoms, serum LDH and ESR as predictive factors for OS [7]. However, no cases presented with all six adverse factors and therefore three risk groups determined by five adverse factors were considered. Although information about B-symptoms was lacking in the CCCW database, the application of the ILI model, using the remaining five variables, was considered appropriate. Like ILI we defined three risk groups in both scoring systems: low (score 0 1), intermediate (score 2) and high (score 3 5) [7]. Statistical analyses were performed using chi-square tests for trends to compare percentages in cross tabulations, and the log-rank test to compare Kaplan Meier survival curves. The prognostic factors were analyzed by Cox proportional hazards regression model. Results The clinical characteristics of 214 patients with FL according to the age groups <60, and 70 years are given in Table 1 for patients with grade I/II FL (n = 171) and in Table 2 for patients with grade III FL (n = 43). Testing for trend between the three age groups in patients with grade I/II FL only showed a statistically significant difference for elevated ESR (P = 0.001), showing more often elevated values in the older age groups. With respect to the prognostic scoring systems, both IPI and ILI showed a delineation between the age groups: a low score (0 1) especially among patients <60 years of age, while a high score (3 5) was most frequently seen among patients >70 years of age. Forty-four percent of all grade I/II FL patients were in the IPI low risk group, 32% in the intermediate risk group and 24% in the high risk group, while for ILI the distribution was 52, 29 and 19%, respectively. Among patients with grade III FL, most patients <60 years of age were in the low risk group, both in the IPI and ILI risk models. Fifty-three percent of all grade III FL patients were in the IPI low risk group, 35% in the intermediate risk group and 12% in the high risk group, while for ILI, the distribution was 49, 32 and 19%, respectively. In the younger age group (<60 years of age) LDH showed significantly more often elevated values in comparison with FL grade I/II patients (P = 0.04). In addition, significantly fewer patients (0 7%) had extranodal localizations compared with the FL I/II group (P = 0.01; Tables 1 and 2). Finally, initial therapy with anthracyclins was given significantly less often to FL grade III patients as age increased (P = 0.006). Table 3 shows outcome of treatment among the three age groups. For the whole cohort of 214 patients the median follow-up was 110 months (range ). The complete remission rate decreased among patients >70 years of age, and this was statistically significant for grade I/II FL (P = 0.03 versus 0.43 for grade III FL). The difference in OS between the age groups was statistically significant among grade I/II

3 1277 Table 1. Clinical characteristics of patients with follicular grade I/II non-hodgkin s lymphoma according to age (n = 171) Characteristics Percentage of No. of patients with Total no. of patients with <60 years years 70 years characteristic known parameter (n = 84) (n = 38) (n = 49) Sex (male) Stage (III IV) ΚΠΣ LDH increased >1 extranodal site ESR 30 mm a IPI 141 Low (score 0 1) Intermediate (score 2) High (score 3 5) ILI 152 Low (score 0 1) Intermediate (score 2) High (score 3 5) Initial treatment 171 No treatment Surgery alone Radiotherapy alone Chemotherapy A Chemotherapy A a P = The other characteristics showed no statistically significant differences between the age groups. A+, anthracycline-containing chemotherapy; ESR, erythrocyte sedimentation rate; IPI, International Prognostic Index; ILI, Italian Lymphoma Intergroup; KPS, Karnofsky performance status; LDH, lactate dehydrogenase. Table 2. Clinical characteristics of patients with follicular grade III non-hodgkin s lymphoma according to age (n = 43) Characteristics Percentage of No. of patients with Total no. of patients with <60 years years 70 years characteristic known parameter (n = 18) (n = 10) (n = 15) Sex (male) Stage (III IV) ΚPS LDH increased >1 extranodal site ESR 30 mm IPI 34 Low (score 0 1) Intermediate (score 2) High (score 3 5) ILI 37 Low (score 0 1) Intermediate (score 2) High (score 3 5) Initial treatment a 43 No treatment Surgery alone Radiotherapy alone Chemotherapy A+ b Chemotherapy A a P = The other characteristics were not significantly different between the age groups. b P = A+, anthracycline-containing chemotherapy; ESR, erythrocyte sedimentation rate; IPI, International Prognostic Index; ILI, Italian Lymphoma Intergroup; KPS, Karnofsky performance status; LDH, lactate dehydrogenase.

4 1278 Table 3. Treatment outcome for follicular lymphomas according to age Treatment outcome a P = b P < c P = d P = SE, standard error. Age classes (years) < Follicular grade I/II (n) Treated initially (n) Complete remission a (%) Overall survival b 5-year, % (SE) 75 (5) 62 (8) 40 (7) 10-year, % (SE) 41 (6) 19 (7) 13 (5) Median survival in months (SE) 103 (10) 73 (7) 46 (4) Follicular grade III (n) Treated initially (n) Complete remission c (%) Overall survival d 5-year, % (SE) 44 (12) 30 (14) 27 (11) 10-year, % (SE) 19 (10) 30 (14) 20 (10) Median survival (months) (SE) 45 (17) 18 (13) 11 (3) patients (P <0.001 versus P = 0.22 for grade III FL). However, by Cox-regression analysis the hazard ratio for death increased in the older age groups, in both grade I/II or grade III FL patients. Among grade I/II patients, the hazard ratio was 1.9 [95% confidence interval (CI) ] and 2.7 (95% CI ), respectively, for age groups and >70 years. Among grade III patients, this was 1.3 (95% CI ) and 1.9 (95% CI ), respectively. Overall survival curves per age group, together with the curves for CSS and survival of the Dutch population at large, are shown in Figure 1A F. The relative survival curves, showing the same trends as the CSS curves, have been omitted for the sake of clarity. For the whole cohort, the difference in OS between grade I/II versus grade III patients was statistically significant (log-rank test P = 0.03). Considering CSS, the cause of death was known for 89% of patients. NHL and toxicity of treatment contributed to the cause of death in 80, 66 and 52% of patients with grade I/II FL in the three respective age groups. Among patients with grade III FL, the rates were 100, 88 and 58%, respectively. Considering grade I/II FL, the CSS was not much different from OS for patients <60 years of age only (Figure 1A). With respect to grade III FL, this held true for the two age groups <60 and years of age (Figure 1D E). In other words: the chance of death from an unrelated disease increases for patients >60 years of age with grade I/II FL and for patients >70 years of age with grade III FL. The prognostic values of the variables LDH, Karnofsky index, stage, gender, more than one extranodal site, ESR and age, determined by a Cox regression analysis, are shown in Table 4. For patients with grade I/II FL, especially serum LDH, Karnofsky score and an age 70 years appeared to be important for prognosis. For patients with grade III FL this held true for serum LDH and stage. Testing for interaction between follicular grade and the other risk factors in the multiple Cox regression model showed that the hazard ratios between grade I/II and grade III FL differed significantly for serum LDH (P = 0.02) and stage III IV (P = 0.001). The OS curves according to IPI and ILI for the low (0 1), intermediate (2) and high (3 5) risk scores are shown in Figure 2A D separately for grade I/II and grade III FL. The ILI risk score discriminated best for grade I/II (log-rank test for trend by ILI 34.1, df = 1, P <0.001; log-rank test for trend by IPI 22.07, df = 1, P <0.001), while the IPI risk score discriminated best for grade III (log-rank test by ILI 4.93, df = 1, P = 0.03; log-rank test by IPI 10.7, df = 1, P = 0.001). Table 5 shows the 5- and 10-year survival rates according to both risk models and separately for grade I/II and grade III FL. Discussion This study stresses the differences between grade I/II versus grade III FL and stipulates the importance of considering agerelated differences in outcome.

5 Figure 1. (A F) Overall survival and cause-specific survival in relation to the survival of the Dutch population at large for the three age groups, separately for grade I/II and grade III follicular non-hodgkin s lymphoma. The data for the expected survival of the Dutch population were obtained from Statistics Netherlands. Cause-specific survival was censored by death not due to non-hodgkin s lymphoma or treatment-related toxicity. 1279

6 1280 Table 4. Prognostic value of variables for follicular lymphoma grade I/II versus grade III in a multivariate Cox model Characteristics Grade I/II Grade III RR of death 95% CI P RR of death 95% CI P LDH > normal KPS Stage III IV >1 extranodal site Male sex ESR 30 mm ESR unknown Age (years) CI, confidence interval; ESR, erythrocyte sedimentation rate; KPS, Karnofsky performance status; LDH, lactate dehydrogenase. A comparison with respect to characteristics and treatment results of follicular lymphoma, reported in different series, is given in Table 6. There is little consistency in the proportion of the characteristics among patients in the different series, which it is assumed is related to the inclusion criteria used for the different treatment protocols. What was especially apparent was the decrease in performance among grade I/II patients in the current series compared with other grade I/II series [8, 9]. Furthermore, the distribution of age groups (more elderly patients) is different between population-based series (current CCCW and Nebraska Lymphoma Study Group [20]) as opposed to institutional or referral-based series. The finding that both IPI and ILI showed the low risk category to be patients <60 years of age and the high risk category those patients >70 years of age can be explained, in part, by the fact that both scoring systems consider age >60 years as a risk factor. In our series the probability of increased LDH among grade I/II patients was especially apparent for patients >60 years of age. With respect to elevated ESR, the CCCW registry showed a statistically significant difference among the three age groups of grade I/II patients. According to ILI an elevated ESR (besides the presence of B-symptoms) forms an important prognostic factor [7], thus probably attributing (as well as increased LDH) to the inferior prognosis among elderly patients in our series. When considering the initial treatment, it appeared that 40% of the patients in the CCCW registry with grade III FL received anthracycline-based chemotherapy, but this dropped to 13% among patients >70 years of age. In this way the difference between a population-based strategy and treatment decisions for patients in other series (institution protocols [10, 12, 20] and trial protocols [11]), where most patients were treated with anthracycline-based chemotherapy, is remarkable. Different complete response (CR) rates are reported in the literature depending on the techniques and diligence used to assess residual disease [21]. The CR rate per age group in our series was roughly the same, irrespective of grade. The chance to obtain complete remission was lower among patients >70 years of age (as was shown by the NHLCP [4]), and this was statistically significant for patients with grade I/II FL. However, the value of establishing complete remission in FL I/II patients, for whom deferred treatment in cases of low tumor burden does not jeopardize the survival rate [21, 22], can be argued. The OS rate was inferior among the older age groups, which is especially apparent from the Cox-regression analysis. The hazard ratios for age were not significantly different between grade I/II and grade III FL (P = 0.97). Among grade I/II patients, the age group <60 years showed a median survival of 103 months, which conforms to the Stanford series [21]. For the age groups and 70 years this dropped to 73 and 46 months, respectively. In their report on age-related differences in the behavior of NHL, the NHLCP showed decreased survival rate in patients >70 years of age (and <35 years of age!) with FL in all grades (29% of that cohort had grade III FL) [4]. The Nebraska group found significantly decreased survival among patients >60 years of age and with stage I/II FL (75% of these patients had grade III FL) [23]. More series confirmed age >60 years as an adverse prognostic factor [6, 9]. With respect to grade III FL, the median survival in our series appeared to be 45, 18 and 11 months for the three respective age groups. The Nebraska group found significantly decreased survival for patients >70 years of age and with stage III/IV disease only [20]. The Group d Etude des Lymphomes d Adulte (GELA) group, in their analysis of patients with grade III FL, found that age >60 years was a statistically significant adverse prognostic factor [11], but other series did not confirm this age cut-off [10, 24]. A plateau forming in the OS curves was especially apparent for grade III FL, foremost in the age group <60 years. Most series with

7 1281 Figure 2. (A D) Survival according to the International Prognostic Index (IPI) and Italian Lymphoma Intergroup (ILI) risk groups, separately for grade I/II and grade III follicular non-hodgkin s lymphoma (FL). Low risk means 0 1 risk factor, intermediate risk 2 factors and high risk 3 5 risk factors. The ILI risk score discriminated best for grade I/II FL (log-rank test for trend by ILI, 34.1, and by IPI, 22.07), while the IPI risk score fitted better for grade III FL (log-rank test for trend by IPI, 10.7, and by ILI, 4.93). Table 5. Survival according to IPI and ILI risk groups, separately for grade I/II and grade III follicular non-hodgkin s lymphoma Risk group Grade I/II Grade III IPI ILI IPI ILI Low (0 1) n = 62 n = 79 n = 18 n = 18 5-year, % (SE) 84 (5) 77 (5) 56 (12) 44 (12) 10-year, % (SE) 49 (7) 40 (6) 30 (12) 26 (11) Intermediate (2) n = 45 n = 44 n = 12 n = 12 5-year, % (SE) 49 (7) 57 (7) 17 (11) 33 (14) 10-year, % (SE) 8 (5) 23 (7) 17 (11) 25 (13) High (3 5) n = 34 n = 29 n = 4 n = 7 5-year, % (SE) 43 (9) 25 (8) 0 14 (13) 10-year, % (SE) 22 (8) (13) IPI, International Prognostic Index; ILI, Italian Lymphoma Intergroup.

8 Table 6. Comparison of characteristics and treatment results among different series of follicular lymphoma CCCW (current) NHLCP a ILI b MDA c,d GELA e,f Stanford g NLSG h MDA i Grade I/II Grade III All grades All grades Grade I/II 8 Grade III 12 Grade I/II 9 Grade III 11 Grade III Grade III Grade III No. of cases Female (%) Stage III/IV (%) KPS 70 (%) LDH (%) >1 EN site (%) ESR 30 mm (%) Age (%) <60 years years years IPI (%) Low Intermediate High ILI (%) Low Intermediate High CR (%) year OS (%) year OS (%) year CSS (%) year CSS (%) a NHLCP, consecutive cases of previously untreated follicular non-hodgkin s lymphoma from 1988 to 1990 from nine institutions in eight countries [5]. b ILI, patients from eight single centers and two cooperative groups, treated according to the rules of prospective clinical trials or to specified guidelines from 1985 to 1996 [7]. c MDA, previously untreated patients, only stage IV, follicular small-cleaved and mixed-cell, all treated with cyclophosphamide, doxorubicin, vincristine, prednisone plus bleomycin (CHOP-bleo) from 1972 to 1982 [8]. d MDA, treated according to protocols for intermediate grade lymphoma by the Working Formulation from 1984 to 1999 [12]. e GELA, patients from 40 centers from October 1986 to May 1995, stage III IV only and <70 years of age, treated according to the rules of two open-label phase III trials [9]. f GELA, 89 patients from 35 centers included in the LNH87 trial (selection for four protocols) from October 1987 to April 1993 [11]. g Stanford, retrospective analysis of patients treated from 1969 to 1991 according to institutional guidelines [10]. h NLSG, previously untreated patients were prospectively treated according to plans of the NLSG from 1980 to 1991 [20]. i MDA, consecutive previously untreated patients were treated according to institutional guidelines from 1973 to 1981 [24]. CCCW, Comprehensive Cancer Center West; CR, complete response; CSS, cause-specific survival; EN, extranodal site; ESR, erythrocyte sedimentation rate; GELA, Group d Etude des Lymphomes d Adulte; NHLCP, non-hodgkin s Lymphoma Classification Project; IPI, International Prognostic Index; ILI, Italian Lymphoma Intergroup; KPS, Karnofsky performance status; LDH, lactate dehydrogenase; MDA, M.D. Anderson Cancer Center; NSLG, Nebraska Lymphoma Study Group; OS, overall survival rate.

9 1283 prolonged follow-up recognize a 20 30% disease-free survival among grade III FL patients, thereby simulating the course of diffuse large B-cell lymphoma [10 12, 24]. However, the South West Oncology Group (SWOG) in their report on follicular lymphomas, only of stages III/IV, could not demonstrate any difference in OS for any subtype after follow-up until 25 years after treatment [25]. The inclusion of stage I/II patients in the OS curves for grade III FL can contribute to the formation of a plateau (in our series 49% of the patients with grade III FL had stage I/II disease). Data from the Nebraska group also supports this [20, 23]. We applied a special technique to correct for natural death in elderly patients by comparing the survival with time-period and age-matched Dutch inhabitants according to Hakulinen s relative survival program [19], as well as a CSS analysis by censoring for death not due to NHL or treatment related toxicity. Compared with OS the CSS (and relative survival) per age group appeared greater >60 years of age among grade I/II FL and >70 years of age among grade III FL patients. A superior CSS was also demonstrated by the MD Anderson series on grade III FL, especially for stage I/II patients [24]. In that series an age-dependent analysis of CSS was not performed. Also in the Nebraska series the difference in survival between patients <60 and >60 years of age with stage I/II FL (75% grade III) was no longer evident when looking at CSS data [23]; this has also been demonstrated for other types of aggressive NHL [26]. However, for grade I/II FL patients a CSS analysis was not different from OS in another MD Anderson series [8]. In that series age (>60 versus <60 years of age) was not significantly correlated with CSS. The prognostic scoring system of the IPI seems less optimal for follicular lymphoma [27], because of the low number of patients in the higher risks groups [10, 12, 24]. The large series of ILI identified age, gender, B-symptoms, number of extranodal sites, serum LDH and elevated ESR as key features for prognosis of FL (irrespective of grade) [7]. In our populationbased series, which lacked information on B-symptoms, we identified elevated LDH, Karnofsky performance status 70 and age >70 years as the main prognostic factors for grade I/II FL patients, whereas elevated LDH and stage III/IV disease were the main prognostic determinants among grade III FL patients. Of course, the low number of patients (especially with respect to grade III FL) contributes to the fact that statistical significance was not reached for the other prognostic factors. By comparing the OS curves according to IPI and ILI, we were able to confirm that the ILI scoring system resulted in a better prognostic differentiation for patients with grade I/II FL than the IPI system. The latter was better at discriminating between patients with grade III FL, thus supporting the concept that grade III FL should be treated as a separate entity to grades I and II, with its properties of aggressive NHLs. Several conclusions can be drawn from the analysis of this population-based cohort of patients with follicular non- Hodgkin s lymphoma. First, the achievement of complete remission was not different for grade I/II versus grade III, both showing decreased CR rate >70 years of age. Secondly, older age groups showed a decrease in the OS, whether grade I/II or grade III (Cox model). Thirdly, the median survival was lower for patients with grade III FL, irrespective of age group. However, the plateau observed in the survival curve, shown foremost by the CSS analysis, confirms the separate status of grade III FL being more akin to diffuse large B-cell lymphoma. Fourthly, the prognostic scoring system according to ILI fitted better to grade I/II FL patients, while the IPI system appeared more appropriate among grade III FL patients. The patients in this population-based cohort have been treated according to classical guidelines of the 1980s. The new treatment options of the 1990s will hopefully lead to a longsought after survival advantage for FL [27]. A thorough understanding of pre-existing risk factors and age-related differences should be helpful in the design of new studies and the interpretation of treatment data. Appendix The following hospitals, internist-oncologists and pathologists in the CCCW NHL registry participated in this study: Leiden University Medical Center (J.C. Kluin-Nelemans, E.M. Noordijk, J.H.J.M. van Krieken, P.M. Kluin); Antoniushove, Leidschendam (M.G. Herben); Bronovo, The Hague (R. Bieger, C. Tinga); Diaconessenhuis, Leiden (H. van Slooten, M. Gorsira); Groene Hart Ziekenhuis Gouda (K.J. Heering, J. Rahder, R.F.A. Simonis, A.M.E. van der Torren- Conze); t Lange Land Ziekenhuis, Zoetermeer (A. Folmer); Ziekenhuis Leyenburg, The Hague (W.B.J. Gerrits, H.L. Haak, P.W. Wijermans, K. van Groningen); Reinier de Graaf Groep, Delft-Voorburg (A. Houwing, C. van Krimpen, E. Maartense, J. de Regt, J.R. van der Mey); Rode Kruis Ziekenhuis, The Hague (W.A. van Deijk, P.J. Spaander); Rijnland Ziekenhuis, Leiderdorp (F.H.M. Cluitmans, W.J. Molendijk, P.C.M. Rosekrans, J. Calame, G.J.P.M. Jonkers); and Westeinde Ziekenhuis, The Hague (G. Booij, R. Vriesendorp, P. Blok). References 1. Harris NL, Jaffe ES, Diebold J et al. World Health Organization classification of neoplasic diseases of the hematopoietic and lymphoid tissues: report of the Clinical Advisory Committee Meeting Airlie House, Virginia, November J Clin Oncol 1999; 17: Nathwani BN, Harris NL, Weisenburger D et al. Follicular lymphoma. In Jaffe ES, Harris NL, Stein H, Vardiman JW (eds): Tumours of Haematopoietic and Lymphoid Tissues: Pathology & Genetics. WHO Classification of Tumours. Lyon: IARC Press 2001; The Non-Hodgkin s Lymphoma Classification Project. A clinical evaluation of the International Lymphoma Study Group classification of non-hodkin s lymphoma. Blood 1997; 89: The Non-Hodgkin s Lymphoma Classification Project. Effect of age on the characteristics and clinical behavior of non-hodgkin s lymphoma patients. Ann Oncol 1997; 8:

10 Armitage JO, Weisenburger DD. New approach to classifying non- Hodgkin s lymphomas: clinical features of the major histologic subtypes. J Clin Oncol 1998; 16: Cameron DA, Leonard RCF, Mao J, Prescott RJ. Identification of prognostic groups in follicular lymphoma. Leuk Lymphoma 1993; 10: Federico M, Vitolo U, Zinzani PL et al. Prognosis of follicular lymphoma: a predictive model based on a retrospective analysis of 987 cases. Blood 2000; 95: Romaguera JE, McLaughlin P, North L et al. Multivariate analysis of prognostic factors in stage IV follicular low-grade lymphoma: a risk model. J Clin Oncol 1991; 9: Decaudin D, Lepage E, Brousse N et al. Low-grade stage III IV follicular lymphoma: multivariate analysis of prognostic factors in 484 patients a study of the Group d Etude des Lymphomes de l Adulte. J Clin Oncol 1999; 17: Bartlett NL, Rizeq M, Dorfman RF et al. Follicular large-cell lymphoma: intermediate or low grade? J Clin Oncol 1994; 12: Wendum D, Sebban C, Gaulard Ph et al. Follicular large-cell lymphoma treated with intensive chemotherapy: an analysis of 89 cases included in the LNH87 trial and comparison with the outcome of diffuse large B-cell lymphoma. J Clin Oncol 1997; 15: Rodriguez J, McLaughlin P, Hagemeister FB et al. Follicular large cell lymphoma; an aggressive lymphoma that often presents with favorable prognostic features. Blood 1999; 93: Maartense E, Hermans J, Kluin-Nelemans JC et al. Elderly patients with non-hodgkin s lymphoma: population-based results in the Netherlands. Ann Oncol 1998; 9: Maartense E, Kluin-Nelemans HC, le Cessie S et al. Different age limits for elderly patients with indolent and aggressive non- Hodgkin s lymphoma and the role of relative survival with rising age; analysis of a population-based non-hodgkin s lymphoma registry. Cancer 2000; 89: International Non-Hodgkin s Lymphoma Prognostic Factors Project. A predictive model for aggressive non-hodgkin s lymphoma. N Engl J Med 1993; 329: Non-Hodgkin s Lymphoma Pathologic Classification Project. A National Cancer Institute sponsored study of classifications of non-hodgkin s lymphomas. Summary and description of a working formulation for clinical usage. Cancer 1982; 49: Harris NL, Jaffe ES, Stein H et al. A revised European American classification of lymphoid neoplasms: a proposal from the International Lymphoma Study Group. Blood 1994; 84: Carbone PP, Kaplan HS, Musshoff K et al. Report of the Committee on Hodgkin s Disease Staging Classification. Cancer Res 1971; 31: Hakulinen T, Abeywickrama KH. A computer program package for relative survival. Comp Prog Biomed 1985; 19: Anderson JR, Vose JM, Bierman PJ et al. Clinical features and prognosis of follicular large-cell lymphoma: a report from the Nebraska Lymphoma Study Group. J Clin Oncol 1993; 11: Horning SJ. Natural history of and therapy for the indolent non- Hodgkin s lymphomas. Semin Oncol 1993; 20 (Suppl 5), S75 S Brice P, Bastion Y, Lepage E et al. Comparison in low tumor burden follicular lymphomas between an initial no-treatment policy, prednimustine, or interferon α: a randomized study from the Groupe d Etude des Lymphomes Folliculaires. J Clin Oncol 1997; 15: Tezcan H, Vose JM, Bast M et al. Limited stage I and II follicular non-hodgkin s lymphoma: the Nebraska Lymphoma Study Group experience. Leuk Lymphoma 1999; 34: Rodriguez J, McLaughlin P, Fayad L et al. Follicular large cell lymphoma: long-term follow-up of 62 patients treated between Ann Oncol 2000; 11: Miller TP, LeBlanc M, Grogan TM, Fisher RI. Follicular lymphomas: do histologic subtypes predict outcome? Hematol Oncol Clin North Am 1997; 11: Vose JM, Armitage JO, Weisenburger DD et al. The importance of age in survival of patients treated with chemotherapy for aggressive non-hodgkin s lymphoma. J Clin Oncol 1988; 6: Horning SJ. Follicular lymphoma: have we made any progress? Ann Oncol 2000; 11 (Suppl 1): S23 S27.