Original article: Non-Hodgkin's lymphoma in the elderly

Annals of Oncology' 3: 379-386, 1992. O 1992 Kluwer Academic Publishers. Pnmcd in the Netherlands.

Original article Non-Hodgkin's lymphoma in the elderly A study of 602 patients aged 70 or older from a Danish population-based registry F. d'Amore,1 H. Brincker,1 B. E. Christensen,1 K. Thorling,2 M. Pedersen,3 J. Lanng Nielsen,4 E. Sandberg,5 N. T. Pedersen6 & E. Sorensen7 - for the Danish LYFO-study group 1 Dept. of Haematology, Odetise University Hospital, Odense; 2Dept. of Oncology, Aarhus Kommunehospital, Aarhits; 3Dept. of Oncology, Aalborg Hospital, Aalborg; 4 Dept. of Medicine, Holstebro Hospital, Holstebro; iDept. of Oncology, Elsbjerg Hospital, Esbjerg; ''Dept. of Pathology, Odense University Hospital, Odense; 1 Dept. of Medical Physics, Aarhus Kommunehospital, Aarhus, Denmark

Summary. Within a 7-year period 1,597 newly diagnosed cases of non-Hodgkin's lymphoma (NHL) were included in a Danish population-based NHL registry. Of these, 602 (38%) were aged 70 years or older (age range 70-94, median: 76.8). They represent the population defined as 'elderly' patients in the present study. The average annual incidence rate for this elderly patient population was 35.7/105, as compared with 6.6/105 for patients aged <70 (overall annual incidence: 9.5/ L05). Localised cases (stage I and II) and extranodal manifestations were found more frequently among elderly patients. The most common sites of extranodal involvement were the stomach (21% of all extranodal cases) and the bone marrow (16%). Histologjcally, follicular centroblastic/centrocytic cases were found to be less frequent (p < 0.01) in elderly patients as compared to their younger counterparts (<70

Introduction

years), who in contrast had a lower occurrence of diffuse centroblastic cases (p < 0.01). Overall 7-year survival for the elderly patient population was 35% (median 1.7 years), and for patients aged <70 it was 57%. This difference persisted after correction for apparently NHL-unrelated deaths (52% vs. 66%, respectively, p < 0.0001). Elderly patients with poor prognosis were characterised by the following features identified in a Cox-regression model: hepatic involvement, presence of B-symptoms, high-grade histology and elevated s-LDH. The corresponding relative risk values were in the order 2.4, 2.2, 1.9 and 1.6.

Key words: lymphoma, non-Hodgkin's, elderly, clinicopathological features, prognosis

be population-based in order to avoid bias due to patient selection. The aim of the present study, to our knowledge the first population-based study of lymphoma patients aged 70 years or older, was to illustrate the course of NHL in these patients by analysing their clinicopathological features and prognostic factors on the basis of the data from a Danish population-based NHL registry.

More than one half of all cases of cancer are diagnosed after age 65 and the incidence of cancer increases 1,000 times from age 40 to 80 [1]. Predictions for the year 2000 indicate that in industrialized countries approximately 60%-70% of all cancers will occur in persons aged 65 years or older [2]. In the case of nonHodgkin's lymphoma (NHL) different published series report 18%-38% of all patients as being elderly (heterogeneously defined by different age cut-points, Patients and methods i.e. 60, 65 or 70 years) at diagnosis [3-9]. However, although the elderly represent a large part of the total A population-based registry of all new cases of NHL in West-Denpatient population with NHL, little attention has been mark (Jutland and Funen, 2.7 million inhabitants), the so-called directed to the course of the disease in this particular LYFO-registry, was started on 1 January 1983, and is still ongoing. group of patients. The few existing clinico-pathological All hospitals and pathological laboratories in the region participated a study panel of 18 specialists from the three regional referstudies are retrospective and based on selected archival through ral centers (Aarhus, Odense, Aalborg). Excluded were cases of material, and only very few treatment protocols have acute lymphoblastic leukaemia, chronic lymphocytic leukaemia and been specifically devised for elderly patients with NHL multiple myeloma. For the present study, the data from 1,597 con[10-12]. As a consequence, the available information secutive cases of NHL registered between 1 January 1983 and 31 about biological behaviour and specific problems December 1988 were analysed. related to management and treatment of NHL in the Clinical and histological assessment elderly is largely insufficient. A more systematic knowledge of NHL in this large subgroup of patients must The registered parameters at the time of diagnosis included the come from prospective series, which should preferably patients name, date of birth, sex, histology, clinical stage, informa-

380 tion about constitutional symptoms, site(s) of involvement and a number of biochemical parameters. Routine clinical staging consisted of a thorough history and physical examination with particular attention to all lymphoid regions including rhinopharynx and Waldeyer's ring, biopsy from involved tissue, bone marrow aspirate and biopsy (Islam needle), chest x-ray, abdominal CT scan and/or lymphangiogram supplemented by a liver and spleen scan. Laparotomy was not used routinely. The following laboratory parameters were considered: s-LDH (normal range 150-500 U/l), s-ca" (normal range 1.19-1.29 mmol/1), s-urate (normal range 0.15-0.45 mmol/1), s-lgA (normal range 0.40-2.80 g/1), s-lgG (normal range 6.2-13.3 g/1) and s-IgM (normal range 0.18-1.30 g/1). The histological diagnosis and classification was reached by consensus among a panel of three haemopathologists. The Kiel Classification [13] was used. Moreover, cases were graded as low, intermediate or high according to the National Cancer Institute Working Formulation for Clinical Usage (WF) [14]. NHL of diffuse centroblastic type was considered as high grade [15|. A miscellaneous group corresponding to that of the WF was also considered.

and vincristin were reduced 25%. In patients with a left ventricular ejection fraction of 50% or less, where CHOP would have been the treatment of choice, doxorubicin was either completely avoided or substituted by 4'-epirubicin or mitoxantrone. After a diagnostic excision biopsy, cases with localized disease of low-grade histology received 'involved-field' RT as 30 Gy in 15 fractions given daily over 21 days. Patients with localized disease of intermediate- or high-grade histology received a combination of RT and CT, where the above mentioned 'involved-field' RT schedule was followed by 6-9 monthly courses of CHOP or analogue regimen. Patients with disseminated low-grade NHL were treated either with single agent CT (chlorambucil or cyclophosphamide) or CVP (cyclophosphamide, vincristine and prednisone) administered once a month for 6-9 courses. Cases of disseminated intermediate- or high-grade NHL were treated with CHOP or analogue regimen every 4 weeks for 8-9 courses. In NHL of lymphoblastic type, treatment was supplemented by CNS prophylaxis with intrathecal methotrexate. Additional RT was given for bulky tumours (>5 cm of max. diameter) or persistent lesions after CT. Nine percent of the patients did not receive any form of cytoreductive treatment.

Extranodal cases The definition criteria for extranodal cases in the LYFO-registry have been described in detail elsewhere |16]. Briefly, cases were defined as extranodal, if one or more extranodal sites were found to be involved without any sign of nodal involvement or along with a 'minor', i.e. <25% of the total tumour volume, nodal component. In accordance with other studies [17-19] tonsils and Waldeyer's ring were classified as nodal sites. Cases with bone marrow involvement were classified as extranodal only on condition that the general definition criteria for extranodal cases were fulfilled. Primary treatment regimens Two randomised trials have been conducted since 1983. One for low-grade lymphomas of advanced stage (III and IV), where patients were randomised to either monotherapy with chlorambucil or polychemotherapy with CHOP (cyclophosphamide, doxorubicin, vincristine, prednisone), and one for intermediate and high-grade lymphomas comparing CHOP with the platin-containing regimen CVBP (cisplatin, VP-16, bleomycin, prednisone). In the first trial, chlorambucil was given as 10 mg per day for 6 weeks followed by a maintenance dose of 2 mg per day up to a total treatment period of 9 months and CHOP as 9 monthly courses of cyclophosphamide 750 mg/m2, doxorubicin 50 mg/m2 and vincristine 1.4 mg/m2 all i.v. on day I and prednisone 100 mg orally on days 1—5. In the trial for intermediate- and high-grade lymphomas the same CHOP regimen was compared with CVBP given as 9 monthly courses consisting of cisplatin 50 mg/m2, bleomycin 30 mg and VP-16 100 mg/m2 all i.v. on day 1, supplemented by VP-16 100 mg/m2 orally on days 2-4 and prednisone 100 mg orally on days 1-4. Moreover, localised cases in the intermediate-/high-grade group were treated with 'involved-field' radiotherapy (RT) prior to chemotherapy (CT). RT consisted of 30 Gy in 15 fractions given by linear accelerators (4-15 MeV) as daily fractions Monday to Friday over 21 days. The randomisation rate for the elderly patient population was 10.1% (61 patients) versus a rate of 24.1% for patients aged less than 70 years. Twenty-two elderly patients entered the trial for low-grade NHL (11 in each treatment arm) and 39 the trial for intermediateand high-grade NHL (19 patients received CHOP and 20 CVBP). The reason for the low absolute number of elderly patients randomised in the different trials was an upper age limit for randomisation of 75 years. Therefore, if one only looks at those elderly patients who could be included in the trials (age group 70-75 years: 225 patients) a randomisation rate of 27.5 can be found. The large majority of non-randomised patients were treated following common general guidelines resulting in a rather homogeneous therapeutic background. As a general rule, for non-randomised patients aged 70 or older where CHOP or CHOP-like regimens were indicated, cyclophosphamide, doxorubicin or doxorubicin-analogues

Follow-up At the time of analysis the median follow-up for the elderly patient population was 45.3 months. The maximum possible follow-up was 92 months. Patients were followed with 2-monthly post-treatment controls within the first 6 months, 3-monthly controls up to 2 years and 6-monthly controls from 2 to 5 years after treatment was discontinued; thereafter once yearly up to a planned total post-treatment observation period of 10 years. Each control consisted routinely of a physical examination and a blood analysis (haematology, electrolytes, liver enzymes). At every second control a routine chest x-ray was taken. Statistical analysis Statistical differences between cross-tabulated values in frequency tables were evaluated by the Pearsson Chi-square test. Time at risk began at the date of conclusive histological diagnosis and ended at the date of last known status or at the date of death. Survival curves were plotted using the method of Kaplan and Meier. Statistical differences in the univariate analysis were evaluated by the Taron-Ware test [20). Prognostic factors for survival were identified by a Cox-regression analysis [21). Variables reaching statistical significance (p < 0.05) at univariate level were included in the multivariate analysis. Only cases without missing variables were included in the regression analysis. The BMDP statistical programme package, Statistical Software, Los Angeles, California was used. Programmes ID, 2D and 4F were used for data description and frequency tables, programmes 1L and 2L for univariate and multivariate analysis.

Results Demographic data

Within a 6-year period (1 January 1983-31 December 1988) 1,597 newly diagnosed NHL cases (826 males and 771 females) were entered into the LYFO registry. The population of Western Denmark during the same period was at a stable 2.8 million (1.39 million males and 1.41 million females) [22]. This corresponded to an average annual NHL incidence rate (IR) of 9.5/105 individuals (9.9 for males and 9.1 for females). The IR was 6.5/105 if age-adjusted for a standard World popu-

381

lation (WP), or 8.8/105 if age-adjusted for a standard European population (EP). Of all patients, 602 (37.7%) were aged 70 or older, 276 (46%) were men and 326 (54%) women. The age range was 70-94 years with a median value of 76.8 years. The median age of the whole patient population was 65.7 years. Figure 1 shows the average annual agespecific IR's for the whole patient population subdivided in 5-year age groups. The values show an exponential rise of NHL incidence as a function of age. The average annual IR for the elderly patient population (3*70 years) was 35.7/10 5 (39.1 for males and 32.3 for females) as compared with 6.6/105 which was the corresponding value for the patient population aged less than 70 years. Age-adjusted IR's were 34.6/105 (WP) or 34.8/105 (EP) for elderly patients and 5.2/105 (WP) or 6.5/105 (EP) for the younger ones. Clinicopathological features

The general characteristics of the elderly patient population (>70 years) are shown in Table 1, where they are compared with those of patients aged less than 70 years. Elderly patients had a significantly higher number of female cases, (proportional to the female predominance seen in the general population for that age group), of low clinical stages (I and II) and of extranodal manifestations. No difference was found between the two groups as to presence of constitutional symptoms or distribution of main histological subgroups (low, intermediate and high grade). Sites of involvement Fifty-two percent of all the elderly patients presented with apparently pure nodal manifestations, while 41% fulfilled the definition criteria for extranodal disease. Seven percent of the cases could not be defined as to this parameter. Table 2 shows the amount of cases in-

Table I. General patient characteristics: >70 years vs. <70 years. Characteristics

No. of patients (%)

X2

p-value

>70 yrs (n - 602)

<70 yrs (n-995)

276 (46) 326 (54)

551 (55) 444 (45)

13.5

0.0002

274 (46) 302 (50) 26(4)

359 (36) 621 (63) 15(1)

18.0

0.0000

378 (63) 198(33) 26(4)

661 (67) 319(32) 15(1)

0.5

0.4608

180 (30) 112(18) 250 (42) 60 (10)

316(32) 185(19) 382 (38) 112(11)

1.3

0.5286

314(52) 245 (41) 43(7)

588 (59) 330 (33) 77(8)

9.2

0.0099

Sex

male female Clinical stage 1-11 1I1-IV Unknown B-symptoms No Yes

Unknown Histology Low

Intermediate High Miscellaneous Localisation Nodal Extranodal Unknown

Table 2. Sites of involvement in extranodal cases (n - 245). Extranodal site

No. of cases"

Stomach Bone marrow Skin Thyroidb

51 39 30 24 21 15 14 12 10 10 8 6 2

Gut

Bones CNS

Testes Salivary glands Orbita and sinuses Lungs Liver Breast Others

11

In excess of 245; > 1 site can be involved in the same case. Male to female ratio: 0.19.

h~ co

Fig. I. Average annual age-specific incidence rates of NHL in Western Denmark (n - 1,597).

volving specific extranodal sites. The stomach was the most frequent site of extranodal involvement followed by the bone marrow. Cases with myeloid involvement were only classified as extranodal on condition that the general definition criteria for extranodal disease were fulfilled (see 'Patients and methods'). The gastric localisation together with the intestinal was found in almost a third of all extranodal cases. As previously reported [16], almost all cases of thyroid and testes lymphoma were found in elderly patients and there was a strong predominance of female cases among thyroid- and salivary glands lymphomas. Histology

Table 3 shows the distribution of histological sub-

382 Table 3. Analysis of histological subtypes.

Histology

No. of cases (c

df

p-value

24.7 10.2

1 1 1 1

n.s. 0.005 0.0000 0.001

0.1 0.1 2.0

1 1 1

n.s. n.s. n.s.

18.2

5.7 1.0

1 1 1 1 1 1 1

0.0000 n.s. n.s. 0.005 0.001 0.02

93.8

13

0.0000

X2

>70 yrs (n - 602)

<70 yrs (n - 995)

Low-grade Lymphocytic Lymphoplasmacytic/-cytoid, non polymorphic Follicular centroblastic/centrocytic Unclassifiable low-grade Intermediate Diffuse centroblastic/centrocytic Centrocytic (follicular and diffuse) Follicular centroblastic High-grade Diffuse centroblastic Immunoblastic Unclassifiable high-grade Lymphoblastic (Burkitt type) Lymphoblastic (convoluted type) Lymphoblastic (non specified) Miscellaneous"

180 (30) 22(4) 65(11)" 59(10)" 34 (6)" 112(19) 47(8) 62(10) 3(1) 250(41) 143(24)" 37(6) 59(10) 2 (0.3)" 1 (0.1)" 8(1)" 60(10)

316(32) 22(2) 67(7) 202 (20) 25(3) 185(19) 74(8) 99(10) 12(1) 382 (38) 145(15) 69(7) 90(9) 22(2) 23(2) 33(3) 112(11)

Total

602 (100)

995(100)

2.9 7.8

0.3 0.3 8.7

11.4

ns.

Degrees of freedom. Main contribution to overall x 2 value (if df — 1,X2 ~ 6.63 corresponds to p - 0.01).

groups among elderly patients and compares it with that of the patient population younger than 70 years. Although the distribution of the main histological categories (low, intermediate and high grade) within the two age groups was similar, some interesting significant differences emerged from the analysis of histological subtypes. Elderly patients within the low-grade category had a significantly lower occurrence of follicular centroblastic/centrocytic (CB/CC) lymphoma compensated by a higher frequency of all the other lowgrade subtypes. Within the high-grade category the older patient group had a significantly higher proportion of centroblastic (CB) diffuse cases along with a lower occurrence of lymphoblastic cases. The CB diffuse subtype was by far the most common among elderly patients, being more than twice as frequent as the second largest group, the lymphocytic/-cytoid nonpolymorphic. No age-related differences were found among the subtypes of the intermediate grade category. Biochemical parameters The number of cases in which some selected biochemical parameters were found to be abnormal is shown in Table 4, where the data are correlated to main histological categories of the WF and to age (3*70 years vs. <70 years). Common for both age groups was a higher frequency of cases presenting with elevated s-LDH within the high-grade and with paraproteinaemia within the low-grade category. However, elderly patients within the low- and high-grade group had more frequently hyperuricaemia than their younger counterparts and those of the low-grade group presented more frequently with high immunoglobulin levels (IgG and IgM).

Table 4. Correlation between biochemical parameters, main histological categories and age. Parameter*

% of cases (within subgroup) Low >70 (n-

<70

Intermediate

High

>70 (n-

<70 (n-

>10 (n-

<70 (n-

112)

185)

250)

382)

180)

(n316)

Elevated s-LDH Elevated s-urate Elevated s-ca""

29" 19C 3

20" 10 c 3

33" 16 4

28" 13 3

49" 2I C 5

47" 10 c 5

Elevated s-IgA Elevated s-IgG Elevated s-IgM

8 19C

19 b c

4 10 c llc

16C 12 7"

7C 8 9

15 10 10"

10 10 8

s-Paraprotein u-Paraprotein

19 b 12

13" 10

8" 8

7" 4

5C

4 h.c

r

• For reference values see 'Patients and methods'. " Significant difference (p < 0.05) between main histological categories within the same age group. c Significant difference between the two age groups within the same histological category.

Treatment and mortality data Preliminary overall results for the low-grade randomisation group showed no significant difference in survival or remission rates between patients treated with CHOP and those treated with chlorambucil. However, the time to progression was significantly different in the two groups, with median values of 45 and 34 months respectively (p = 0.03). This also applied to the randomised elderly patients (11 treated with chlorambucil and 11 treated with CHOP).

383

Overall results from the high-grade randomisation group showed that CHOP was superior to CVBP for inducing complete and overall response [23]. This did not apply to randomised elderly patients (>70 years), where CR rates for CHOP and CVBP were 58% vs. 55% and overall response rates 74%> vs. 80%, respectively. The median response duration was also similar in the two treatment groups. However, a significant difference in survival values was found. Elderly patients treated with CVBP (n = 20) had a 7-year survival of 54% (75th quantile: 1.8 years), while the same value for those treated with CHOP (n = 19) was 20% (median: 1.4 years; p - 0.0213). In the LYFO-registry toxicity data were only recorded for randomised patients. Chemotherapy was generally well tolerated (23]. A total of 322 polychemotherapy courses (198 CHOP and 124 CVBP) were administered to elderly patients. This corresponded to an average number of 6.6 courses/elderly patient in the CHOP arm and 6.2 courses/elderly patient in the CVBP arm. Median values were 9/elderly patient and 7/elderly patient, respectively. During a 7-year observation period 347 (58%) deaths were registered among the 602 elderly patients of the LYFO registry. Fifty-five percent of the cases of death were reported to be the result of progressive lymphoma, while 4% had apparently direct relation to treatment toxicity. Eighty-six (25%) patients died of causes not related to lymphoma and in 58 (16%) patients the cause of death was unknown. Among the 190 patients where lymphoma was the reported cause of death 98 had high-grade, 34 intermediate and 40 low-grade histology, while 18 belonged to the miscellaneous group. In 83% of the deceased patients a disease-status at the time of death was available: 213 (61%) had disseminated and 39 (11%) localised residual disease, whereas no residual disease was detectable in 36 (11%) patients. Autopsy was performed in 134 cases (39% of all deaths). Survival and prognostic factors

ment of >1 extranodal site, hypercalcaemia, splenic involvement, stage IV disease and hyperuricaemia. Only patients, where information about all these variables was available were included in the Cox regression analysis. Out of 602 elderly patients 446 satisfied this requirement. As shown in Table 6, the variables identified by the prognostic model were in order of relative risk (RR) magnitude: hepatic involvement, presence of B-symptoms, high-grade histology and elevated s-LDH. Clinical stage, splenic involvement and number of extranodal sites were correlated to B-symptoms and hepatic involvement. Hyperuricaemia was linked to elevated s-LDH and hepatic involvement, and hypercalcaemia to elevated s-LDH and hyperuricaemia. Survival curves for main histological categories are shown in Fig. 2. No significant difference in 7-year survival could be found between the low- and the intermediate grade group (48% vs. 38%, p - 0.1375), whereas the values for these two groups differed significantly from that of the high-grade category (low vs. high: 48% vs. 22%, p < 0.0001; intermediate vs. high: 38% vs. 22%, p - 0.0008). Fig. 3 shows survival curves for some specific histological subtypes. The lymphocytic type had the most favourable prognosis (68% survival at 6 years) while the immunoblastic type had the poorest survival with a 5-year value of 11% and a Table 5. NHL in the elderly: Univariate analysis of prognostic factors. Parameter

x2

B-symptoms Hepatic involvement Elevated s-LDH High-grade histology > 1 extranodal site Hypercalcaemia Splenic involvement Hyperuricaemia Disseminated disease Elevated s-IgA Elevated s-IgM

72.7 48.6 35.6 33.8 24.7 20.9 14.5 13.5 13.4

Sex

The overall 7-year survival of the patient population aged 70 or older was 35% (75th quantile 0.4 years, median 1.7 years) as compared to 57% (75th quantile 1.4 years) for patients younger than 70 (p < 0.0001). This expectable difference persisted after correction of survival data for those causes of death, which did not appear to be related to NHL. The corrected 7-year survival for patients aged 70 or older was 52% (75th quantile 0.9 years) versus 66% (75th quantile 2.3 years) for those aged less than 70 (p < 0.0001). Table 5 shows the predictive value for survival of some pretreatment clinical and biochemical parameters as assessed by univariate analysis. Those parameters reaching statistical significance level (p < 0.05) were subsequently included in the Cox regression analysis. They were: presence of B-symptoms, liver involvement, elevated s-LDH, high-grade histology, involve-

Positive Coomb's test Bone marrow involvement Paraproteinaemia Elevated s-IgG

jf

;I :I

3.2 2.9 1.3 0.7 0.7 0.2 0.0

p-value <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 0.0001 0.0002 0.0002 n.s. ns.

n.s. n.s. n.s. n.s. n.s.

n.s. - not significant at 0.05 level. Table 6. NHL in the elderly: Multivariate analysis of prognostic factors. Parameter

Coeff. var.'

SE b

p-value

RRC

Hepatic involvement B-symptoms High-grade histology Elevated s-LDH

0.8836 0.7955 0.6803 0.4751

0.2341 0.1337 0.1307 0.0003

0.0002 0.0000 0.0000 0.0003

2.4 2.2 1.9 1.6

• Coefficient of variation. b Standard error. c Relative risk.

384 100

Discussion

Over a 7-year period 1,597 newly diagnosed NHL cases were included in the population-based LYFOregistry. Half of the patients were over 65 years of age and more than one third (38%) were aged 70 years or more. These figures are higher than those of previously published series [3-9], which differed from the present so study by not being population-based. A populationbased cohort is not subject to selection due to specific eligibility criteria or local referral policies and can therefore provide a more realistic picture of the occurrence, presentation and prognosis of a given nosologic entity. This aspect becomes particularly important when the patient group to be analysed, e.g. elderly patients, is frequently selected and underrepresented in clinical trials [24]. It would therefore be of particular interest to compare the present incidence data with those from other population-based NHL registries. This is to a certain extent possible in the case of the I H Dutch NHL-registry from the Comprehensive Cancer Fig. 2. 7-Year survival of elderly patients according to the main Centre West (CCCW), which registered 640 new NHL histological categories of the WF (L — low grade, I — intermediate cases over a period of 5 years within a region comgrade and H — high grade). prising 1.58 million inhabitants [19]. The overall NHLincidence for this cohort was 8/105/year, i.e. slightly lower than that of the LYFO-cohort (9.5/105/year). To our knowledge, no age-specific incidence data from the CCCW-registry have yet been reported, thus not allowing specific comparisons between the elderly patient populations from the two registries. However, the reported overall median age of the Dutch series was identical to that found in our series, i.e. 65 years. BO

As to clinical stage at presentation, the elderly patient population of the present series had an equal representation of localised (stages I and U) and disseminated (stages III and IV) cases (46% vs. 50%), hence differing significantly from patients younger than 70 years, where localised and disseminated cases were 37% and 63%, respectively. However, for elderly patients one should bear in mind the possibility of a certain degree of 'understaging' which may lead to misa A registration, i.e. truly disseminated cases registered as ymmrm localised. In our series elderly patients with localised disease had a lower 7-year survival than their younger Ly CB/CC t IC C8 d • • • • IB counterparts (42% vs. 68%, p < 0,0001) and this Fig. 3. 7-Year survival of elderly patients according to specific histo- difference persisted after correction of the survival valogical subtypes (Ly — lymphocytic, CB/CC f - centroblastic/cen- lues for apparently NHL-unrelated deaths (63% vs. trocytic follicular, IC - lymphoplasmacytic/-cytoid, non-polymor- 76%, p = 0,0005). In a retrospective study conducted phic, CC — centrocytic, CB d ~ centroblastic diffuse and IB ~ immuby the Lymphoma Study Group of the European Organoblastic). nisation for Research and Treatment of Cancer on 137 patients aged 70 years or more referred from 13 Euromedian survival of 7 months. Interestingly, no differ- pean institutes, the clinical stage at presentation was ence could be seen between the centrocytic and the dif- localised in 60% and advanced in 37% of the patients fuse centroblastic type (7-year survival: 33% and 26% [9]. No data for patients younger than 70 years were rerespectively). The 7-year survival of both the follicular ported. In an older retrospective study performed at centroblastic/centrocytic type and the lymphoplasma- the Division of Oncology, Stanford University Medical Center [6], patients aged more than 70 years reprecytic/-cytoid type was 48%. sented less than 20% of a total of 337 adult NHL patients and no cases with localised disease were found

385

within this age group. Such differences in clinical stage distribution between different series are most likely explained by a varying degree of patient selection and differences in staging procedures. Therefore a comparison of clinical stage data is only meaningful, if the patient populations to be compared are in fact comparable as to basic parameters, e.g. age and sex, and staging procedures. In agreement with previous reports [3, 5, 8] we observed a clear predominance of 'diffuse' over 'follicular' histological patterns (ratio 6 :1) in elderly patients. No age-related difference in the distribution of the main WF-categories could be detected. However, some interesting differences emerged from the analysis of histological subtypes: patients aged 70 years or more had a lower occurrence of follicular centroblastic/centrocytic and a higher occurrence of diffuse centroblastic cases than their younger counterparts (<70 years). It is unclear whether these findings could reflect an underlying age-related biologic diversity of NHL. In a study of Chow et al. [25], who analysed the morphology and immunophenotype of spontaneously occurring immunocytoma-like tumours in SJL mice, advancing age was associated with loss of plasmacytoid characteristics and of y-heavy chain expression and with the appearence of less differentiated cells resembling centrocytes, centroblasts and immunoblasts. A progressive loss of differentiation in these B-lymphomas as a function of age was hypothesized. The Cox-regression analysis performed on the entire elderly patient population (all stages, all histologies) identified some pretreatment clinical features that were important independent predictors of outcome for elderly patients with NHL. These features were indicators of tumour burden (hepatic involvement, B-symptoms and elevated s-LDH) and of tumour-cell type (high-grade histology). Interestingly, the elderly patients with NHL of high-grade histology, who were randomised to receive a platin- and etoposide-containing regimen (CVBP) had a significantly (p - 0.02) better survival than the CHOP-treated group. A closer analysis of the presentation characteristics of these randomised patients revealed an overrepresentation of cases with elevated s-LDH in the CHOP-treated group (14 out of 19 cases in the CHOP group vs. 5 out of 20 cases in the CVBP group, p =0.0035). As new therapeutic approaches [10-12] and supportive measures become available (e.g. anthracyclines combined with cardioprotective drugs or substituted by less cardiotoxic analogs; colony stimulating factors), a patient stratification according to the above mentioned prognostic model could, if the efficacy of the model is confirmed in unrelated series, prove itself useful for the planning and the analysis of future therapeutic trials specifically devised for elderly patients.

Acknowledgements

The authors wish to thank Mr. Leif Spange Mortensen, UNI-C, Aarhus, Denmark, for expert statistical advice and Ms. Jacinta Kendellen, Ms. Inge Lise Neerholt and Ms. Vibeke Klebe for skillful secretarial assistance. References 1. Kennedy BJ. Needed: Clinical trials for older patients. J Clm Oncol 1991; 9: 718-20. 2. Monfardini S, Chabner B. Joint NCI-EORTC consensus meeting on neoplasia in the elderly. Eur J Cancer 1991; 27:653—4. 3. Jones SE, Fuks Z, Bull M et al. Non-Hodgkin's lymphomas IV. Clinicopathologic correlation in 405 cases. Cancer 1973; 31: 806-23. 4. Patchefsky AS, Brodovsky HS, Menduke H et al. Non-Hodgkin's lymphomas: A clinicopathologic study of 293 cases. Cancer 1974; 34: 1173-86. 5. Narwani BM, Kim H, Rappaport H et al. Non-Hodgkin's lymphomas. A clinicopathologic study comparing two classifications. Cancer 1978; 41: 303-13. 6. Elias L. Differences in age and sex distributions among patients with non-Hodgkin's lymphoma Cancer 1979; 43: 2540-6. 7. Anderson T, Chabner BA, Young RC et al. Malignant lymphoma I. The history and staging of 473 patients at the National Cancer Institute. Cancer 1982; 50: 2693-707. 8. Carbone A, Tirelli U, Volpe R et al. Non-Hodgkin's lymphoma in the elderly. A retrospective clinicopathologic study of 50 patients. Cancer 1986; 57: 2185-9. 9. Tirelli U, Zagonel V, Serraino D et al. Non-Hodgkin's lymphomas in 137 patients aged 70 years or olden A retrospective European Organization for Research and Treatment of Cancer Lymphoma Group study. J Clin Oncol 1988; 6: 1708-13. 10. Tirelli U, Carbone A, Zagonel V et al. Non-Hodgkin's lymphomas in the elderly: Prospective studies with specifically devised chemotherapy regimens in 66 patients. Eur J Cancer Clin Oncol 1987; 23: 535-40. 11. Sonneveld P, Michiels JJ. Full dose chemotherapy in elderly patients with non-Hodgkin's lymphoma: A feasibility study using a mitoxantrone containing regimen. Br J Cancer 1990; 62: 105-8. 12. O'Reilly SE, Klimo P, Connors JM. Low-dose ACOP-B and VABE: Weekly chemotherapy for elderly patients with advanced stage diffuse large-cell lymphoma. J Clin Oncol 1991; 9:741-7. 13. Lennert K. Malignant lymphomas other than Hodgkin's disease. Berlin: Springer 1978. 14. 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: 2112-35. 15. Kriiger GRF, Medina JR, Klein O et al. A new working formulation of non-Hodgkin's lymphomas. Cancer 1983; 52: 83340. 16. d'Amore F, Christensen BE, Brincker H et al. Clinicopathological features and prognostic factors in extranodal non-Hodgkin lymphomas. Eur J Cancer 1991; 27: 1201-8. 17. EORTC Lymphoma Cooperative Group. Protocol No. 20855. Phase III trial comparing CHVmP/VCR-BLM vs. modified ProMACE/MOPP regimen for patients with stages II, III, IV lymphomas of intermediate and high-grade malignancy. Brussels, 1985. 18. EORTC Lymphoma Cooperative Group. Protocol No. 20856. Maintenance of remission with recombinant alpha-2 interferon in patients with stages III and IV low grade malignant nonHodgkin's lymphoma. Brussels, 1985.

386 19. Otter R, Gerrits WBJ, v.d. Sandt MM, Hermans J, Willemze R. Primary extranodal and nodal non-Hodgkin's lymphoma. Eur J Cancer Clin Oncol 1989; 25: 1203-10. 20. Tarone RE, Ware J. On distribution-free tests for equality of survival distribution. Biometrika 1977; 64: 156-60. 21. Cox DR. Regression models and life tables. J Roy Statist Soc 1972; 34: 187-220. 22. Danmarks Statistik. Statistical Yearbooks 1983-1988. Copenhagen, Danmarks Statistik, 1983-1988. 23. Andersen J, Thorling K, Bentzen SM et al. Phase III trial of cyclophosphamide, doxorubicin, vincristine and prednisone (CHOP) versus cisplatin, etoposide, bleomycin and prednisone (CisEBP) for the treatment of advanced non-Hodgkin's lymphoma of high grade malignancy. Acta Oncol 1990; 29:995-9.

Book review Complications of cancer management P. N. Plowman,

T. J. McElwain, A. T. Meadows (eds). ButterworthHeinemann, Guildford, 1991. 504 pp, ill., £98.00. A modest but definite improvement in the treatment of cancer patients has occurred within the last decades. At the same time, the management has become much more complex and intensive, and the detailed knowledge of the complications related to the three main treatment modalities - surgery, radiotherapy and chemotherapy - is important. The latter is the subject of this book. Have the authors then succeeded in giving a critical and comprehensive overview? The answer is yes, but it is only a qualified yes, although the blemish could easily be remedied by the editors in the next issue. Of the three modalities, complications secondary to surgery have been left a little bit out in the cold by being discussed in only two of the 33 chapters that the book contains. Otherwise, the book is rather well balanced, with important up-to-date contributions mainly

24. Tirelli U, Zagonel V, Monfardini S. Common errors in conducting and reporting clinical trials in non-Hodgkin's lymphomasand patients'age. Eur J Cancer 1991; 27:811. 25. Chow EYW, Ho FCS. Age-related changes in the morphology and immunophenotype of spontaneous lymphomas of SJL/N mice. J Pathol 1988; 156: 331-9. Received 10 December 1991; accepted 31 January 1992. Correspondence to: F. dAmore, M.D. Dept. of Haematology Odense University Hospital 5000 Odense C, Denmark

Annals of Oncology 3: 386, 1992.

from North America and Europe, describing the clinical organ toxicity as related to the use of chemotherapy and radiotherapy. Sequelae to the combined applications are also given. Experimental pathophysiology, be it clinical or preclinical, is unfortunately rarely mentioned. A certain overlapping takes place - especially in section 1 - where repetition occurs between the chapers dealing with complications in adult and childhood malignancies. Some of the authors also have difficulties in distinguishing between complications secondary to the disease versus secondary to the treatment. Altogether, however, it is a useful reference book, which nicely complements the large series of existing textbooks in oncology. It should be found on the shelves of the conference room of all clinical oncology departments. H. Hansen Copenhagen