- Email: [email protected]
Biological markers may add to prediction of outcome achieved by the international prognostic score in Hodgkin's disease
Annals of Oncology 11: 1405-1411, 2000. O 2000 Kluwer Academic Publishers. Printed tn the Netherlands
Original article Biological markers may add to prediction of outcome achieved by the international prognostic score in Hodgkin's disease U. Axdorph,1 J. Sjoberg,1 G. Grimfors,1 O. Landgren,' A. Porwit-MacDonald2 & M. Bjorkholm1 'Division of Hematology, Department of Medicine, 2 Department of Pathology and Cytology, Karolinska Hospital and Institutet, Stockholm, Sweden
survival (CSS) in univariate analyses were: increased serum levels of IL-10, sCD30 and CRP, anaemia, low levels of Background: The International Prognostic Score (IPS) identi- albumin (P < 0.001); stage TV (P = 0.003), age 3= 45 years fies seven independent factors predicting progression-free and (P = 0.006), increased serum levels of sCD25 (P = 0.010), overall survival in advanced stage Hodgkin's disease (HD). low lymphocyte counts (P = 0.020). Serum IL-10 added progThe IPS is also applicable in limited disease. However, the IPS nostic information to that achieved by the IPS: patients with a does not identify patients with a very poor prognosis. The aim high score and increased serum IL-10 had a very poor outcome of this study was to define biological markers which may add with a five-year CSS of 38%. Patients with increased serum levels of sCD30 and a high score also had a poor outcome with to the IPS in predicting outcome. Patients and methods: One hundred forty-five patients afive-yearCSS of 54%. (> 15 years) with HD of all stages and histopathology subConclusion: Serum levels of IL-10 and sCD30 may add to groups were included. In addition to factors included in the IPS in prediction of outcome in HD, and should be validated IPS, serum levels of CRP, sCD4, sCD8, sCD25, sCD30, in large, prospective studies. sCD54, interleukin (IL)-10, |32-microglobulin and thymidine kinase were analysed. Key words: Hodgkin's disease, interleukin-10, prognosis, progResults: The strongest predictors of a poor cause-specific nostic score, sCD30. Summary
Introduction
Current therapies fail to cure about one third of patients with advanced Hodgkin's disease (HD). However, longterm remissions and cure occur even in patients who stop treatment after two to four cycles of chemotherapy [1]. This may suggest over-treatment of an undefined fraction of patients with increasing risk of late complications such as secondary malignancies, cardiac toxicity and infertility [2-4]. In early stage disease, short-term chemotherapy is today commonly added to radiotherapy due to a high relapse rate among patients treated with the latter modality alone [5-8]. The dose of radiotherapy is often reduced in these combined modality treatments with the hope of diminishing the late toxicity of irradiation [5, 6]. Thus, it is of great importance to identify additional markers, which may further characterise patients with a very dismal prognosis, who may benefit from aggressive upfront treatment, and, in addition, patients with a good prognosis, who can be given limited treatment without jeopardising outcome. A number of mainly biological markers have been reported to predict outcome in the literature: Abnormal lymphocyte function [9-11], C-reactive protein (CRP) [12], serum levels of soluble (s)CD4 (Grimfors, data to be published), sCD8 [13, 14], sCD25 [15-17], sCD30
[14, 18, Axdorph, data to be published], sCD54 [19, 20], IL-10 [21], p2-microglobulin ((32M) [22], thymidine kinase (TK) [23], IL-6 [24] and tumour necrosis factor (TNF) [25]. These biological markers were, for obvious reasons, not analysed in the International Prognostic Factors Project [26]. In that retrospective, multicenter project, 19 routinely documented demographic and clinical features were analysed at diagnosis in 1618 patients with HD receiving combination chemotherapy. Seven factors (age ^ 45 years, male sex, stage IV, albumin <40 g/1, haemoglobin (Hb) < 105 g/1, white blood cell (WBQ counts ^ 15 x 109, lymphocyte counts <0.6 x 109 or <8% of WBC counts) related to poor progression-free and overall survival were combined to give a prognostic score, the IPS. This score is also applicable for HD patients with limited disease [27]. However, neither IPS nor previously published prognostic indices [28] allow the identification of a distinct group of patients with a very poor prognosis, defined as a five-year survival rate of < 50%. The aim of this study was to identify biological markers, which may add to the IPS in predicting outcome in HD for further validation in large, prospective studies.
1406 tigations for serum markers [13, 16, 20]. The median age at diagnosis was 36 years (range 15-87 years) and the median follow-up for surviving patients was 143 months (range 44-272 months). The extent of Biopsies and methods disease was evaluated according to the Ann Arbor staging classificaThe histopathology and immunophenotyping of biopsies were reviewed tion [31] with the following investigations: chest roentgenograms, and, when necessary, complementary immunostainings for CD15, radionuclide scanning of liver and spleen (1974-1992) and bone (1975— 1992) and computed tomographic scanning (1977-1992) of the abdoCD30, CD20, LN-1, CD79a, CD3, UCHL-1 and EMA (avidin-biotinperoxidase complex (ABC) technique) were performed to confirm the men and chest, when clinically indicated. Bone marrow involvement diagnosis. The R.E.A.L. Classification [29] was applied, including the was assessed by bone marrow biopsies and aspirates. Bulky disease was division of nodular sclerosis (NS) according to the British National defined as a mediastinal mass with a diameter exceeding one third of the maximal mediastinal width or any tumour manifestation with a Lymphoma Investigation [30]. diameter of > 10 cm. Complete remission (CR) corresponded to complete regression of all palpable or histologically documented tumours and resolution of all radiographic and biochemical abnormalities due Patient characteristics to HD for a minimum of three months. In this retrospective study, 145 adults (>15 years) with previously Treatment (Table 2) untreated HD, diagnosed between 1974 and 1992, were included (Table 1). Criteria for inclusion were known results of previous invesDetails of treatment have been described previously [1, 11, 13, 32, 33]. In short, patients with limited disease (stage I-IIA) were given radioTable 1. Patient characteristics. therapy [11, 13, 32]. However, patients diagnosed 1974-1979 were included in a study regarding the value of early splenectomy. In this n (%) study, patients mainly in stage IIA-IIIA were given total nodal irradiation [33]. During this period, patients with stage IIIB-IV 145 (100) All patients disease were given MOPP or CCNU-OPP chemotherapy [33]. From Sex 1979, patients with stage IIB-IV disease received 6-8 cycles of MOPP/ Male 80 (55) ABVD chemotherapy with irradiation given to bulky disease [1]. Female 65 (45) All clinical and immunological studies were approved by the Ethics Age Committee at Karolinska Institutet and patients gave informed consent. < 45 years 103 (71) > 45 years 42 (29) Clinical stage Biological markers I 38 (26) II 46 (32) The analysed samples for each biological marker vary between 86 and III 32 (22) 123 (Table 3). Serum samples were collected before start of treatment IV 29 (20) and stored at -70 °C until use. Sera were collected prospectively, and B-symptoms were analysed on the same occasion. Serum levels of soluble markers Yes 57 (39) were determined by means of commercially available sandwich immuNo 88 (61) noassay-methods: sCD4, sCD8 [13], sCD25 [16], sCD54 [20] (CellBulky disease free® test kits, T-cell Science, Cambridge, Massachusetts), sCD30 Yes 33 (23) (Dako CD30 (K.i-1 Antigen) ELISA) and IL-10 (Quantikine human No 112 (77) IL-10, R&D Systems Europe). The minimal detectable value of IL-10 Histopathology was 7.8 pg/ml. Thus, IL-10 was dichotomized at 10 pg/ml in accordLP 12 (8) ance with Sarris et al. [21]. P2-microglobuhn (P2M) was analysed by NS unclassified 2 (1) a competitive luminometric assay (LIA-mat® p2-microglobulin, CamNS1 44 (30) bridge Life Sciences pic). Thymidine kinase (TK.) was measured in a NS2 31 (21) radio-enzyme assay (Prolifigen® TK-REA, AB Sangtec Medical). MC 50 (35) LD 1 (1) Controls Unclassified" 5 (4) The control population consisted of healthy blood donors and mem* Slides or biopsies were not available forreviewin three of thesefivecases. bers of the laboratory staff. Patients had significantly higher median
Patients and methods
Table 2. Treatment according to clinical stage. Treatment
Total number oi patients
Number of patients Stage I (n = 38)
Stage II (n = 46)
Stage III (n = 32)
Stage IV (/i = 29)
Radiotherapy Total nodal irradiation Subtotal nodal irradiation Mantle field, inverted Yfield,involved field
72 21 17 34
3 9 23
10 7 11
8 1 0
0 0 0
Chemotherapy MOPP or CCNU-OPP MOPP/ABVD
56 10 46
1 1
1 7
6 15
2 23
Combined modality treatment MOPP/ABVD and radiotherapy
17 17
1
10
2
4
1407 Table 3. Patients with biological markers analysed in percentage of all patients (/i = 145) in the study. Biological marker
No. patients analysed (%) patii
IL-10 sCD4 sCD8 sCD25 sCD30 sCD54 CRP
123 (85) 99(68) 86 (59) 113 (78) 93(64) 115(79) 99 (68) 99 (68) 97 (67)
P2M TK
Cause-specific Survival by Prognostic Score
100
Q.
I
O
8 values than controls regarding sCD25 (P < 0.001,17 controls), sCD30 (P < 0.001,43 controls), f32M (/> < 0.001,79 controls),TK (P < 0.001, 160 controls), sCD54 (P = 0.054, 20 controls). The patient population did not statistically differ from controls with regard to IL-10 (8 controls), sCD4 (10 controls) and sCD8 (20 controls). CRP, Hb, WBC and total lymphocyte counts and albumin were determined according to standard methods.
Statistical methods The Mann-Whitney U-test was used to calculate differences in distributions of serum markers between groups. All laboratory variables were initially analysed as continuous except IL-10, which was analysed as dichotomized only. In congruence with the IPS, Hb, WBC and total lymphocyte counts, albumin, age and stage were also dichotomized. Cut-off points were chosen according to the literature [21, 26]. When combining sCD30 with the IPS, sCD30 was dichotomized. The cutoff point for sCD30 was chosen as the median value. Data skewly distributed were logarithmically transformed. Linear regression and logistic regression analyses were used when analysing the association between year of diagnosis with the biological variables. Cause-specific survival (CSS) was defined as the time from diagnosis to death from HD or death related to treatment of the disease including second malignancies. Observations were censored by end of follow-up or death without signs of HD. In complete remitters, disease-free survival (DFS) was defined as the time from last day of treatment to relapse or death from HD or death related to treatment of the disease including second malignancies. Observations were censored by end of follow-up or death without signs of HD. Survival curves were constructed by the Kaplan-Meier procedure [34]. Differences in the survival for categorical factors were calculated by Gehan's Wilcoxon test [35]. Cox regression analyses were used for continuous factors univariately and for the multivariate analyses [36]. In the latter analysis, only markers, significant in univariate analyses, with less than 15% missing values, were analysed. Thus, serum levels of sCD25, sCD30 and CRP were not analysed with respect to their independent significance.
Results At follow-up, 57 patients had died. Forty patients died from progressive HD or from complications to the treatment including three patients, who developed myelodysplastic syndrome/acute leukaemia. The year of diagnosis was not associated with CSS or DFS (data not shown). The CR rate was 86%. Thirty-eight of the one hundred twenty-five patients (30%) who achieved a CR relapsed during the observation period. In the whole series, the five-year CSS was 82%. Among patients achieving a CR, the five-year DFS was 69%.
10
12
14
16
18
20
Years after diagnosis Figure 1. Cause-specific survival. IPS divided in a high-risk group (3-7 factors, 47 patients) and a low-risk group (0-2 factors, 96 patients). P < 0.001, Gehan's test, df = 1.
Cause-specific survival Evaluation of the International Prognostic Score Low serum albumin (P < 0.001), anaemia (P < 0.001), stage IV (P = 0.003), age ^ 45 years (P = 0.006), low lymphocyte counts (P — 0.020) but not male sex and increased WBC counts predicted a poor CSS in univariate analyses. In the multivariate analysis of IPS factors reaching significance in univariate analyses, albumin (P = 0.007), Hb (P = 0.026) and stage (P = 0.041) retained significance with a borderline value for age (P - 0.053). In the multivariate analysis of IPS factors, 143 patients were analysed. Two patients had missing data regarding albumin and total lymphocyte count, respectively. High-risk patients (score 3-7) had a significantly worse survival as compared to low-risk patients (score 0-2, P < 0.001, Figure 1). Biological markers Increased serum levels of sCD30 (P < 0.001), CRP (P < 0.001) and sCD25 (P - 0.010) as well as serum IL-10 ^ 10 pg/ml (P < 0.001) predicted a poor CSS in univariate analyses. The predicted prognostic capacity of serum levels of sCD4, sCD8, sCD54, p"2M and TK did not reach statistical significance. We did as well perform survival analyses with control for year of diagnosing. Controlling for this factor did not change the results (data not shown). In the multivariate analysis including significant factors from the IPS, IL-10 retained an independent prognostic significance (Table 4). For IL-10, controlling for Hb and albumin, the hazards were not different for the two values considered as judged by log minus log plots (data not shown). Serum IL-10, sCD30 and the International Prognostic Factor Score Patients were subgrouped according to IPS and dichotomized serum IL-10 values (P < 0.001, Figure 2). IL-10
1408 Prognostic Score and sCD30
Prognostic Score and IL-10 100
Score 0-2 4ILKK10 I
1 60
CO
40
Score 0-2 & sCD30<40
1
80 3 in
100
Score 0-2 &IL1>10
Score 3-7 &ILK10 o
a.
I
Q. in
| Score 3-7 & IL1>10
i,
at
20
CD
o
0 0
2
4
6
8
10
12
14
16
18
20
0
Years after diagnosis
score 0-2 and score 0-2 and score 3-7 and score 3-7 and
IL-10 <10 pg/mL: IL-10 >10 p g / raL IL-10 <10 pg/mL: IL-10 >10 pg/mL:
4
6
8
10
12
14
16
18 20
Years after diagnosis
No of patients
Five-year survival
Ten-year survival
80 7 26 8
91% 86%
88% 43% 52% 25%
69% 38%
2
score 0-2 and sCD30 <40 U/mL: score 0-2 and sCD30 >40 U/mL score 3-7 and sCD30 <40 U/ mLscore 3-7 and sCD30 >40 U/mL
No of patients
Five-year survival
Ten-year survival
38 22 9 24
97% 77% 89% 54%
91% 77% 89% 30%
Figure 2. Cause-specific survival. IPS and serum IL-10. P < 0.001, Gehan's test for heterogeneity. x2-statistic = 26.0, df = 1.
Figure 3 Cause-specific survival. IPS and serum sCD30. P < 0.001, Gehan's test for heterogeneity. z2-statistic = 26.4, df = 3.
added prognostic information to that achieved by the IPS: patients with a high score and increased IL-10 levels had a very poor outcome with a five-year CSS of 38%, while patients with a low score and low IL-10 levels had a five-year CSS of 91%. Atfiveyears, the CSS did nor differ in patients with low score and high and low IL-10 levels, respectively. However, the 10-year CSS was 43% for patients with a low score and increased IL-10 levels as compared to 88% for patients with a low score and low IL-10 levels. Subgrouping according to IPS and dichotomized serum sCD30 values showed a five-year CSS for patients with a high score and increased sCD30 levels of 54%, being significantly worse than that of other subgroups (P < 0.001, Figure 3).
Disease-free survival
Table 4. Cox multivariate (forward stepwise) regression with respect to poor cause-specific survival. Variable
Relative hazards
95% Confidence interval
Evaluation of the International Prognostic Factor Score Increased WBC counts (P - 0.014) predicted a poor DFS in univariate analysis. Low levels of albumin, anaemia, stage IV, age ^45 years, low lymphocyte counts or male sex did not reach statistical significance. Biological markers Increased serum levels of sCD25 (P = 0.009), sCD30 (P = 0.028), CRP (P = 0.034) and sCD8 (P = 0.051) predicted a poor DFS in univariate analyses, as well as serum IL-10 ^ 10 pg/ml {P - 0.010). The prognostic discriminatory capacity of serum levels of sCD4, sCD54, P2M, TK did not reach statistical significance. Serum IL-10 had an independent impact on DFS in the multivariate analysis (P = 0.007) with the WBC count of the IPS as a competing factor. Discussion
P
In spite of considering established prognostic factors such as age, stage, haemoglobin values and others, there is a great variability in outcome of patients with ad3.33 Serum IL-10 5= lOpg/ml 1.56-7.08 0.002 vanced stage HD. The search for additional prognostic 2.33 Haemoglobin < 105 g/1 0.035 1.06-5.0 factors has been intensive for many years. The aim has 2.56 Albumin <40 g/l 0.037 1.05-6.25 been to identify patients with a very poor prognosis, Competing factors from the IPS were: stage IV, age > 45 years, low who may benefit from aggressive upfront treatment. lymphocyte counts. Also, to find patients, who can be treated less aggresIn this multivariate analysis of IPS factors and IL-10,122 patients were sively without compromising the anti-tumour effect, but analysed; 1 of the 123 patients with values of IL-10 had missing data with less acute and late toxic complications. Recently, regarding albumin. All competing factors had less than 15% missing values. Sarris et al. reported that serum levels of IL-10 was a
1409 strong biological prognostic factor [21], and other groups have described the same results [37, 38]. In limited disease, high relapse rates after radiotherapy alone have led to the use of a combination of chemotherapy and radiotherapy in many centres. It is likely, that this approach will lead to over-treatment of some patients, increasing the incidence of late complications. Thus, apt predictors of prognosis are of equal importance in patients with limited disease. Previously known prognostic factors have been included in different prognostic indices for survival [28], but these indices have failed to identify patients at very high risk. An important work was the IPS study, where seven factors related to progression-free and overall survival in HD patients receiving combination chemotherapy were identified [26]. Interestingly, some of the previously known variables, among them extranodal disease, histologic type, erythrocyte sedimentation rate, were found not to add prognostic information in that study. This may in part be due to an improved success rate during recent years. The score predicted five-year rates of freedom from progression of disease in the range of 42% to 84%, and 56% to 89% in overall survival. Thus, even this prognostic score could not identify a subgroup of patients with a very poor prognosis (afive-yearsurvival < 50%) and a potential need of more aggressive upfront treatment. Despite the limited number of patients included in the present study, the results were in a good accordance with those of the IPS study: five of the seven factors significantly predicted a poor CSS in univariate analyses, and four factors remained in the multivariate analysis. A very good prognostic discriminatory capacity was also seen, when the patients where divided into a low-risk and a high-risk group (score 0-2 and 3-7, respectively). Fifty-seven patients (38%) in our study had limited disease (stage I-IIA, treated with radiotherapy alone). In comparison, the inclusion criterion in the international study was advanced disease, treated with chemotherapy ± radiotherapy. However, Franklin et al. has reported that the IPS also is applicable for patients with limited disease [27]. We analysed CSS as end point instead of freedom from progression, which was the main end point reported in the IPS study. In agreement with Jack et al. [39], who also evaluated the score with CSS, we find CSS to constitute a definitive measure of failure. In addition, we also analysed prognostic factors in relation to DFS. An important factor to consider with regard to DFS is the greater risk of relapse following radiotherapy alone [6]. However, we found no difference in time to relapse or disease-free survival among CR patients according to initial treatment (radiotherapy, chemotherapy or combined modality treatment; data not shown). In the IPS study, analyses were restricted to patients between 15 and 65 years. In the search for prognostic variables, that will help the clinician in individualising treatment, we find it important not to exclude older patients. Optimising treatment in older patients with intercurrent diseases constitutes a major
therapeutic challenge. To avoid misinterpretation of survival because of intercurrent diseases, we analysed cause-specific survival. In conclusion: in spite of a restricted number of patients, inclusion of patients of all ages and stages and the use of other end points, this patient population showed a good concordance with the results of the IPS study. A useful prognostic marker has to be 1) easy to analyse and 2) show reliable reproducibility, and of course, be a strong predictor of prognosis, identifiable also in smaller groups of patients. In a series of previous reports we have evaluated biological markers relevant for HD with the goal to refine our ability to predict prognosis [9-11, 13, 16, 20]. There are certain other variables, for example IL-6 and TNF/TNF-receptors, that would also be of interest to evaluate in association to IPS [24, 25]. However, this is already a rather variable intense study, and we had to limit the number of variables. Partly due to the same reason, we did not include blood lymphocyte function tests, although these have in previous studies been shown to be a strong predictor of prognosis [9-11]. In addition, tests of lymphocyte function are rather complex, and their reproducibility may be questioned. In order to address the issue whether cytokine levels remain stable during storage at -70 °C over several years, we tested the association between year of diagnosis with CRP, IL-10 and with the log values of sCD4, sCD8, sCD25, sCD30, sCD54, P2M and TK. sCD4, sCD54, 02M and TK showed positive association with calendar time on 5% risk level. As a consequence of this, we repeated all survival analyses with control for year of diagnosis. However, the overall results where not changed (data not shown). The univariate analyses regarding the significant correlations of a poor CSS and increased serum levels of IL-10, sCD30, sCD25 and CRP confirmed results of previous studies concerning biological markers and prognosis [12, 14-16, 18, 21, 37, 38]. In the multivariate analysis, IL-10 emerged as an independent variable, which is also in agreement with previous studies [21, 37, 38]. IL-10 seems to be helpful in adding prognostic information to that achieved by the IPS: patients with a high score and increased IL-10 had a very poor outcome, patients with a low score and low levels of IL-10 had a good prognosis, and patients with the combination of a low score and high IL-10 as well as a high score and low IL-10 showed an intermediate pattern. In addition, a high serum level of IL-10 also predicted a poor DFS. Patients with the combination of high serum levels of sCD30 and a high score according to IPS also had a poor prognosis. Serum levels of sCD25, sCD30 and CRP were not analysed with respect to their independent significance because of > 15% missing data. We conclude that IL-10 and sCD30 may add to prediction of outcome in HD, and thereby be helpful in identifying high- and low-risk patients with the goal to optimise treatment. Hopefully these results will encourage others to measure biological factors such as IL-10
1410 and sCD30 in order to allow a future multivariate analysis in a substantial number of patients. Acknowledgements This study was supported by grants from the Swedish Cancer Society, the Stockholm County Council and Karolinska Institutet Foundations. We thank Professor Derek Crowther for critical review of the manuscript. References 1. BjSrkholm M, Axdorph U, Grimfors G et al. Fixed versus response adapted MOPP/ABVD chemotherapy in Hodgkin's disease. Ann Oncol 1995; 6: 895-9. 2. Leyiand-Jones B, O'Dwyer P, Alonso T. Late effects of chemotherapeutic agents. In Lacher MJ, Redman JR (eds): Hodgkin's Disease: The Consequences of Survival. Philadelphia, Pennsylvania: Lea & Febiger 1990; 47-62. 3. Van Leeuwen FE, Klokman WJ, Veer MB et al. Long-term risk of second malignancy in survivors of Hodgkin's disease treated during adolescence or young adulthood. J Clin Oncol 2000; 18: 487-97 4. Swerdlow AJ, Barber JA, Hudson GV et al. Risk of second malignancy after Hodgkin's disease in a collaborative British cohort: The relation to age at treatment. J Clin Oncol 2000; 18: 498-509. 5. Noordijk E. Radiotherapy in early stage Hodgkin's disease. Principles and results of recent clinical trials. Ann Oncol 1998; 9 (Suppl 5): 63-5. 6. Cosset JM, Mauch PM. The role of radiotherapy for early stage Hodgkin's disease: Limitations and perspectives. Ann Oncol 1998; 9 (Suppl 5): 57-62. 7. Vlachaki M, Hagemeister F, Fuller L et al. Long-term outcome of treatment for Ann Arbor stage I Hodgkin's disease: Prognostic factors for survival and freedom from progression. Int J Radiat Oncol Biol Phys 1997; 38: 593-9. 8. Glimelius B, Kalkner M, Enblad G et al. Treatment of early and intermediate stages of supradiaphragmatic Hodgkin's disease: The Swedish National Care Program experience. Ann Oncol 1994; 5: 809-16. 9. BjSrkholm M, Holm G, Mellstedt H et al. Prognostic factors in Hodgkin's disease. II. Role of the lymphocyte defect. Scand J Haematol 1978; 20: 306-18. 10. Wedelin C, Bjorkholm M, Holm G et al. Lymphocyte function in untreated Hodgkin's disease: An important predictor of prognosis. Br J Cancer 1982; 45: 70-9. 11. Tullgren O, Grimfors G, Holm G et al. Lymphocyte abnormalities predicting a poor prognosis in Hodgkin's disease. Cancer 1991; 68: 768-75. 12. Zielinski C, Preis P, Aiginger P, Eibl M. Acute phase-proteins and parameters of humoral immunity in patients with advanced Hodgkin's disease. J Cancer Res C1in Oncol 1985; 110: 65-70. 13. Grimfors G, Andersson B, Tullgren O et al. Increased serum CD8 soluble antigen level is associated with blood lymphocyte abnormalities and other established indicators of poor prognosis in adult Hodgkin's disease. Br J Haematol 1992; 80: 166-71. 14. Gause A, Jung W, Schmits R et aJ. Soluble CD8, CD25 and CD30 antigens as prognostic markers in patients with untreated Hodgkin's lymphoma. Ann Oncol 1992; 3 (Suppl 4): 49-52. 15. Enblad G, Sundstrom C, Gronowitz S et al. Serum levels of interieukin-2 receptor (CD25) in patients with Hodgkin's disease, with special reference to age and prognosis. Ann Oncol 1995; 6: 65-70.
16. Axdorph U, Grimfors G, Landgren O et al. Serum levels of soluble Interleukin-2 receptor (sCD25) are increased in untreated patients with Hodgkin's disease with impaired blood lymphocyte function and a poor prognosis. Second meeting of the European Haematology Association 1996; Paris (Abstr). 17. Viviani S, Camerini E, Bonfante V et al. Soluble interleukin-2 receptors (sIL-2R) in Hodgkin's disease: Outcome and clinical implications. Br J Cancer 1998; 77: 992-7. 18. Nadali G, Tavecchia L, Zanolin E et al. Serum level of the soluble form of the CD30 molecule identifies patients with Hodgkin's disease at high risk of unfavorable outcome. Blood 1998; 91: 3011-6. 19. Christiansen I, Enblad G, Kalkner K et al. Soluble ICAM-1 in Hodgkin's disease: A promising independent predictive marker for survival. Leuk Lymph 1995; 19: 243-51. 20. Axdorph U, Grimfors G, Landgren O et al. Serum levels of sICAM-1 are correlated to tumour burden and blood lymphocyte functions in untreated Hodgkin's disease. Third International Symposium on Hodgkin's Lymphoma 1995; KSln (Abstr). 21. Sarris AH, KJiche KO, Pethambaram P et al. Interleukin-10 levels are often elevated in serum of adults with Hodgkin's disease and are associated with inferior failure-free survival. Ann Oncol 1999; 10: 433-40. 22. Dimopoulos MA, Cabanillas F, Lee J et al. Prognostic role of serum p2-microglobulin in Hodgkin's disease. J Clin Oncol 1993; 11: 1108-11. 23. Eriksson B, Hagberg H, Glimelius B et al. Serum thymidine kinase as a prognostic marker in Hodgkin's disease. Acta Radiol Oncol 1985; 24: 167-71. 24. Kurzrock R, Redman J, Cabanillas F et al. Serum interleukin-6 levels are elevated in lymphoma patients and correlate with survival in advanced Hodgkin's disease and with B symptoms. Cancer Res 1993; 53: 2118-22. 25. Warzocha K, Bienvenu J, Ribeiro P et al. Plasma levels of tumour necrosis factor and its soluble receptors correlate with clinical features and outcome of Hodgkin's disease patients. Br J Cancer 1998; 77: 2357-62. 26. Hasenclever D, Diehl V, Armitage JO et al. A prognostic score for advanced Hodgkin's Disease. N Engl J Med 1998; 339: 150614. 27. Franklin J, Paulus U, Lieberz D et al. Is the international prognostic score for advanced stage Hodgkin's disease applicable to early stage patients? Ann Oncol 2000; 11. 617-23. 28. Specht L, Hasenclever D. Prognostic factors of Hodgkin's disease. In Mauch P, Armitage J, Diehl Vet al. (eds): Hodgkin's Disease. Philadelphia, Pennsylvania: Lippincott Williams & Wilkins 1999; 295-325. 29. Harris LN. A practical approach to the pathology of lymphoid neoplasms: A revised European-American classification from the international lymphoma study group. In De Vita V, Hellman S, Rosenberg S (eds): Important Advances in Oncology. Philadelphia, Pennsylvania: J.B. Lippincott Company 1995; 111-40. 30. MacLennan KA, Bennett MH, Tu A et al. Relationship of histopathologic features to survival and relapse in nodular sclerosing Hodgkin's disease. Cancer 1989; 64: 1686-93. 31. Carbone P, Kaplan H, Musshoff K et al. Report of the committee on Hodgkin's disease staging classification. Cancer Res 1971; 31: 1860-1. 32. Bjorkholm M, Holm G, Mellstedt H et al. Prognostic factors in Hodgkin's disease. I. Analysis of histopathology, stage distribution and results of therapy. Scand J Haematol 1977; 19: 487-95. 33. Askergren J, Bjorkholm M, Holm G et al. Prognostic influence of early diagnostic splenectomy in Hodgkin's disease. A long-term follow-up. Acta Med Scand 1986; 219: 315-22. 34. Kaplan E, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Soc 1958; 53: 457-81. 35. Gehan E. A generalised Wilcoxon test for comparing arbitrarily single-censored samples. Biometrika 1965; 52: 203-23. 36. Cox DR. Regression models and life tables. J R Stat Soc 1972; B34: 187-220.
1411 37. Bohlen H, Kessler M, Sextro M et al. Poor clinical outcome of patients with Hodgkin's disease and elevated interleukin-10 serum levels. Ann Hematol 2000; 79: 110-113. 38. Viviani S, Notti P, Bonfante V et al. Elevated pretreatment serum levels of IL-10 are associated with a poor prognosis in Hodgkin's disease, the Milan Cancer Institute Experience. Med Oncol 2000; 17: 59-63. 39. Jack F, Angus B, Taylor P. Prognostic score for Hodgkin's disease. N Engl J Med 1999; 340: 1288-90 (Letter).
Received 11 May 2000; accepted 31 July 2000. Correspondence to: U. Axdorph, MD Division of Hematology Department of Medicine Karolinska Hospital 171 76 Stockholm, Sweden E-mail: [email protected]
Original article Biological markers may add to prediction of outcome achieved by the international prognostic score in Hodgkin's disease U. Axdorph,1 J. Sjoberg,1 G. Grimfors,1 O. Landgren,' A. Porwit-MacDonald2 & M. Bjorkholm1 'Division of Hematology, Department of Medicine, 2 Department of Pathology and Cytology, Karolinska Hospital and Institutet, Stockholm, Sweden
survival (CSS) in univariate analyses were: increased serum levels of IL-10, sCD30 and CRP, anaemia, low levels of Background: The International Prognostic Score (IPS) identi- albumin (P < 0.001); stage TV (P = 0.003), age 3= 45 years fies seven independent factors predicting progression-free and (P = 0.006), increased serum levels of sCD25 (P = 0.010), overall survival in advanced stage Hodgkin's disease (HD). low lymphocyte counts (P = 0.020). Serum IL-10 added progThe IPS is also applicable in limited disease. However, the IPS nostic information to that achieved by the IPS: patients with a does not identify patients with a very poor prognosis. The aim high score and increased serum IL-10 had a very poor outcome of this study was to define biological markers which may add with a five-year CSS of 38%. Patients with increased serum levels of sCD30 and a high score also had a poor outcome with to the IPS in predicting outcome. Patients and methods: One hundred forty-five patients afive-yearCSS of 54%. (> 15 years) with HD of all stages and histopathology subConclusion: Serum levels of IL-10 and sCD30 may add to groups were included. In addition to factors included in the IPS in prediction of outcome in HD, and should be validated IPS, serum levels of CRP, sCD4, sCD8, sCD25, sCD30, in large, prospective studies. sCD54, interleukin (IL)-10, |32-microglobulin and thymidine kinase were analysed. Key words: Hodgkin's disease, interleukin-10, prognosis, progResults: The strongest predictors of a poor cause-specific nostic score, sCD30. Summary
Introduction
Current therapies fail to cure about one third of patients with advanced Hodgkin's disease (HD). However, longterm remissions and cure occur even in patients who stop treatment after two to four cycles of chemotherapy [1]. This may suggest over-treatment of an undefined fraction of patients with increasing risk of late complications such as secondary malignancies, cardiac toxicity and infertility [2-4]. In early stage disease, short-term chemotherapy is today commonly added to radiotherapy due to a high relapse rate among patients treated with the latter modality alone [5-8]. The dose of radiotherapy is often reduced in these combined modality treatments with the hope of diminishing the late toxicity of irradiation [5, 6]. Thus, it is of great importance to identify additional markers, which may further characterise patients with a very dismal prognosis, who may benefit from aggressive upfront treatment, and, in addition, patients with a good prognosis, who can be given limited treatment without jeopardising outcome. A number of mainly biological markers have been reported to predict outcome in the literature: Abnormal lymphocyte function [9-11], C-reactive protein (CRP) [12], serum levels of soluble (s)CD4 (Grimfors, data to be published), sCD8 [13, 14], sCD25 [15-17], sCD30
[14, 18, Axdorph, data to be published], sCD54 [19, 20], IL-10 [21], p2-microglobulin ((32M) [22], thymidine kinase (TK) [23], IL-6 [24] and tumour necrosis factor (TNF) [25]. These biological markers were, for obvious reasons, not analysed in the International Prognostic Factors Project [26]. In that retrospective, multicenter project, 19 routinely documented demographic and clinical features were analysed at diagnosis in 1618 patients with HD receiving combination chemotherapy. Seven factors (age ^ 45 years, male sex, stage IV, albumin <40 g/1, haemoglobin (Hb) < 105 g/1, white blood cell (WBQ counts ^ 15 x 109, lymphocyte counts <0.6 x 109 or <8% of WBC counts) related to poor progression-free and overall survival were combined to give a prognostic score, the IPS. This score is also applicable for HD patients with limited disease [27]. However, neither IPS nor previously published prognostic indices [28] allow the identification of a distinct group of patients with a very poor prognosis, defined as a five-year survival rate of < 50%. The aim of this study was to identify biological markers, which may add to the IPS in predicting outcome in HD for further validation in large, prospective studies.
1406 tigations for serum markers [13, 16, 20]. The median age at diagnosis was 36 years (range 15-87 years) and the median follow-up for surviving patients was 143 months (range 44-272 months). The extent of Biopsies and methods disease was evaluated according to the Ann Arbor staging classificaThe histopathology and immunophenotyping of biopsies were reviewed tion [31] with the following investigations: chest roentgenograms, and, when necessary, complementary immunostainings for CD15, radionuclide scanning of liver and spleen (1974-1992) and bone (1975— 1992) and computed tomographic scanning (1977-1992) of the abdoCD30, CD20, LN-1, CD79a, CD3, UCHL-1 and EMA (avidin-biotinperoxidase complex (ABC) technique) were performed to confirm the men and chest, when clinically indicated. Bone marrow involvement diagnosis. The R.E.A.L. Classification [29] was applied, including the was assessed by bone marrow biopsies and aspirates. Bulky disease was division of nodular sclerosis (NS) according to the British National defined as a mediastinal mass with a diameter exceeding one third of the maximal mediastinal width or any tumour manifestation with a Lymphoma Investigation [30]. diameter of > 10 cm. Complete remission (CR) corresponded to complete regression of all palpable or histologically documented tumours and resolution of all radiographic and biochemical abnormalities due Patient characteristics to HD for a minimum of three months. In this retrospective study, 145 adults (>15 years) with previously Treatment (Table 2) untreated HD, diagnosed between 1974 and 1992, were included (Table 1). Criteria for inclusion were known results of previous invesDetails of treatment have been described previously [1, 11, 13, 32, 33]. In short, patients with limited disease (stage I-IIA) were given radioTable 1. Patient characteristics. therapy [11, 13, 32]. However, patients diagnosed 1974-1979 were included in a study regarding the value of early splenectomy. In this n (%) study, patients mainly in stage IIA-IIIA were given total nodal irradiation [33]. During this period, patients with stage IIIB-IV 145 (100) All patients disease were given MOPP or CCNU-OPP chemotherapy [33]. From Sex 1979, patients with stage IIB-IV disease received 6-8 cycles of MOPP/ Male 80 (55) ABVD chemotherapy with irradiation given to bulky disease [1]. Female 65 (45) All clinical and immunological studies were approved by the Ethics Age Committee at Karolinska Institutet and patients gave informed consent. < 45 years 103 (71) > 45 years 42 (29) Clinical stage Biological markers I 38 (26) II 46 (32) The analysed samples for each biological marker vary between 86 and III 32 (22) 123 (Table 3). Serum samples were collected before start of treatment IV 29 (20) and stored at -70 °C until use. Sera were collected prospectively, and B-symptoms were analysed on the same occasion. Serum levels of soluble markers Yes 57 (39) were determined by means of commercially available sandwich immuNo 88 (61) noassay-methods: sCD4, sCD8 [13], sCD25 [16], sCD54 [20] (CellBulky disease free® test kits, T-cell Science, Cambridge, Massachusetts), sCD30 Yes 33 (23) (Dako CD30 (K.i-1 Antigen) ELISA) and IL-10 (Quantikine human No 112 (77) IL-10, R&D Systems Europe). The minimal detectable value of IL-10 Histopathology was 7.8 pg/ml. Thus, IL-10 was dichotomized at 10 pg/ml in accordLP 12 (8) ance with Sarris et al. [21]. P2-microglobuhn (P2M) was analysed by NS unclassified 2 (1) a competitive luminometric assay (LIA-mat® p2-microglobulin, CamNS1 44 (30) bridge Life Sciences pic). Thymidine kinase (TK.) was measured in a NS2 31 (21) radio-enzyme assay (Prolifigen® TK-REA, AB Sangtec Medical). MC 50 (35) LD 1 (1) Controls Unclassified" 5 (4) The control population consisted of healthy blood donors and mem* Slides or biopsies were not available forreviewin three of thesefivecases. bers of the laboratory staff. Patients had significantly higher median
Patients and methods
Table 2. Treatment according to clinical stage. Treatment
Total number oi patients
Number of patients Stage I (n = 38)
Stage II (n = 46)
Stage III (n = 32)
Stage IV (/i = 29)
Radiotherapy Total nodal irradiation Subtotal nodal irradiation Mantle field, inverted Yfield,involved field
72 21 17 34
3 9 23
10 7 11
8 1 0
0 0 0
Chemotherapy MOPP or CCNU-OPP MOPP/ABVD
56 10 46
1 1
1 7
6 15
2 23
Combined modality treatment MOPP/ABVD and radiotherapy
17 17
1
10
2
4
1407 Table 3. Patients with biological markers analysed in percentage of all patients (/i = 145) in the study. Biological marker
No. patients analysed (%) patii
IL-10 sCD4 sCD8 sCD25 sCD30 sCD54 CRP
123 (85) 99(68) 86 (59) 113 (78) 93(64) 115(79) 99 (68) 99 (68) 97 (67)
P2M TK
Cause-specific Survival by Prognostic Score
100
Q.
I
O
8 values than controls regarding sCD25 (P < 0.001,17 controls), sCD30 (P < 0.001,43 controls), f32M (/> < 0.001,79 controls),TK (P < 0.001, 160 controls), sCD54 (P = 0.054, 20 controls). The patient population did not statistically differ from controls with regard to IL-10 (8 controls), sCD4 (10 controls) and sCD8 (20 controls). CRP, Hb, WBC and total lymphocyte counts and albumin were determined according to standard methods.
Statistical methods The Mann-Whitney U-test was used to calculate differences in distributions of serum markers between groups. All laboratory variables were initially analysed as continuous except IL-10, which was analysed as dichotomized only. In congruence with the IPS, Hb, WBC and total lymphocyte counts, albumin, age and stage were also dichotomized. Cut-off points were chosen according to the literature [21, 26]. When combining sCD30 with the IPS, sCD30 was dichotomized. The cutoff point for sCD30 was chosen as the median value. Data skewly distributed were logarithmically transformed. Linear regression and logistic regression analyses were used when analysing the association between year of diagnosis with the biological variables. Cause-specific survival (CSS) was defined as the time from diagnosis to death from HD or death related to treatment of the disease including second malignancies. Observations were censored by end of follow-up or death without signs of HD. In complete remitters, disease-free survival (DFS) was defined as the time from last day of treatment to relapse or death from HD or death related to treatment of the disease including second malignancies. Observations were censored by end of follow-up or death without signs of HD. Survival curves were constructed by the Kaplan-Meier procedure [34]. Differences in the survival for categorical factors were calculated by Gehan's Wilcoxon test [35]. Cox regression analyses were used for continuous factors univariately and for the multivariate analyses [36]. In the latter analysis, only markers, significant in univariate analyses, with less than 15% missing values, were analysed. Thus, serum levels of sCD25, sCD30 and CRP were not analysed with respect to their independent significance.
Results At follow-up, 57 patients had died. Forty patients died from progressive HD or from complications to the treatment including three patients, who developed myelodysplastic syndrome/acute leukaemia. The year of diagnosis was not associated with CSS or DFS (data not shown). The CR rate was 86%. Thirty-eight of the one hundred twenty-five patients (30%) who achieved a CR relapsed during the observation period. In the whole series, the five-year CSS was 82%. Among patients achieving a CR, the five-year DFS was 69%.
10
12
14
16
18
20
Years after diagnosis Figure 1. Cause-specific survival. IPS divided in a high-risk group (3-7 factors, 47 patients) and a low-risk group (0-2 factors, 96 patients). P < 0.001, Gehan's test, df = 1.
Cause-specific survival Evaluation of the International Prognostic Score Low serum albumin (P < 0.001), anaemia (P < 0.001), stage IV (P = 0.003), age ^ 45 years (P = 0.006), low lymphocyte counts (P — 0.020) but not male sex and increased WBC counts predicted a poor CSS in univariate analyses. In the multivariate analysis of IPS factors reaching significance in univariate analyses, albumin (P = 0.007), Hb (P = 0.026) and stage (P = 0.041) retained significance with a borderline value for age (P - 0.053). In the multivariate analysis of IPS factors, 143 patients were analysed. Two patients had missing data regarding albumin and total lymphocyte count, respectively. High-risk patients (score 3-7) had a significantly worse survival as compared to low-risk patients (score 0-2, P < 0.001, Figure 1). Biological markers Increased serum levels of sCD30 (P < 0.001), CRP (P < 0.001) and sCD25 (P - 0.010) as well as serum IL-10 ^ 10 pg/ml (P < 0.001) predicted a poor CSS in univariate analyses. The predicted prognostic capacity of serum levels of sCD4, sCD8, sCD54, p"2M and TK did not reach statistical significance. We did as well perform survival analyses with control for year of diagnosing. Controlling for this factor did not change the results (data not shown). In the multivariate analysis including significant factors from the IPS, IL-10 retained an independent prognostic significance (Table 4). For IL-10, controlling for Hb and albumin, the hazards were not different for the two values considered as judged by log minus log plots (data not shown). Serum IL-10, sCD30 and the International Prognostic Factor Score Patients were subgrouped according to IPS and dichotomized serum IL-10 values (P < 0.001, Figure 2). IL-10
1408 Prognostic Score and sCD30
Prognostic Score and IL-10 100
Score 0-2 4ILKK10 I
1 60
CO
40
Score 0-2 & sCD30<40
1
80 3 in
100
Score 0-2 &IL1>10
Score 3-7 &ILK10 o
a.
I
Q. in
| Score 3-7 & IL1>10
i,
at
20
CD
o
0 0
2
4
6
8
10
12
14
16
18
20
0
Years after diagnosis
score 0-2 and score 0-2 and score 3-7 and score 3-7 and
IL-10 <10 pg/mL: IL-10 >10 p g / raL IL-10 <10 pg/mL: IL-10 >10 pg/mL:
4
6
8
10
12
14
16
18 20
Years after diagnosis
No of patients
Five-year survival
Ten-year survival
80 7 26 8
91% 86%
88% 43% 52% 25%
69% 38%
2
score 0-2 and sCD30 <40 U/mL: score 0-2 and sCD30 >40 U/mL score 3-7 and sCD30 <40 U/ mLscore 3-7 and sCD30 >40 U/mL
No of patients
Five-year survival
Ten-year survival
38 22 9 24
97% 77% 89% 54%
91% 77% 89% 30%
Figure 2. Cause-specific survival. IPS and serum IL-10. P < 0.001, Gehan's test for heterogeneity. x2-statistic = 26.0, df = 1.
Figure 3 Cause-specific survival. IPS and serum sCD30. P < 0.001, Gehan's test for heterogeneity. z2-statistic = 26.4, df = 3.
added prognostic information to that achieved by the IPS: patients with a high score and increased IL-10 levels had a very poor outcome with a five-year CSS of 38%, while patients with a low score and low IL-10 levels had a five-year CSS of 91%. Atfiveyears, the CSS did nor differ in patients with low score and high and low IL-10 levels, respectively. However, the 10-year CSS was 43% for patients with a low score and increased IL-10 levels as compared to 88% for patients with a low score and low IL-10 levels. Subgrouping according to IPS and dichotomized serum sCD30 values showed a five-year CSS for patients with a high score and increased sCD30 levels of 54%, being significantly worse than that of other subgroups (P < 0.001, Figure 3).
Disease-free survival
Table 4. Cox multivariate (forward stepwise) regression with respect to poor cause-specific survival. Variable
Relative hazards
95% Confidence interval
Evaluation of the International Prognostic Factor Score Increased WBC counts (P - 0.014) predicted a poor DFS in univariate analysis. Low levels of albumin, anaemia, stage IV, age ^45 years, low lymphocyte counts or male sex did not reach statistical significance. Biological markers Increased serum levels of sCD25 (P = 0.009), sCD30 (P = 0.028), CRP (P = 0.034) and sCD8 (P = 0.051) predicted a poor DFS in univariate analyses, as well as serum IL-10 ^ 10 pg/ml {P - 0.010). The prognostic discriminatory capacity of serum levels of sCD4, sCD54, P2M, TK did not reach statistical significance. Serum IL-10 had an independent impact on DFS in the multivariate analysis (P = 0.007) with the WBC count of the IPS as a competing factor. Discussion
P
In spite of considering established prognostic factors such as age, stage, haemoglobin values and others, there is a great variability in outcome of patients with ad3.33 Serum IL-10 5= lOpg/ml 1.56-7.08 0.002 vanced stage HD. The search for additional prognostic 2.33 Haemoglobin < 105 g/1 0.035 1.06-5.0 factors has been intensive for many years. The aim has 2.56 Albumin <40 g/l 0.037 1.05-6.25 been to identify patients with a very poor prognosis, Competing factors from the IPS were: stage IV, age > 45 years, low who may benefit from aggressive upfront treatment. lymphocyte counts. Also, to find patients, who can be treated less aggresIn this multivariate analysis of IPS factors and IL-10,122 patients were sively without compromising the anti-tumour effect, but analysed; 1 of the 123 patients with values of IL-10 had missing data with less acute and late toxic complications. Recently, regarding albumin. All competing factors had less than 15% missing values. Sarris et al. reported that serum levels of IL-10 was a
1409 strong biological prognostic factor [21], and other groups have described the same results [37, 38]. In limited disease, high relapse rates after radiotherapy alone have led to the use of a combination of chemotherapy and radiotherapy in many centres. It is likely, that this approach will lead to over-treatment of some patients, increasing the incidence of late complications. Thus, apt predictors of prognosis are of equal importance in patients with limited disease. Previously known prognostic factors have been included in different prognostic indices for survival [28], but these indices have failed to identify patients at very high risk. An important work was the IPS study, where seven factors related to progression-free and overall survival in HD patients receiving combination chemotherapy were identified [26]. Interestingly, some of the previously known variables, among them extranodal disease, histologic type, erythrocyte sedimentation rate, were found not to add prognostic information in that study. This may in part be due to an improved success rate during recent years. The score predicted five-year rates of freedom from progression of disease in the range of 42% to 84%, and 56% to 89% in overall survival. Thus, even this prognostic score could not identify a subgroup of patients with a very poor prognosis (afive-yearsurvival < 50%) and a potential need of more aggressive upfront treatment. Despite the limited number of patients included in the present study, the results were in a good accordance with those of the IPS study: five of the seven factors significantly predicted a poor CSS in univariate analyses, and four factors remained in the multivariate analysis. A very good prognostic discriminatory capacity was also seen, when the patients where divided into a low-risk and a high-risk group (score 0-2 and 3-7, respectively). Fifty-seven patients (38%) in our study had limited disease (stage I-IIA, treated with radiotherapy alone). In comparison, the inclusion criterion in the international study was advanced disease, treated with chemotherapy ± radiotherapy. However, Franklin et al. has reported that the IPS also is applicable for patients with limited disease [27]. We analysed CSS as end point instead of freedom from progression, which was the main end point reported in the IPS study. In agreement with Jack et al. [39], who also evaluated the score with CSS, we find CSS to constitute a definitive measure of failure. In addition, we also analysed prognostic factors in relation to DFS. An important factor to consider with regard to DFS is the greater risk of relapse following radiotherapy alone [6]. However, we found no difference in time to relapse or disease-free survival among CR patients according to initial treatment (radiotherapy, chemotherapy or combined modality treatment; data not shown). In the IPS study, analyses were restricted to patients between 15 and 65 years. In the search for prognostic variables, that will help the clinician in individualising treatment, we find it important not to exclude older patients. Optimising treatment in older patients with intercurrent diseases constitutes a major
therapeutic challenge. To avoid misinterpretation of survival because of intercurrent diseases, we analysed cause-specific survival. In conclusion: in spite of a restricted number of patients, inclusion of patients of all ages and stages and the use of other end points, this patient population showed a good concordance with the results of the IPS study. A useful prognostic marker has to be 1) easy to analyse and 2) show reliable reproducibility, and of course, be a strong predictor of prognosis, identifiable also in smaller groups of patients. In a series of previous reports we have evaluated biological markers relevant for HD with the goal to refine our ability to predict prognosis [9-11, 13, 16, 20]. There are certain other variables, for example IL-6 and TNF/TNF-receptors, that would also be of interest to evaluate in association to IPS [24, 25]. However, this is already a rather variable intense study, and we had to limit the number of variables. Partly due to the same reason, we did not include blood lymphocyte function tests, although these have in previous studies been shown to be a strong predictor of prognosis [9-11]. In addition, tests of lymphocyte function are rather complex, and their reproducibility may be questioned. In order to address the issue whether cytokine levels remain stable during storage at -70 °C over several years, we tested the association between year of diagnosis with CRP, IL-10 and with the log values of sCD4, sCD8, sCD25, sCD30, sCD54, P2M and TK. sCD4, sCD54, 02M and TK showed positive association with calendar time on 5% risk level. As a consequence of this, we repeated all survival analyses with control for year of diagnosis. However, the overall results where not changed (data not shown). The univariate analyses regarding the significant correlations of a poor CSS and increased serum levels of IL-10, sCD30, sCD25 and CRP confirmed results of previous studies concerning biological markers and prognosis [12, 14-16, 18, 21, 37, 38]. In the multivariate analysis, IL-10 emerged as an independent variable, which is also in agreement with previous studies [21, 37, 38]. IL-10 seems to be helpful in adding prognostic information to that achieved by the IPS: patients with a high score and increased IL-10 had a very poor outcome, patients with a low score and low levels of IL-10 had a good prognosis, and patients with the combination of a low score and high IL-10 as well as a high score and low IL-10 showed an intermediate pattern. In addition, a high serum level of IL-10 also predicted a poor DFS. Patients with the combination of high serum levels of sCD30 and a high score according to IPS also had a poor prognosis. Serum levels of sCD25, sCD30 and CRP were not analysed with respect to their independent significance because of > 15% missing data. We conclude that IL-10 and sCD30 may add to prediction of outcome in HD, and thereby be helpful in identifying high- and low-risk patients with the goal to optimise treatment. Hopefully these results will encourage others to measure biological factors such as IL-10
1410 and sCD30 in order to allow a future multivariate analysis in a substantial number of patients. Acknowledgements This study was supported by grants from the Swedish Cancer Society, the Stockholm County Council and Karolinska Institutet Foundations. We thank Professor Derek Crowther for critical review of the manuscript. References 1. BjSrkholm M, Axdorph U, Grimfors G et al. Fixed versus response adapted MOPP/ABVD chemotherapy in Hodgkin's disease. Ann Oncol 1995; 6: 895-9. 2. Leyiand-Jones B, O'Dwyer P, Alonso T. Late effects of chemotherapeutic agents. In Lacher MJ, Redman JR (eds): Hodgkin's Disease: The Consequences of Survival. Philadelphia, Pennsylvania: Lea & Febiger 1990; 47-62. 3. Van Leeuwen FE, Klokman WJ, Veer MB et al. Long-term risk of second malignancy in survivors of Hodgkin's disease treated during adolescence or young adulthood. J Clin Oncol 2000; 18: 487-97 4. Swerdlow AJ, Barber JA, Hudson GV et al. Risk of second malignancy after Hodgkin's disease in a collaborative British cohort: The relation to age at treatment. J Clin Oncol 2000; 18: 498-509. 5. Noordijk E. Radiotherapy in early stage Hodgkin's disease. Principles and results of recent clinical trials. Ann Oncol 1998; 9 (Suppl 5): 63-5. 6. Cosset JM, Mauch PM. The role of radiotherapy for early stage Hodgkin's disease: Limitations and perspectives. Ann Oncol 1998; 9 (Suppl 5): 57-62. 7. Vlachaki M, Hagemeister F, Fuller L et al. Long-term outcome of treatment for Ann Arbor stage I Hodgkin's disease: Prognostic factors for survival and freedom from progression. Int J Radiat Oncol Biol Phys 1997; 38: 593-9. 8. Glimelius B, Kalkner M, Enblad G et al. Treatment of early and intermediate stages of supradiaphragmatic Hodgkin's disease: The Swedish National Care Program experience. Ann Oncol 1994; 5: 809-16. 9. BjSrkholm M, Holm G, Mellstedt H et al. Prognostic factors in Hodgkin's disease. II. Role of the lymphocyte defect. Scand J Haematol 1978; 20: 306-18. 10. Wedelin C, Bjorkholm M, Holm G et al. Lymphocyte function in untreated Hodgkin's disease: An important predictor of prognosis. Br J Cancer 1982; 45: 70-9. 11. Tullgren O, Grimfors G, Holm G et al. Lymphocyte abnormalities predicting a poor prognosis in Hodgkin's disease. Cancer 1991; 68: 768-75. 12. Zielinski C, Preis P, Aiginger P, Eibl M. Acute phase-proteins and parameters of humoral immunity in patients with advanced Hodgkin's disease. J Cancer Res C1in Oncol 1985; 110: 65-70. 13. Grimfors G, Andersson B, Tullgren O et al. Increased serum CD8 soluble antigen level is associated with blood lymphocyte abnormalities and other established indicators of poor prognosis in adult Hodgkin's disease. Br J Haematol 1992; 80: 166-71. 14. Gause A, Jung W, Schmits R et aJ. Soluble CD8, CD25 and CD30 antigens as prognostic markers in patients with untreated Hodgkin's lymphoma. Ann Oncol 1992; 3 (Suppl 4): 49-52. 15. Enblad G, Sundstrom C, Gronowitz S et al. Serum levels of interieukin-2 receptor (CD25) in patients with Hodgkin's disease, with special reference to age and prognosis. Ann Oncol 1995; 6: 65-70.
16. Axdorph U, Grimfors G, Landgren O et al. Serum levels of soluble Interleukin-2 receptor (sCD25) are increased in untreated patients with Hodgkin's disease with impaired blood lymphocyte function and a poor prognosis. Second meeting of the European Haematology Association 1996; Paris (Abstr). 17. Viviani S, Camerini E, Bonfante V et al. Soluble interleukin-2 receptors (sIL-2R) in Hodgkin's disease: Outcome and clinical implications. Br J Cancer 1998; 77: 992-7. 18. Nadali G, Tavecchia L, Zanolin E et al. Serum level of the soluble form of the CD30 molecule identifies patients with Hodgkin's disease at high risk of unfavorable outcome. Blood 1998; 91: 3011-6. 19. Christiansen I, Enblad G, Kalkner K et al. Soluble ICAM-1 in Hodgkin's disease: A promising independent predictive marker for survival. Leuk Lymph 1995; 19: 243-51. 20. Axdorph U, Grimfors G, Landgren O et al. Serum levels of sICAM-1 are correlated to tumour burden and blood lymphocyte functions in untreated Hodgkin's disease. Third International Symposium on Hodgkin's Lymphoma 1995; KSln (Abstr). 21. Sarris AH, KJiche KO, Pethambaram P et al. Interleukin-10 levels are often elevated in serum of adults with Hodgkin's disease and are associated with inferior failure-free survival. Ann Oncol 1999; 10: 433-40. 22. Dimopoulos MA, Cabanillas F, Lee J et al. Prognostic role of serum p2-microglobulin in Hodgkin's disease. J Clin Oncol 1993; 11: 1108-11. 23. Eriksson B, Hagberg H, Glimelius B et al. Serum thymidine kinase as a prognostic marker in Hodgkin's disease. Acta Radiol Oncol 1985; 24: 167-71. 24. Kurzrock R, Redman J, Cabanillas F et al. Serum interleukin-6 levels are elevated in lymphoma patients and correlate with survival in advanced Hodgkin's disease and with B symptoms. Cancer Res 1993; 53: 2118-22. 25. Warzocha K, Bienvenu J, Ribeiro P et al. Plasma levels of tumour necrosis factor and its soluble receptors correlate with clinical features and outcome of Hodgkin's disease patients. Br J Cancer 1998; 77: 2357-62. 26. Hasenclever D, Diehl V, Armitage JO et al. A prognostic score for advanced Hodgkin's Disease. N Engl J Med 1998; 339: 150614. 27. Franklin J, Paulus U, Lieberz D et al. Is the international prognostic score for advanced stage Hodgkin's disease applicable to early stage patients? Ann Oncol 2000; 11. 617-23. 28. Specht L, Hasenclever D. Prognostic factors of Hodgkin's disease. In Mauch P, Armitage J, Diehl Vet al. (eds): Hodgkin's Disease. Philadelphia, Pennsylvania: Lippincott Williams & Wilkins 1999; 295-325. 29. Harris LN. A practical approach to the pathology of lymphoid neoplasms: A revised European-American classification from the international lymphoma study group. In De Vita V, Hellman S, Rosenberg S (eds): Important Advances in Oncology. Philadelphia, Pennsylvania: J.B. Lippincott Company 1995; 111-40. 30. MacLennan KA, Bennett MH, Tu A et al. Relationship of histopathologic features to survival and relapse in nodular sclerosing Hodgkin's disease. Cancer 1989; 64: 1686-93. 31. Carbone P, Kaplan H, Musshoff K et al. Report of the committee on Hodgkin's disease staging classification. Cancer Res 1971; 31: 1860-1. 32. Bjorkholm M, Holm G, Mellstedt H et al. Prognostic factors in Hodgkin's disease. I. Analysis of histopathology, stage distribution and results of therapy. Scand J Haematol 1977; 19: 487-95. 33. Askergren J, Bjorkholm M, Holm G et al. Prognostic influence of early diagnostic splenectomy in Hodgkin's disease. A long-term follow-up. Acta Med Scand 1986; 219: 315-22. 34. Kaplan E, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Soc 1958; 53: 457-81. 35. Gehan E. A generalised Wilcoxon test for comparing arbitrarily single-censored samples. Biometrika 1965; 52: 203-23. 36. Cox DR. Regression models and life tables. J R Stat Soc 1972; B34: 187-220.
1411 37. Bohlen H, Kessler M, Sextro M et al. Poor clinical outcome of patients with Hodgkin's disease and elevated interleukin-10 serum levels. Ann Hematol 2000; 79: 110-113. 38. Viviani S, Notti P, Bonfante V et al. Elevated pretreatment serum levels of IL-10 are associated with a poor prognosis in Hodgkin's disease, the Milan Cancer Institute Experience. Med Oncol 2000; 17: 59-63. 39. Jack F, Angus B, Taylor P. Prognostic score for Hodgkin's disease. N Engl J Med 1999; 340: 1288-90 (Letter).
Received 11 May 2000; accepted 31 July 2000. Correspondence to: U. Axdorph, MD Division of Hematology Department of Medicine Karolinska Hospital 171 76 Stockholm, Sweden E-mail: [email protected]