- Email: [email protected]
The prognostic impact of tumour-associated macrophages and Reed-Sternberg cells in paediatric Hodgkin lymphoma
European Journal of Cancer (2013) xxx, xxx– xxx
Available at www.sciencedirect.com
journal homepage: www.ejcancer.com
The prognostic impact of tumour-associated macrophages and Reed-Sternberg cells in paediatric Hodgkin lymphoma q Sumit Gupta a,⇑, Stacy Yeh b, Rose Chami c, Angela Punnett a, Catherine Chung d a
Division of Hematology/Oncology, Hospital for Sick Children, 555 University Avenue, Toronto, Ontario, Canada M5G 1X8 Faculty of Medicine, University of Toronto, 1 Kings College Circle, Toronto, Ontario, Canada M5S 1A8 c Department of Pathology, CHU Sainte-Justine, 3175 Chemin de la Coˆte-Sainte-Catherine, Montreal, Quebec, Canada H3T 1C5 d Division of Pathology, Hospital for Sick Children, 555 University Avenue, Toronto, Ontario, Canada M5G 1X8 b
KEYWORDS Child Hodgkin lymphoma Macrophages Prognosis Reed-Sternberg cells
Abstract Background: Tumour-associated macrophages (TAM) are associated with treatment failure in adults with Hodgkin lymphoma (HL). Equivalent data in paediatric HL are sparse. We aimed to determine the prognostic significance of TAM and Reed-Sternberg (RS) cells in paediatric HL. Methods: All children aged 0–18 with HL between 1980 and 2009 with available diagnostic biopsy material were identified. A treatment failure-enriched cohort was assembled. Demographic, disease and outcome data were abstracted. Tissue microarrays with duplicate cores were constructed from diagnostic biopsy material and stained with immunohistochemical markers for TAM (CD68, CD163) and RS (CD30). A high score was defined as >5% positive cells relative to overall cellularity in any core. The association of candidate variables with event-free survival (EFS) was determined using Cox proportional hazards. Results: The final study cohort comprised 96 patients with a median age of 14 years (interquartile range 11–15). Agreement on scores between cores from the same biopsy revealed weighted kappas of 0.60, 0.68 and 0.73 for CD30, CD68 and CD163 respectively, indicating moderate tumour heterogeneity. In univariate analysis, a high CD30 score was significantly associated with treatment failure (hazard ratio (HR) 2.27; 95th confidence interval 1.01– 5.11; p < 0.05). High CD68 and CD163 scores were not associated with EFS. Conclusions: Unlike adult HL, a higher percentage of RS cells was associated with poor outcome, while a higher percentage of TAM was not. Adult HL findings may not extend to paediatric HL. Cooperative group trials of paediatric HL should prospectively determine the association of different components of the tumour microenvironment with outcome. Ó 2013 Elsevier Ltd. All rights reserved.
q Funding sources: This study was supported through an operating Grant from the Physicians’ Services Incorporated Foundation. SG was supported through a doctoral fellowship award from the Canadian Institutes of Health Research. ⇑ Corresponding author: Tel.: +1 416 813 7742. E-mail address: [email protected] (S. Gupta).
0959-8049/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ejca.2013.05.024
Please cite this article in press as: Gupta S. et al., The prognostic impact of tumour-associated macrophages and Reed-Sternberg cells in paediatric Hodgkin lymphoma, Eur J Cancer (2013), http://dx.doi.org/10.1016/j.ejca.2013.05.024
2
S. Gupta et al. / European Journal of Cancer xxx (2013) xxx–xxx
1. Introduction Hodgkin lymphoma (HL) is commonly encountered by paediatric oncologists.1 Multimodality treatment with chemotherapy and radiation has led to cure rates in excess of 80%, but at the cost of significant therapyrelated late effects including infertility, cardiac and respiratory dysfunction and second malignancies.2,3,4 Novel prognosticators available at the time of diagnosis may have significant impact by identifying children in whom treatment de-intensification is warranted. Alternatively, such prognosticators could also identify the children who are likely to fail treatment, allowing for upfront intensification. Recently, several studies have examined whether various aspects of the tumour microenvironment may play such a role. Uniquely among malignancies, the malignant cells in HL, Reed-Sternberg (RS) cells, can comprise less than 1% of the tumour cell population. The composition of the remaining 99%, called the tumour microenvironment, is determined by the complex expression of various cytokines and chemokines by both RS and reactive cells.5 The most promising potential prognosticator seems to be the presence of tumourassociated macrophages (TAM), which multiple investigators have found to be associated with inferior survival among adults with HL.6,7,8 As of yet, equivalent data for paediatric HL patients are sparse. The prognostic value of the burden of RS cells is relatively unexplored in both populations. Our primary objective was therefore to determine the prognostic significance of the percentage of both TAM and RS cells using the immunohistochemical markers CD68 and CD163, and CD30, respectively, in diagnostic biopsy specimens among children with HL. 2. Methods 2.1. Study population The study followed a retrospective cohort design. In order to increase statistical power, a cohort enriched with cases of treatment failure was assembled. This technique has been used in past studies of tumour microenvironment.6 All patients aged 0–18 years and diagnosed at The Hospital for Sick Children between 1980 and 2009 were identified using a local electronic patient database. Patients were excluded for the following reasons: (1) nodular lymphocyte predominant Hodgkin lymphoma (NLPHL), (2) post transplant lymphoproliferative disease, (3) prior immunodeficiency, (4) synchronous additional malignant diagnoses and (5) initial diagnostic biopsy material insufficient or unavailable for immunohistochemical studies. In cases where the initial biopsy was performed at an outside hospital, every effort was made to obtain the original biopsy material.
Using the same electronic database, all patients experiencing disease progression, relapse or death were identified and were added to the study cohort. The remaining patients (i.e. those who experienced long-term progression-free and relapse-free survival) were then identified. Patients from this remaining population were then randomly selected using a computer-generated random number sequence generator and added to the study cohort until a total sample size of 100 patients was reached; this total was determined a priori based on feasibility and study resources. 2.2. Immunohistochemistry Tissue microarrays (TMAs) were constructed from duplicate 1.0 mm cores from formalin-fixed paraffinembedded blocks of diagnostic biopsy tissue. Cores were obtained from representative areas containing RS cells as evaluated by morphology and immunohistochemical stains that were performed for routine diagnosis. A total of 22 samples had insufficient material to be included in the TMA but sufficient material for whole section immunohistochemical staining. These 22 patients were retained in the study cohort, with immunohistochemical staining performed on whole sections. Three arrays were built from the remaining 78 cases in duplicate with cases arranged randomly across the TMAs. Both four micron sections from the TMAs and whole sections were cut and stained with haematoxylin and eosin (H&E) and the immunohistochemical markers CD68 (predilute, clone KP1, Ventana), CD163 (1:100, clone 10D6, Leica) and CD30 (1:50, clone BerH2, Dako Canada). The immunohistochemical stains were performed on an automated stainer, Ventana Benchmark XT (Ventana Medical Systems, Tucson, Arizona), using a multimer detection kit (ultraView Universal DAB, Ventana Medical Systems, Tucson, Arizona). Two paediatric pathologists (C.C. and R.C.) independently scored the TMA and whole section slides. Two areas of whole section slides were scored. Both pathologists were blinded to patients’ clinical characteristics and treatment outcomes. Immunoreactivity was scored based on the percentage of cells showing cytoplasmic (CD68, CD163) and membranous and golgi (CD30) staining, relative to overall cellularity as follows: 1 = <5%, 2 = 5–25%, 3 = 26–50%, 4 = >50%. Discrepancies between the two reviewers were discussed until consensus was reached. 2.3. Data abstraction The charts of patients were obtained and data relating to patient characteristics, disease, treatment and outcome were abstracted. Patients were classified as belonging to the early time period if diagnosed between 1980 and 2000, and to the late period if diagnosed
Please cite this article in press as: Gupta S. et al., The prognostic impact of tumour-associated macrophages and Reed-Sternberg cells in paediatric Hodgkin lymphoma, Eur J Cancer (2013), http://dx.doi.org/10.1016/j.ejca.2013.05.024
S. Gupta et al. / European Journal of Cancer xxx (2013) xxx–xxx
between 2001 and 2009. Risk groups were defined according to the current Children’s Oncology Group classification: low (stages IA and IIA), intermediate (stages IB, IIB, IIIA and IVA) and high (stages IIIB and IVB). Various cutoffs for immunoreactivity have been used in the HL microenvironment literature.6,9,7 Based on the distribution of immunoreactivity scores between patients, a cutoff of greater or equal to 5% positive cells relative to overall cellularity was chosen for each immunohistochemical marker. Upon the discovery of moderate heterogeneity between cores from the same diagnostic biopsy (see below), a cutoff of 5% in any core was deemed sufficient to be considered positive. These decisions were made prior to any outcome analyses. 2.4. Statistical analysis The distribution of immunoreactivity was compared between risk groups for each immunohistochemical marker using the Chi square test or Fisher’s exact test as appropriate. The agreement between cores from the same diagnostic biopsy, and between CD68 and CD163 immunoreactivity on the same core, were both determined using weighted kappa statistics. The primary outcome was event-free survival (EFS), described using the Kaplan–Meier method and defined as the time from diagnosis to disease progression, relapse or death. Patients were censored at the time of last follow-up. Univariate and multivariable Cox proportional hazards models were used to explore predictors of EFS. The proportional hazards assumption was examined and verified. Variables with a P value < 0.05 on univariate analysis were examined in the multiple regression models. Statistical analyses were performed using SAS-PC software (version 9.2; SAS Institute, Cary, NC). Statistical significance was defined as P < 0.05. The study was approved by the institutional research ethics board, which also waived the need for written informed consent. 3. Results A total of 361 patients were identified as being diagnosed with HL during the study period. A total of 81 patients were excluded, leaving 280 patients in the final study cohort. Reasons for exclusion included insufficient biopsy material available for immunohistochemical staining (n = 39), NLPHL (n = 37), PTLD (n = 2), immunodeficiency (n = 2) and simultaneous diagnosis of non-Hodgkin lymphoma (n = 1). Of the final study cohort, 35 experienced progressive disease, relapse or death and met the inclusion criteria. Only one patient experienced death as a first event; sensitivity analyses excluding this patient did not significantly change any of the results (data not shown).
3
Table 1 Characteristics of study cohort patients. N (%) Gender Male Female
Median (IQR)
50 (52.1) 46 (47.9)
Age (years)
14 (11–15)
Stage I II III IV
5 (5.3) 45 (47.4) 15 (15.8) 30 (31.6)
B-symptoms Absent Present
48 (50.5%) 47 (49.5%)
Pathology Nodular sclerosis Mixed cellularity Lymphocyte deplete
87 (91.6) 5 (5.3) 3 (3.2)
Time period Early (1980–2000) Late (2001–2009)
41 (42.7) 55 (57.3)
Risk groupa Low Intermediate High
29 (30.9) 37 (39.4) 28 (29.8)
Radiation received Yes No
73 (77.7) 37 (22.3)
IQR, interquartile range; N, number. a See text for risk group definitions.
Table 2 Average percentage of CD30, CD68 and CD163 positive cells relative to total cellularity in diagnostic biopsy samples of study cohort patients. N (%) CD30 <5% 5 to <25% 25 to <50% >50%
53 (55.2) 37 (38.5) 6 (6.3) 0 (0.0)
CD68 <5% 5 to <25% 25 to <50% >50%
22 (22.9) 63 (65.5) 10 (10.4) 1 (1.0)
CD163 <5% 5 to <25% 25 to <50% >50%
57 (59.4) 30 (31.3) 9 (9,4) 0 (0.0)
IQR, interquartile range; N, number.
Another 65 patients, who experienced long-term progression-free and relapse-free survival, were randomly selected from the remaining patients and added to the cohort, bringing the total cohort sample size to 100 as a priori specified. The charts of four cohort patients
Please cite this article in press as: Gupta S. et al., The prognostic impact of tumour-associated macrophages and Reed-Sternberg cells in paediatric Hodgkin lymphoma, Eur J Cancer (2013), http://dx.doi.org/10.1016/j.ejca.2013.05.024
4
S. Gupta et al. / European Journal of Cancer xxx (2013) xxx–xxx
could not be located. As the TMAs had already been constructed, the final study cohort comprised of 96 children, the characteristics of which are shown in Table 1. A higher percentage of patients with nodular sclerosing histology was seen than expected, likely secondary to the treatment-failure enriched nature of the cohort; younger children with mixed cellularity histology have been suggested to have a superior prognosis.1 Fewer patient experiencing an event had received prior radiation than had randomly selected patients without an event (19/31– 61.3% versus 54/65–85.7%; p = 0.01). The percentages of CD30, CD68 and CD163 positive cells relative to total cellularity, as an average of both duplicate cores, are illustrated in Table 2. The weighted kappa statistic for agreement between cores from the same patient biopsy for CD30, CD68 and CD163 was 0.60 (95% confidence interval (CI) 0.43–0.77), 0.68 (95%CI 0.52–0.84) and 0.73 (0.61–0.84), respectively. The agreement between CD68 and CD163 positivity in
Table 3 Univariate predictors of event-free survival among study cohort patients. HR
95% CI
P value
Gender Male Female
– 1.13
– 0.55–2.32
– 0.74
Age (per year)
1.11
0.96–1.25
0.16
Risk group Low Intermediate High
– 1.92 2.05
– 0.73–5.04 0.76–5.54
– 0.19 0.16
Time period Early (1980–2000) Late (2001–2009)
– 1.28
– 0.61–2.70
– 0.51
CD30 All cores <5% positive cellularity Any core P5% positive cellularity
– 2.27
– 1.01–5.11
– 0.048
CD68 All cores <5% positive cellularity Any core P5% positive cellularity
– 1.09
– 0.42–2.85
– 0.86
CD163 All cores <5% positive cellularity Any core P5% positive cellularity
– 1.53
– 0.73–3.17
– 0.26
CI, confidence interval; HR, hazard ratio.
the same sample was 0.49 (0.35–0.63). An example of within-tumour heterogeneity is demonstrated in Fig. 1. Immunoreactivity was not significantly associated with disease risk group for any of the immunohistochemical markers examined (p = 0.17, 0.56 and 0.30 for CD30, CD68 and CD163, respectively). The 5-year EFS of the entire cohort was 65.0%, reflecting its enrichment with cases of treatment failure. Univariate predictors of EFS among the study cohort are shown in Table 3. Only CD30 positivity (immunoreactivity greater than or equal to 5% in any core) was significantly associated with worse outcome (hazard ratio (HR) 2.27, 95%CI 1.01–5.11; p = 0.048). Neither CD68 nor CD163 positivity was significantly associated with EFS. Sensitivity analyses of CD68 and CD163 positivity using a higher cutoff of 25% remained not significantly predictive of EFS (data not shown). Though only CD30 positivity reached statistical significance in univariate analyses, a multivariable model with CD30 immunoreactivity and risk group (itself incorporating stage and B symptoms) was constructed. In this model, the trend towards worse outcome with CD30 positivity was maintained (HR 2.19, 95%CI 0.9– 4.98; p = 0.06), though the threshold of statistical significance was not reached. Fig. 1. Example of intra-tumour heterogeneity in CD163 immunoreactivity. Panels (a) and (b) represent two different cores from the same biopsy sample taken from a patient with Hodgkin lymphoma, nodular sclerosis subtype. The percentage of CD163 positive macrophages is much higher in panel (a) compared to panel (b). Both images captured with 20 objective magnification (200 total magnification).
4. Discussion This study is one of the first to look at the prognostic significance of either TAM or RS cells among children with HL. We found that the impact of both
Please cite this article in press as: Gupta S. et al., The prognostic impact of tumour-associated macrophages and Reed-Sternberg cells in paediatric Hodgkin lymphoma, Eur J Cancer (2013), http://dx.doi.org/10.1016/j.ejca.2013.05.024
S. Gupta et al. / European Journal of Cancer xxx (2013) xxx–xxx
5
Table 4 Recent studies examining the prognostic impact of immunohistochemically identified tumour-associated macrophages among patients with Hodgkin lymphoma. N
Marker
Thresholda
Univariate prognostic impact
Multivariable prognostic impact
Deau et al. (2013) Greaves et al. (2013) Barros et al. (2012)c
59 122 95
Harris et al. (2012)
41
Sanchez-Espiridion et al. (2012)
369
Tan et al. (2012)
247
Azambuja et al. (2012)
265
Jakovic et al. (2011) Kamper et al. (2011)
52 288
Yoon et al. (2011)
144
Zaki et al. (2011)
82
Steidl et al. (2010)
166
CD68 CD68 CD68 CD163 CD68 CD163 CD68 CD163 CD68 CD163 CD68 CD163 CD68 CD68 CD163 CD68 CD163 CD68 CD163 CD68
25% <5%, 5–15%, >15% 25%, 50% 25%, 50% <5%, 5–25%, >25% <5%, 5–25%, >25% Multiple thresholds Multiple thresholds 12.70% 16.80% Both 5% and 25% Both 5% and 25% 25% 7.8% 21.1% 20% 20% Median Median 5%
Adverseb Adverse None Adverse Noneb Noneb Adversed None Adverse Adverse None None Adverse Adverse Adverse Adverse Adverse Adverse Nonee Adverse
Adverse Adverse None None – – – – Adverse Adverse – – Adverse Adverse None Adverse Adverse Adverse Nonee Adverse
N, number. a All threshold values represent the percentage of positive cells relative to total cellularity, with the exception of Barros et al. and Zaki et al., where the absolute number of was used. b Case–control analysis used. c Pediatric cohort. d Adverse impact was seen in disease-specific survival only, and only in one of two cohorts examined. e A trend towards adverse impact was seen, with p = 0.08 and p = 0.07 in univariate and multivariable analyses respectively.
microenvironment components differed from that reported in adults with HL. While an increase in RS cells as measured by CD30 positivity was associated with decreased EFS, an increase in macrophages, measured by either CD68 or CD163, did not affect outcome in this population. Multiple studies have now examined the significance of TAM among adult HL patients with most showing TAM to be associated with inferior survival (Table 4). 10–14,8,9,6,7,15,16 Though all retrospective, several have been multicentre or have involved both training and validation cohorts, lending credibility to their results.9 By contrast, the predictive value of the number of RS cells has been far less studied; Steidl et al. found CD30 positivity to have no impact upon outcome in adult HL.6 Very few studies have examined the impact of the tumour microenvironment in paediatric HL cohorts. To our knowledge, the only group to have done so found that among Brazilian children who were EBV negative, a higher percentage of either TIA1 or Granzyme B positive lymphocytes was associated with worse survival.17 In a separate paper using the same cohort, CD68 positivity did not impact outcome, while CD163 positivity adversely impacted progression-free survival in univariate, but not multivariable analysis.18 Neither marker significantly predicted overall survival. The
impact of RS positivity was not examined. In contrast to most TAM studies in the literature, the authors quantified TAM as an absolute number rather than the percentage relative to total cellularity. Quantifying actual numbers of TAM does not control for variability in tumour cellularity; high numbers of TAM may therefore simply be a reflection of high tumour cellularity. In addition, the authors examined the effect of multiple markers and ratios of markers within several subgroups. The possibility of false positives arising from multiple comparisons cannot be excluded. Our findings contradict the predominant message emerging from the adult literature. Several explanations for this discrepancy are possible. It is likely that true microenvironment-related prognostic differences exist between children and adults. Barros et al. recently showed that the composition of the tumour environment in HL differs between these two populations, with B lymphocytes outnumbering CD4 positive T lymphocytes in children but not in adults.17 Both Tan et al. and Yoon et al. found older age to be associated with higher percentages of CD68 or CD163 positive TAM.7,15 Several studies have also shown that the effect of EBV positivity differs among children and adults, impacting upon survival favorably in the former but adversely in the latter.19,20,21 Interestingly, in our cohort only 11/96 (11.4%) children had CD68 positivity of 25%
Please cite this article in press as: Gupta S. et al., The prognostic impact of tumour-associated macrophages and Reed-Sternberg cells in paediatric Hodgkin lymphoma, Eur J Cancer (2013), http://dx.doi.org/10.1016/j.ejca.2013.05.024
6
S. Gupta et al. / European Journal of Cancer xxx (2013) xxx–xxx
or higher, as opposed to 48/166 (28.9%) adults in the validation cohort of Steidl et al.6 Taken together, the above findings suggest that the impact of CD68, CD163, or CD30 positivity may plausibly vary between these two populations. Furthermore, the discrepancies in the literature may be a consequence of differences in methodology and limitations inherent in studies of tumour microenvironment such as differences in tissue fixation, immunohistochemical techniques and antibodies used, scoring methodology and intra-tumour heterogeneity. Both CD68 and CD163 have been used to identify TAM in the adult literature. In our population, the agreement between CD68 and CD163 positivity in the same sample was only moderate, with a weighted kappa value of 0.49. Yoon et al. found that while CD68 and CD163 positivity correlated, 20% of their cohort were designated positive by one marker but not by the other.15 CD68 has been noted to be a relatively non-specific macrophage marker that also stains fibroblasts and dendritic cells.13 As CD163 is a more specific marker of TAM, several authors have suggested that it may be a better prognosticator.13,10 Nonetheless, CD68 has been linked to survival in adults, while in our study neither CD68 nor CD163 showed prognostic significance. Another source of heterogeneity highlighted by this study involves intra-tumour variability. In our cohort, agreement between two cores from the same diagnostic specimen, both thought to be representative of the tumour, ranged from a weighted kappa of 0.60–0.73, indicating moderate heterogeneity. This limitation has received little attention in the literature, with very few studies attempting quantification. While Kamper et al. found intra-observer agreement to be ‘excellent’ with Pearson’s correlation coefficients of 0.95–0.97, Pearson correlation coefficients do not measure agreement, but rather trend.8 Two studies have attempted to overcome intra-tumour heterogeneity by analysing entire tissue cores using automated computer scanning.7,9 Interestingly, only one of these studies showed a consistent association between TAMs and prognosis.7 It is important to note that despite these inherent difficulties, the use of the microenvironment as a novel prognosticator remains promising. Our study found that positivity for any of the immunohistochemical markers studied was not associated with disease risk group, suggesting that microenvironment may be independent of traditional prognosticators and may therefore add to traditional risk stratification systems. Indeed, we found that the effect of CD30 positivity in a multivariable model including disease risk group was similar to that in univariate analyses (HR 2.19 and 2.27, respectively), though the threshold of statistical significance was not reached in the former. Strengths of this study include the sample size and the treatment-failure enriched nature, both of which
allowed for greater statistical power than obtainable by most single-institution studies. It is worth noting that including all 280 patients initially identified would not have greatly increased the statistical power as compared to the treatment-failure enriched cohort, as the number of events would not have changed. In addition to the general methodologic differences inherent to tumour microenvironment studies mentioned above, this study entailed other specific limitations. It is possible that despite cohort enrichment with cases of treatmentfailure, our study may still have remained underpowered to detect a smaller, but real adverse effect of TAMs than has been found by adult studies. In addition, in order to increase our sample size, we included patients diagnosed over a 29-year period treated with various regimens. Our study lacked the power to ascertain whether the impact of TAM or RS varied by treatment protocol. Despite these limitations, we were still able to show the impact of a higher percentage of RS cells upon survival. In conclusion, we found that in this cohort of paediatric patients with HL, a higher percentage of RS cells had an adverse effect on survival while a higher percentage of TAM did not. This suggests that the results of adult HL microenvironment studies may not extend to children. Further studies of the prognostic significance of various components of the tumour microenvironment are warranted in children with HL, ideally incorporating standardised and computerised assessment techniques into prospective multicentre trials. Role of funding source This study was supported through an operating Grant from the Physicians’ Services Incorporated Foundation. The funding agency had no role in the study design, collection, analysis and interpretation of the data, in the writing of the manuscript, and in the decision to submit the manuscript for publication. SG was supported through a doctoral fellowship award from the Canadian Institutes of Health Research and an operating Grant from the Canadian Child Health Clinician Scientist Program. Conflict of interest statement None declared. References 1. Punnett A, Tsang RW, Hodgson DC. Hodgkin lymphoma across the age spectrum: epidemiology, therapy and late effects. Semin Radiat Oncol 2010;20:30–44. 2. Sieniawski M, Rieneke T, Josting A, et al. Assessment of male fertility in patients with Hodgkin’s lymphoma treated in the German Hodgkin Study Group (GHSG) clinical trials. Ann Oncol 2008;19(10):1795–801.
Please cite this article in press as: Gupta S. et al., The prognostic impact of tumour-associated macrophages and Reed-Sternberg cells in paediatric Hodgkin lymphoma, Eur J Cancer (2013), http://dx.doi.org/10.1016/j.ejca.2013.05.024
S. Gupta et al. / European Journal of Cancer xxx (2013) xxx–xxx 3. Aleman BM, van den Belt-Dusebout AW, Klokman WJ, et al. Long-term cause-specific mortality of patients treated for Hodgkin’s disease. J Clin Oncol 2003;21(18):3431–9. 4. O’Brien MM, Donaldson SS, Balise RR, Whittlemore AS, Link MP. Second malignant neoplasms in survivors of pediatric Hodgkin’s lymphoma treated with low-dose radiation and chemotherapy. J Clin Oncol 2010;28(7):1232–9. 5. Steidl C, Connors JM, Gascoyne RD. Molecular pathogenesis of Hodgkin’s lymphoma: increasing evidence of the importance of the microenvironment. J Clin Oncol 2011;29(14):1812–26. 6. Steidl C, Lee T, Shah SP, et al. Tumor-associated macrophages and survival in classic Hodgkin’s lymphoma. N Engl J Med 2010;362:875–85. 7. Tan KL, Scott DW, Hong F, et al. Tumor-associated macrophages predict inferior outcomes in classic Hodgkin lymphoma: a correlative study from the E2496 Intergroup trial. Blood 2012;120:3280–7. 8. Kamper P, Bendix K, Hamilton-Dutoit S, et al. Tumor-infiltrating macrophages correlate with adverse prognosis and Epstein–Barr virus status in classical Hodgkin’s lymphoma. Haematologica 2011;96(2):269–76. 9. Sanchez-Espiridion B, Martin-Moreno AM, Montalban C, et al. Immunohistochemical markers for tumor associated macrophages and survival in advanced classical Hodgkin’s lymphoma. Haematologica 2012;97(7):1080–4. 10. Azambuja D, Natkunam Y, Biasoli I, et al. Lack of association of tumor-associated macrophages with clinical outcome in patients with classical Hodgkin’s lymphoma. Ann Oncol 2012;23:736–42. 11. Deau B, Bachy E, Ribrag V, et al. Macrophage, mast cell and T lymphocyte infiltrations are independent predictive biomarkers of primary refractoriness or early relapse in classical Hodgkin lymphom. Leuk Lymphoma 2013;54(1):41–5. 12. Greaves P, Clear A, Coutinho R, et al. Expression of FOXP3, CD68, and CD20 at diagnosis in the microenvironment of classical
13.
14.
15.
16.
17.
18.
19.
20.
21.
7
Hodgkin lymphoma is predictive of outcome. J Clin Oncol 2013;31:256–62. Harris JA, Jain S, Ren Q, et al. CD163 versus CD68 in tumor associated macrophages of classical Hodgkin lymphoma. Diagn Pathol 2012;7(12). Jakovic LR, Mihaljevic BS, Jovanovic MDP, et al. The prognostic relevance of tumor associated macrophages in advanced stage classical Hodgkin lymphoma. Leuk Lymphoma 2011;52(10): 1913–9. Yoon DH, Koh YW, Kang HJ, et al. CD68 and CD163 as prognostic factors for Korean patients with Hodgkin lymphoma. Eur J Haematol 2011;88:292–305. Zaki MAA, Wada N, Ikeda J, et al. Prognostic implication of types of tumor-associated macrophages in Hodgkin lymphoma. Virchows Arch 2011;459:361–6. Barros MHM, Vera-Lozada G, Soares FA, Niedobitek G, Hassan R. Tumor microenvironment composition in pediatric classical Hodgkin lymphoma is modulated by age and Epstein–Barr virus infection. Int J Cancer 2012;131:1142–52. Barros MH, Hassan R, Niedobitek G. Tumor-associated macrophages in pediatric classical Hodgkin lymphoma: association with Epstein–Barr virus, lymphocyte subsets, and prognostic impact. Clin Cancer Res 2012;18(14):3762–71. Engel M, Essop MF, Close P, et al. Improved prognosis of Epstein–Barr virus associated childhood Hodgkin’s lymphoma: study of 47 South African cases. J Clin Pathol 2000;53(3):182–6. Keegan TH, Claser SL, Clarke CA, et al. Epstein–Barr virus as a marker of survival after Hodgkin’s lymphoma: a population-based study. J Clin Oncol 2005;23(30):7604–13. Diepstra A, van Imhoff GW, Schaapveld M, et al. Latent Epstein– Barr virus infection of tumor cells in classical Hodgkin’s lymphoma predicts adverse outcome in older adult patients. J Clin Oncol 2009;27(23):3815–21.
Please cite this article in press as: Gupta S. et al., The prognostic impact of tumour-associated macrophages and Reed-Sternberg cells in paediatric Hodgkin lymphoma, Eur J Cancer (2013), http://dx.doi.org/10.1016/j.ejca.2013.05.024
Available at www.sciencedirect.com
journal homepage: www.ejcancer.com
The prognostic impact of tumour-associated macrophages and Reed-Sternberg cells in paediatric Hodgkin lymphoma q Sumit Gupta a,⇑, Stacy Yeh b, Rose Chami c, Angela Punnett a, Catherine Chung d a
Division of Hematology/Oncology, Hospital for Sick Children, 555 University Avenue, Toronto, Ontario, Canada M5G 1X8 Faculty of Medicine, University of Toronto, 1 Kings College Circle, Toronto, Ontario, Canada M5S 1A8 c Department of Pathology, CHU Sainte-Justine, 3175 Chemin de la Coˆte-Sainte-Catherine, Montreal, Quebec, Canada H3T 1C5 d Division of Pathology, Hospital for Sick Children, 555 University Avenue, Toronto, Ontario, Canada M5G 1X8 b
KEYWORDS Child Hodgkin lymphoma Macrophages Prognosis Reed-Sternberg cells
Abstract Background: Tumour-associated macrophages (TAM) are associated with treatment failure in adults with Hodgkin lymphoma (HL). Equivalent data in paediatric HL are sparse. We aimed to determine the prognostic significance of TAM and Reed-Sternberg (RS) cells in paediatric HL. Methods: All children aged 0–18 with HL between 1980 and 2009 with available diagnostic biopsy material were identified. A treatment failure-enriched cohort was assembled. Demographic, disease and outcome data were abstracted. Tissue microarrays with duplicate cores were constructed from diagnostic biopsy material and stained with immunohistochemical markers for TAM (CD68, CD163) and RS (CD30). A high score was defined as >5% positive cells relative to overall cellularity in any core. The association of candidate variables with event-free survival (EFS) was determined using Cox proportional hazards. Results: The final study cohort comprised 96 patients with a median age of 14 years (interquartile range 11–15). Agreement on scores between cores from the same biopsy revealed weighted kappas of 0.60, 0.68 and 0.73 for CD30, CD68 and CD163 respectively, indicating moderate tumour heterogeneity. In univariate analysis, a high CD30 score was significantly associated with treatment failure (hazard ratio (HR) 2.27; 95th confidence interval 1.01– 5.11; p < 0.05). High CD68 and CD163 scores were not associated with EFS. Conclusions: Unlike adult HL, a higher percentage of RS cells was associated with poor outcome, while a higher percentage of TAM was not. Adult HL findings may not extend to paediatric HL. Cooperative group trials of paediatric HL should prospectively determine the association of different components of the tumour microenvironment with outcome. Ó 2013 Elsevier Ltd. All rights reserved.
q Funding sources: This study was supported through an operating Grant from the Physicians’ Services Incorporated Foundation. SG was supported through a doctoral fellowship award from the Canadian Institutes of Health Research. ⇑ Corresponding author: Tel.: +1 416 813 7742. E-mail address: [email protected] (S. Gupta).
0959-8049/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ejca.2013.05.024
Please cite this article in press as: Gupta S. et al., The prognostic impact of tumour-associated macrophages and Reed-Sternberg cells in paediatric Hodgkin lymphoma, Eur J Cancer (2013), http://dx.doi.org/10.1016/j.ejca.2013.05.024
2
S. Gupta et al. / European Journal of Cancer xxx (2013) xxx–xxx
1. Introduction Hodgkin lymphoma (HL) is commonly encountered by paediatric oncologists.1 Multimodality treatment with chemotherapy and radiation has led to cure rates in excess of 80%, but at the cost of significant therapyrelated late effects including infertility, cardiac and respiratory dysfunction and second malignancies.2,3,4 Novel prognosticators available at the time of diagnosis may have significant impact by identifying children in whom treatment de-intensification is warranted. Alternatively, such prognosticators could also identify the children who are likely to fail treatment, allowing for upfront intensification. Recently, several studies have examined whether various aspects of the tumour microenvironment may play such a role. Uniquely among malignancies, the malignant cells in HL, Reed-Sternberg (RS) cells, can comprise less than 1% of the tumour cell population. The composition of the remaining 99%, called the tumour microenvironment, is determined by the complex expression of various cytokines and chemokines by both RS and reactive cells.5 The most promising potential prognosticator seems to be the presence of tumourassociated macrophages (TAM), which multiple investigators have found to be associated with inferior survival among adults with HL.6,7,8 As of yet, equivalent data for paediatric HL patients are sparse. The prognostic value of the burden of RS cells is relatively unexplored in both populations. Our primary objective was therefore to determine the prognostic significance of the percentage of both TAM and RS cells using the immunohistochemical markers CD68 and CD163, and CD30, respectively, in diagnostic biopsy specimens among children with HL. 2. Methods 2.1. Study population The study followed a retrospective cohort design. In order to increase statistical power, a cohort enriched with cases of treatment failure was assembled. This technique has been used in past studies of tumour microenvironment.6 All patients aged 0–18 years and diagnosed at The Hospital for Sick Children between 1980 and 2009 were identified using a local electronic patient database. Patients were excluded for the following reasons: (1) nodular lymphocyte predominant Hodgkin lymphoma (NLPHL), (2) post transplant lymphoproliferative disease, (3) prior immunodeficiency, (4) synchronous additional malignant diagnoses and (5) initial diagnostic biopsy material insufficient or unavailable for immunohistochemical studies. In cases where the initial biopsy was performed at an outside hospital, every effort was made to obtain the original biopsy material.
Using the same electronic database, all patients experiencing disease progression, relapse or death were identified and were added to the study cohort. The remaining patients (i.e. those who experienced long-term progression-free and relapse-free survival) were then identified. Patients from this remaining population were then randomly selected using a computer-generated random number sequence generator and added to the study cohort until a total sample size of 100 patients was reached; this total was determined a priori based on feasibility and study resources. 2.2. Immunohistochemistry Tissue microarrays (TMAs) were constructed from duplicate 1.0 mm cores from formalin-fixed paraffinembedded blocks of diagnostic biopsy tissue. Cores were obtained from representative areas containing RS cells as evaluated by morphology and immunohistochemical stains that were performed for routine diagnosis. A total of 22 samples had insufficient material to be included in the TMA but sufficient material for whole section immunohistochemical staining. These 22 patients were retained in the study cohort, with immunohistochemical staining performed on whole sections. Three arrays were built from the remaining 78 cases in duplicate with cases arranged randomly across the TMAs. Both four micron sections from the TMAs and whole sections were cut and stained with haematoxylin and eosin (H&E) and the immunohistochemical markers CD68 (predilute, clone KP1, Ventana), CD163 (1:100, clone 10D6, Leica) and CD30 (1:50, clone BerH2, Dako Canada). The immunohistochemical stains were performed on an automated stainer, Ventana Benchmark XT (Ventana Medical Systems, Tucson, Arizona), using a multimer detection kit (ultraView Universal DAB, Ventana Medical Systems, Tucson, Arizona). Two paediatric pathologists (C.C. and R.C.) independently scored the TMA and whole section slides. Two areas of whole section slides were scored. Both pathologists were blinded to patients’ clinical characteristics and treatment outcomes. Immunoreactivity was scored based on the percentage of cells showing cytoplasmic (CD68, CD163) and membranous and golgi (CD30) staining, relative to overall cellularity as follows: 1 = <5%, 2 = 5–25%, 3 = 26–50%, 4 = >50%. Discrepancies between the two reviewers were discussed until consensus was reached. 2.3. Data abstraction The charts of patients were obtained and data relating to patient characteristics, disease, treatment and outcome were abstracted. Patients were classified as belonging to the early time period if diagnosed between 1980 and 2000, and to the late period if diagnosed
Please cite this article in press as: Gupta S. et al., The prognostic impact of tumour-associated macrophages and Reed-Sternberg cells in paediatric Hodgkin lymphoma, Eur J Cancer (2013), http://dx.doi.org/10.1016/j.ejca.2013.05.024
S. Gupta et al. / European Journal of Cancer xxx (2013) xxx–xxx
between 2001 and 2009. Risk groups were defined according to the current Children’s Oncology Group classification: low (stages IA and IIA), intermediate (stages IB, IIB, IIIA and IVA) and high (stages IIIB and IVB). Various cutoffs for immunoreactivity have been used in the HL microenvironment literature.6,9,7 Based on the distribution of immunoreactivity scores between patients, a cutoff of greater or equal to 5% positive cells relative to overall cellularity was chosen for each immunohistochemical marker. Upon the discovery of moderate heterogeneity between cores from the same diagnostic biopsy (see below), a cutoff of 5% in any core was deemed sufficient to be considered positive. These decisions were made prior to any outcome analyses. 2.4. Statistical analysis The distribution of immunoreactivity was compared between risk groups for each immunohistochemical marker using the Chi square test or Fisher’s exact test as appropriate. The agreement between cores from the same diagnostic biopsy, and between CD68 and CD163 immunoreactivity on the same core, were both determined using weighted kappa statistics. The primary outcome was event-free survival (EFS), described using the Kaplan–Meier method and defined as the time from diagnosis to disease progression, relapse or death. Patients were censored at the time of last follow-up. Univariate and multivariable Cox proportional hazards models were used to explore predictors of EFS. The proportional hazards assumption was examined and verified. Variables with a P value < 0.05 on univariate analysis were examined in the multiple regression models. Statistical analyses were performed using SAS-PC software (version 9.2; SAS Institute, Cary, NC). Statistical significance was defined as P < 0.05. The study was approved by the institutional research ethics board, which also waived the need for written informed consent. 3. Results A total of 361 patients were identified as being diagnosed with HL during the study period. A total of 81 patients were excluded, leaving 280 patients in the final study cohort. Reasons for exclusion included insufficient biopsy material available for immunohistochemical staining (n = 39), NLPHL (n = 37), PTLD (n = 2), immunodeficiency (n = 2) and simultaneous diagnosis of non-Hodgkin lymphoma (n = 1). Of the final study cohort, 35 experienced progressive disease, relapse or death and met the inclusion criteria. Only one patient experienced death as a first event; sensitivity analyses excluding this patient did not significantly change any of the results (data not shown).
3
Table 1 Characteristics of study cohort patients. N (%) Gender Male Female
Median (IQR)
50 (52.1) 46 (47.9)
Age (years)
14 (11–15)
Stage I II III IV
5 (5.3) 45 (47.4) 15 (15.8) 30 (31.6)
B-symptoms Absent Present
48 (50.5%) 47 (49.5%)
Pathology Nodular sclerosis Mixed cellularity Lymphocyte deplete
87 (91.6) 5 (5.3) 3 (3.2)
Time period Early (1980–2000) Late (2001–2009)
41 (42.7) 55 (57.3)
Risk groupa Low Intermediate High
29 (30.9) 37 (39.4) 28 (29.8)
Radiation received Yes No
73 (77.7) 37 (22.3)
IQR, interquartile range; N, number. a See text for risk group definitions.
Table 2 Average percentage of CD30, CD68 and CD163 positive cells relative to total cellularity in diagnostic biopsy samples of study cohort patients. N (%) CD30 <5% 5 to <25% 25 to <50% >50%
53 (55.2) 37 (38.5) 6 (6.3) 0 (0.0)
CD68 <5% 5 to <25% 25 to <50% >50%
22 (22.9) 63 (65.5) 10 (10.4) 1 (1.0)
CD163 <5% 5 to <25% 25 to <50% >50%
57 (59.4) 30 (31.3) 9 (9,4) 0 (0.0)
IQR, interquartile range; N, number.
Another 65 patients, who experienced long-term progression-free and relapse-free survival, were randomly selected from the remaining patients and added to the cohort, bringing the total cohort sample size to 100 as a priori specified. The charts of four cohort patients
Please cite this article in press as: Gupta S. et al., The prognostic impact of tumour-associated macrophages and Reed-Sternberg cells in paediatric Hodgkin lymphoma, Eur J Cancer (2013), http://dx.doi.org/10.1016/j.ejca.2013.05.024
4
S. Gupta et al. / European Journal of Cancer xxx (2013) xxx–xxx
could not be located. As the TMAs had already been constructed, the final study cohort comprised of 96 children, the characteristics of which are shown in Table 1. A higher percentage of patients with nodular sclerosing histology was seen than expected, likely secondary to the treatment-failure enriched nature of the cohort; younger children with mixed cellularity histology have been suggested to have a superior prognosis.1 Fewer patient experiencing an event had received prior radiation than had randomly selected patients without an event (19/31– 61.3% versus 54/65–85.7%; p = 0.01). The percentages of CD30, CD68 and CD163 positive cells relative to total cellularity, as an average of both duplicate cores, are illustrated in Table 2. The weighted kappa statistic for agreement between cores from the same patient biopsy for CD30, CD68 and CD163 was 0.60 (95% confidence interval (CI) 0.43–0.77), 0.68 (95%CI 0.52–0.84) and 0.73 (0.61–0.84), respectively. The agreement between CD68 and CD163 positivity in
Table 3 Univariate predictors of event-free survival among study cohort patients. HR
95% CI
P value
Gender Male Female
– 1.13
– 0.55–2.32
– 0.74
Age (per year)
1.11
0.96–1.25
0.16
Risk group Low Intermediate High
– 1.92 2.05
– 0.73–5.04 0.76–5.54
– 0.19 0.16
Time period Early (1980–2000) Late (2001–2009)
– 1.28
– 0.61–2.70
– 0.51
CD30 All cores <5% positive cellularity Any core P5% positive cellularity
– 2.27
– 1.01–5.11
– 0.048
CD68 All cores <5% positive cellularity Any core P5% positive cellularity
– 1.09
– 0.42–2.85
– 0.86
CD163 All cores <5% positive cellularity Any core P5% positive cellularity
– 1.53
– 0.73–3.17
– 0.26
CI, confidence interval; HR, hazard ratio.
the same sample was 0.49 (0.35–0.63). An example of within-tumour heterogeneity is demonstrated in Fig. 1. Immunoreactivity was not significantly associated with disease risk group for any of the immunohistochemical markers examined (p = 0.17, 0.56 and 0.30 for CD30, CD68 and CD163, respectively). The 5-year EFS of the entire cohort was 65.0%, reflecting its enrichment with cases of treatment failure. Univariate predictors of EFS among the study cohort are shown in Table 3. Only CD30 positivity (immunoreactivity greater than or equal to 5% in any core) was significantly associated with worse outcome (hazard ratio (HR) 2.27, 95%CI 1.01–5.11; p = 0.048). Neither CD68 nor CD163 positivity was significantly associated with EFS. Sensitivity analyses of CD68 and CD163 positivity using a higher cutoff of 25% remained not significantly predictive of EFS (data not shown). Though only CD30 positivity reached statistical significance in univariate analyses, a multivariable model with CD30 immunoreactivity and risk group (itself incorporating stage and B symptoms) was constructed. In this model, the trend towards worse outcome with CD30 positivity was maintained (HR 2.19, 95%CI 0.9– 4.98; p = 0.06), though the threshold of statistical significance was not reached. Fig. 1. Example of intra-tumour heterogeneity in CD163 immunoreactivity. Panels (a) and (b) represent two different cores from the same biopsy sample taken from a patient with Hodgkin lymphoma, nodular sclerosis subtype. The percentage of CD163 positive macrophages is much higher in panel (a) compared to panel (b). Both images captured with 20 objective magnification (200 total magnification).
4. Discussion This study is one of the first to look at the prognostic significance of either TAM or RS cells among children with HL. We found that the impact of both
Please cite this article in press as: Gupta S. et al., The prognostic impact of tumour-associated macrophages and Reed-Sternberg cells in paediatric Hodgkin lymphoma, Eur J Cancer (2013), http://dx.doi.org/10.1016/j.ejca.2013.05.024
S. Gupta et al. / European Journal of Cancer xxx (2013) xxx–xxx
5
Table 4 Recent studies examining the prognostic impact of immunohistochemically identified tumour-associated macrophages among patients with Hodgkin lymphoma. N
Marker
Thresholda
Univariate prognostic impact
Multivariable prognostic impact
Deau et al. (2013) Greaves et al. (2013) Barros et al. (2012)c
59 122 95
Harris et al. (2012)
41
Sanchez-Espiridion et al. (2012)
369
Tan et al. (2012)
247
Azambuja et al. (2012)
265
Jakovic et al. (2011) Kamper et al. (2011)
52 288
Yoon et al. (2011)
144
Zaki et al. (2011)
82
Steidl et al. (2010)
166
CD68 CD68 CD68 CD163 CD68 CD163 CD68 CD163 CD68 CD163 CD68 CD163 CD68 CD68 CD163 CD68 CD163 CD68 CD163 CD68
25% <5%, 5–15%, >15% 25%, 50% 25%, 50% <5%, 5–25%, >25% <5%, 5–25%, >25% Multiple thresholds Multiple thresholds 12.70% 16.80% Both 5% and 25% Both 5% and 25% 25% 7.8% 21.1% 20% 20% Median Median 5%
Adverseb Adverse None Adverse Noneb Noneb Adversed None Adverse Adverse None None Adverse Adverse Adverse Adverse Adverse Adverse Nonee Adverse
Adverse Adverse None None – – – – Adverse Adverse – – Adverse Adverse None Adverse Adverse Adverse Nonee Adverse
N, number. a All threshold values represent the percentage of positive cells relative to total cellularity, with the exception of Barros et al. and Zaki et al., where the absolute number of was used. b Case–control analysis used. c Pediatric cohort. d Adverse impact was seen in disease-specific survival only, and only in one of two cohorts examined. e A trend towards adverse impact was seen, with p = 0.08 and p = 0.07 in univariate and multivariable analyses respectively.
microenvironment components differed from that reported in adults with HL. While an increase in RS cells as measured by CD30 positivity was associated with decreased EFS, an increase in macrophages, measured by either CD68 or CD163, did not affect outcome in this population. Multiple studies have now examined the significance of TAM among adult HL patients with most showing TAM to be associated with inferior survival (Table 4). 10–14,8,9,6,7,15,16 Though all retrospective, several have been multicentre or have involved both training and validation cohorts, lending credibility to their results.9 By contrast, the predictive value of the number of RS cells has been far less studied; Steidl et al. found CD30 positivity to have no impact upon outcome in adult HL.6 Very few studies have examined the impact of the tumour microenvironment in paediatric HL cohorts. To our knowledge, the only group to have done so found that among Brazilian children who were EBV negative, a higher percentage of either TIA1 or Granzyme B positive lymphocytes was associated with worse survival.17 In a separate paper using the same cohort, CD68 positivity did not impact outcome, while CD163 positivity adversely impacted progression-free survival in univariate, but not multivariable analysis.18 Neither marker significantly predicted overall survival. The
impact of RS positivity was not examined. In contrast to most TAM studies in the literature, the authors quantified TAM as an absolute number rather than the percentage relative to total cellularity. Quantifying actual numbers of TAM does not control for variability in tumour cellularity; high numbers of TAM may therefore simply be a reflection of high tumour cellularity. In addition, the authors examined the effect of multiple markers and ratios of markers within several subgroups. The possibility of false positives arising from multiple comparisons cannot be excluded. Our findings contradict the predominant message emerging from the adult literature. Several explanations for this discrepancy are possible. It is likely that true microenvironment-related prognostic differences exist between children and adults. Barros et al. recently showed that the composition of the tumour environment in HL differs between these two populations, with B lymphocytes outnumbering CD4 positive T lymphocytes in children but not in adults.17 Both Tan et al. and Yoon et al. found older age to be associated with higher percentages of CD68 or CD163 positive TAM.7,15 Several studies have also shown that the effect of EBV positivity differs among children and adults, impacting upon survival favorably in the former but adversely in the latter.19,20,21 Interestingly, in our cohort only 11/96 (11.4%) children had CD68 positivity of 25%
Please cite this article in press as: Gupta S. et al., The prognostic impact of tumour-associated macrophages and Reed-Sternberg cells in paediatric Hodgkin lymphoma, Eur J Cancer (2013), http://dx.doi.org/10.1016/j.ejca.2013.05.024
6
S. Gupta et al. / European Journal of Cancer xxx (2013) xxx–xxx
or higher, as opposed to 48/166 (28.9%) adults in the validation cohort of Steidl et al.6 Taken together, the above findings suggest that the impact of CD68, CD163, or CD30 positivity may plausibly vary between these two populations. Furthermore, the discrepancies in the literature may be a consequence of differences in methodology and limitations inherent in studies of tumour microenvironment such as differences in tissue fixation, immunohistochemical techniques and antibodies used, scoring methodology and intra-tumour heterogeneity. Both CD68 and CD163 have been used to identify TAM in the adult literature. In our population, the agreement between CD68 and CD163 positivity in the same sample was only moderate, with a weighted kappa value of 0.49. Yoon et al. found that while CD68 and CD163 positivity correlated, 20% of their cohort were designated positive by one marker but not by the other.15 CD68 has been noted to be a relatively non-specific macrophage marker that also stains fibroblasts and dendritic cells.13 As CD163 is a more specific marker of TAM, several authors have suggested that it may be a better prognosticator.13,10 Nonetheless, CD68 has been linked to survival in adults, while in our study neither CD68 nor CD163 showed prognostic significance. Another source of heterogeneity highlighted by this study involves intra-tumour variability. In our cohort, agreement between two cores from the same diagnostic specimen, both thought to be representative of the tumour, ranged from a weighted kappa of 0.60–0.73, indicating moderate heterogeneity. This limitation has received little attention in the literature, with very few studies attempting quantification. While Kamper et al. found intra-observer agreement to be ‘excellent’ with Pearson’s correlation coefficients of 0.95–0.97, Pearson correlation coefficients do not measure agreement, but rather trend.8 Two studies have attempted to overcome intra-tumour heterogeneity by analysing entire tissue cores using automated computer scanning.7,9 Interestingly, only one of these studies showed a consistent association between TAMs and prognosis.7 It is important to note that despite these inherent difficulties, the use of the microenvironment as a novel prognosticator remains promising. Our study found that positivity for any of the immunohistochemical markers studied was not associated with disease risk group, suggesting that microenvironment may be independent of traditional prognosticators and may therefore add to traditional risk stratification systems. Indeed, we found that the effect of CD30 positivity in a multivariable model including disease risk group was similar to that in univariate analyses (HR 2.19 and 2.27, respectively), though the threshold of statistical significance was not reached in the former. Strengths of this study include the sample size and the treatment-failure enriched nature, both of which
allowed for greater statistical power than obtainable by most single-institution studies. It is worth noting that including all 280 patients initially identified would not have greatly increased the statistical power as compared to the treatment-failure enriched cohort, as the number of events would not have changed. In addition to the general methodologic differences inherent to tumour microenvironment studies mentioned above, this study entailed other specific limitations. It is possible that despite cohort enrichment with cases of treatmentfailure, our study may still have remained underpowered to detect a smaller, but real adverse effect of TAMs than has been found by adult studies. In addition, in order to increase our sample size, we included patients diagnosed over a 29-year period treated with various regimens. Our study lacked the power to ascertain whether the impact of TAM or RS varied by treatment protocol. Despite these limitations, we were still able to show the impact of a higher percentage of RS cells upon survival. In conclusion, we found that in this cohort of paediatric patients with HL, a higher percentage of RS cells had an adverse effect on survival while a higher percentage of TAM did not. This suggests that the results of adult HL microenvironment studies may not extend to children. Further studies of the prognostic significance of various components of the tumour microenvironment are warranted in children with HL, ideally incorporating standardised and computerised assessment techniques into prospective multicentre trials. Role of funding source This study was supported through an operating Grant from the Physicians’ Services Incorporated Foundation. The funding agency had no role in the study design, collection, analysis and interpretation of the data, in the writing of the manuscript, and in the decision to submit the manuscript for publication. SG was supported through a doctoral fellowship award from the Canadian Institutes of Health Research and an operating Grant from the Canadian Child Health Clinician Scientist Program. Conflict of interest statement None declared. References 1. Punnett A, Tsang RW, Hodgson DC. Hodgkin lymphoma across the age spectrum: epidemiology, therapy and late effects. Semin Radiat Oncol 2010;20:30–44. 2. Sieniawski M, Rieneke T, Josting A, et al. Assessment of male fertility in patients with Hodgkin’s lymphoma treated in the German Hodgkin Study Group (GHSG) clinical trials. Ann Oncol 2008;19(10):1795–801.
Please cite this article in press as: Gupta S. et al., The prognostic impact of tumour-associated macrophages and Reed-Sternberg cells in paediatric Hodgkin lymphoma, Eur J Cancer (2013), http://dx.doi.org/10.1016/j.ejca.2013.05.024
S. Gupta et al. / European Journal of Cancer xxx (2013) xxx–xxx 3. Aleman BM, van den Belt-Dusebout AW, Klokman WJ, et al. Long-term cause-specific mortality of patients treated for Hodgkin’s disease. J Clin Oncol 2003;21(18):3431–9. 4. O’Brien MM, Donaldson SS, Balise RR, Whittlemore AS, Link MP. Second malignant neoplasms in survivors of pediatric Hodgkin’s lymphoma treated with low-dose radiation and chemotherapy. J Clin Oncol 2010;28(7):1232–9. 5. Steidl C, Connors JM, Gascoyne RD. Molecular pathogenesis of Hodgkin’s lymphoma: increasing evidence of the importance of the microenvironment. J Clin Oncol 2011;29(14):1812–26. 6. Steidl C, Lee T, Shah SP, et al. Tumor-associated macrophages and survival in classic Hodgkin’s lymphoma. N Engl J Med 2010;362:875–85. 7. Tan KL, Scott DW, Hong F, et al. Tumor-associated macrophages predict inferior outcomes in classic Hodgkin lymphoma: a correlative study from the E2496 Intergroup trial. Blood 2012;120:3280–7. 8. Kamper P, Bendix K, Hamilton-Dutoit S, et al. Tumor-infiltrating macrophages correlate with adverse prognosis and Epstein–Barr virus status in classical Hodgkin’s lymphoma. Haematologica 2011;96(2):269–76. 9. Sanchez-Espiridion B, Martin-Moreno AM, Montalban C, et al. Immunohistochemical markers for tumor associated macrophages and survival in advanced classical Hodgkin’s lymphoma. Haematologica 2012;97(7):1080–4. 10. Azambuja D, Natkunam Y, Biasoli I, et al. Lack of association of tumor-associated macrophages with clinical outcome in patients with classical Hodgkin’s lymphoma. Ann Oncol 2012;23:736–42. 11. Deau B, Bachy E, Ribrag V, et al. Macrophage, mast cell and T lymphocyte infiltrations are independent predictive biomarkers of primary refractoriness or early relapse in classical Hodgkin lymphom. Leuk Lymphoma 2013;54(1):41–5. 12. Greaves P, Clear A, Coutinho R, et al. Expression of FOXP3, CD68, and CD20 at diagnosis in the microenvironment of classical
13.
14.
15.
16.
17.
18.
19.
20.
21.
7
Hodgkin lymphoma is predictive of outcome. J Clin Oncol 2013;31:256–62. Harris JA, Jain S, Ren Q, et al. CD163 versus CD68 in tumor associated macrophages of classical Hodgkin lymphoma. Diagn Pathol 2012;7(12). Jakovic LR, Mihaljevic BS, Jovanovic MDP, et al. The prognostic relevance of tumor associated macrophages in advanced stage classical Hodgkin lymphoma. Leuk Lymphoma 2011;52(10): 1913–9. Yoon DH, Koh YW, Kang HJ, et al. CD68 and CD163 as prognostic factors for Korean patients with Hodgkin lymphoma. Eur J Haematol 2011;88:292–305. Zaki MAA, Wada N, Ikeda J, et al. Prognostic implication of types of tumor-associated macrophages in Hodgkin lymphoma. Virchows Arch 2011;459:361–6. Barros MHM, Vera-Lozada G, Soares FA, Niedobitek G, Hassan R. Tumor microenvironment composition in pediatric classical Hodgkin lymphoma is modulated by age and Epstein–Barr virus infection. Int J Cancer 2012;131:1142–52. Barros MH, Hassan R, Niedobitek G. Tumor-associated macrophages in pediatric classical Hodgkin lymphoma: association with Epstein–Barr virus, lymphocyte subsets, and prognostic impact. Clin Cancer Res 2012;18(14):3762–71. Engel M, Essop MF, Close P, et al. Improved prognosis of Epstein–Barr virus associated childhood Hodgkin’s lymphoma: study of 47 South African cases. J Clin Pathol 2000;53(3):182–6. Keegan TH, Claser SL, Clarke CA, et al. Epstein–Barr virus as a marker of survival after Hodgkin’s lymphoma: a population-based study. J Clin Oncol 2005;23(30):7604–13. Diepstra A, van Imhoff GW, Schaapveld M, et al. Latent Epstein– Barr virus infection of tumor cells in classical Hodgkin’s lymphoma predicts adverse outcome in older adult patients. J Clin Oncol 2009;27(23):3815–21.
Please cite this article in press as: Gupta S. et al., The prognostic impact of tumour-associated macrophages and Reed-Sternberg cells in paediatric Hodgkin lymphoma, Eur J Cancer (2013), http://dx.doi.org/10.1016/j.ejca.2013.05.024