Discussion
The results of our study add significantly to the limited data available for pediatric patients. This is due to the high percentage of relapsed patients in our cohort and to the high number of markers analyzed8-10,15. We found an association between PD-L1 expression and interim remission status, which is not reported in the literature before. PD-L1 plays an important role in the pathogenesis of cHL. It is upregulated through amplification of chromosome 9p24.1 and may support cHL proliferation 18. PD-L1 also contributes to the immune escape of the tumor 19. This may explain why patients with cHL respond so well to PD-1/PD-L1 checkpoint inhibition 19-21. Our findings imply that PD-L1 expression in pediatric cHL is correlated with outcome. This is an important finding that should be further investigated in a larger cohort of pediatric patients.
Remarkably, there are some differences between our data and previously published findings in children with cHL (See Supplemental Table S5 for a summary of the findings of previous studies). Barros et al. found a significant association between the expression of CD163 and the progression-free survival (PFS) 8, however, our study and the study of Gupta et al. found no significant association. Gupta et al. and Dinand et al. found a significant association between CD30 and CD15 expression and EFS 9,15. We were not able to confirm this finding. These differences may be caused by differences in treatment protocols between the different studies. Second, our study contained a much higher percentage of events than the study of Barros and Dinand 8,15. Third, expression patterns in our study were based on lager areas and exact percentages were calculated.
Interestingly, our results are in contrast with previous findings in adult patients. They mostly reported an adverse association between OS and EFS and the expression of PAX5 on the HRS cells, and CD68+, CD163+, PD-1+, and PD-L1+ in the TME 4-7. These differences between pediatric cHL and adult cHL may be due to differences in pathogenesis, resulting from differences in the composition of the TME10-14. Moreover, differences may be due to the use of different antibodies and/or scoring methods.
We found high expression of CD30 and PD-L1 on HRS cells, and PD-1 and PD-L1 in the TME in all patients. CD30, PD-L1 and PD-1 are all markers that can be targeted therapeutically (e.g. by brentuximab vedotin binding to CD30). The expression in pediatric patients underscores the need for further investigations of these novel therapies in first line treatment in children.
Our study is the first to examine eight different markers simultaneously. Another strength of our data is the event-enriched setting particularly in pediatric patients. However, due to this event-enriched setup, we included patients diagnosed over a period of 18 years with different treatment regimens. Despite this event-enriched cohort, the study still lacked the statistical power to perform a sub-analysis per treatment protocol for the primary outcome.
A possible limitation of the study is the use of immunohistochemical staining. This can lead to inconsistent results depending on differences in tissue fixation, the type of antibody and staining method, the scoring method, and the observer’s interpretation. To minimize these inconsistencies, we based our scoring method on published previous studies 4,7,14,22,23 and we tried to overcome the interobserver variability by using a dual-head microscope and reviewing tissue samples together.
In conclusion, we found an association between PD-L1 expression and interim remission status, which should be further investigated. Furthermore, our data demonstrate differences in the expression patterns and prognostic impact of immunohistochemical markers between pediatric and adult patients with cHL. Further research into these differences may lead to specific prognostic factors in pediatric cHL, indispensable for improvement of treatment in this population.