To the Editor:

We read with interest the recent article “PD-L1 expression is associated with ALK positivity and STAT3 activation, but not outcome in patients with systemic anaplastic large cell lymphoma” [1]. Shen et al. examined PD-L1 expression by immunohistochemistry in 95 systemic anaplastic large cell lymphomas (ALCLs) and concluded that PD-L1 expression has no prognostic significance in ALCL, including ALK-negative ALCL.

In a previous study that examined PD-L1 expression in 152 ALCLs, we found that ALK-negativeALCLs with DUSP22 rearrangements expressed minimal PD-L1 and pSTAT3Tyr705 [2], a finding subsequently replicated [3]. We hypothesized that the lack of PD-L1 might be one component of an immunogenic phenotype contributing to the excellent outcomes usually seen in DUSP22-rearranged ALCLs [4, 5]. In light of the data of Shen et al. we examined the association of PD-L1 expression and outcome in our previous series [2]. Of the 152 ALCLs stained for PD-L1, 62 were systemic ALCLs with available outcome data (18 ALK-positive, 44 ALK-negative; M:F, 38:24; mean age, 55 years; median follow-up, 43 months). Using the cutoff of 5% for PD-L1 positivity employed by Shen et al. we found that PD-L1 positivity was associated with inferior outcomes in ALK-negative ALCL (median survival, 38 vs. 262 months in PD-L1-negative cases; P = 0.03, log-rank test; Fig. 1a).

Fig. 1: Overall survival in systemic ALK-negative anaplastic large cell lymphoma, stratified by PD-L1 expression.
figure 1

a Based on a cutoff for positivity of 5% of the neoplastic cells staining, as per Shen et al. [1]. b Based on a cutoff of 55%, the median PD-L1 value in our series.

Full size image

Several factors might account for the different results in our series and that of Shen et al. While used in several studies, the 5% cutoff for PD-L1 positivity may have difficulties with reproducibility, potentially accentuated by antibody/protocol differences ([2]; not specified in Shen et al.). Only 27% of our ALK-negative ALCLs were PD-L1-negative using this cutoff, compared with 58% in Shen et al. Analyzing our data with a cutoff based on the median PD-L1 value of 55%, we found an even stronger association of PD-L1 positivity with outcome (median survival, 7 vs. 216 months in PD-L1-negative cases; P = 0.002; Fig. 1b). Differences in the composition of the study groups may also contribute to the different results. Although median follow-up was only 20 months in Shen et al. and 5-year overall survival in ALK-negative ALCL is not provided, from Fig. 4 of Shet et al., it appears to be ~25%, which is considerably poorer than the 52% in our data and references they cite (e.g., [6], 49%). Possibly, the more clinically aggressive disease in the series of Shen et al. masked contributions of PD-L1. Though not reported, the series by Shen et al. might have fewer DUSP22-rearranged ALCLs than ours (24%), or might be enriched for “high-risk” DUSP22-rearranged ALCLs as recently described [3].

Based on these discrepancies and the features of the series analyzed by Shen et al. including limited follow-up, we believe the conclusion that PD-L1 is not associated with outcome in ALK-negative ALCL is premature. The collective literature is in agreement that ALCLs demonstrate variable expression of PD-L1. As Shen et al. note, the most important implication of this variable expression is the hypothesis that PD-L1 functionally promotes immune escape in some ALCLs but not others and might serve as a biomarker to individualize immunotherapeutic approaches. Our data showing poorer outcomes in PD-L1-positive ALK-negative ALCLs are consistent with this hypothesis and with data from other cancers as reviewed by Shen et al. Future work should examine the functional role of PD-L1 in ALCL and its relationship to immunotherapy response in preclinical models and clinical trials.