Background
The standard of care for metastatic melanoma is immune checkpoint blockade (ICB) or targeted therapy [
1]. In the absence of brain metastases, pembrolizumab and nivolumab monotherapy yield durable responses in approx. 30–40% of patients; the complete response (CR) rate is approx. 15% [
2,
3]. Additional radiotherapy (RT) may increase the rate of deep and durable responses. Preclinical work has shown that localized RT can induce CD8+ T cells, which contribute to the control of the irradiated tumor and sometimes elicit abscopal effects in non-irradiated metastases, particularly when combined with ICB [
4‐
7]. Case reports [
8] and clinical trials have also provided evidence for RT-induced abscopal effects [
9‐
14]. However, it is not fully clear how to best induce an RT-mediated abscopal response and whether pretreatment biomarkers can predict which patients respond to combined RT/ICB.
Melanoma has been regarded as radioresistant, and, in melanoma, RT is currently mainly used in the adjuvant setting or to palliate symptoms [
15]. Advanced age is usually a disadvantage for immunotherapy, mainly because of immunosenescence, but anecdotal clinical experience suggests that advanced age does not result in poorer responses in elderly patients treated with ICB [
16].
Here, we report on two elderly patients with oligometastatic melanoma who mounted a deep (long-lasting complete or radiologic complete) response to stereotactic body radiotherapy (SBRT) and anti-PD-1. Patient 1 had a favorable pretreatment immune signature and immediately responded with a CR, now already lasting for more than 4.5 years. Patient 2 had an unfavorable pretreatment immune signature. Nonetheless, he showed a radiologic CR after a second (late) SBRT delivered after more than 10 months of progression on anti-PD-1 and the first SBRT. However, eventually, his disease progressed. Besides various potential pre- and on-treatment biomarkers, we also characterized CD8+ T cell responses to melanoma differentiation antigens and neoantigens. The latter could be useful for the design of additional immunotherapies that might further deepen responses, including patients with unfavorable immune signature.
Materials and methods
Patient study
All human samples were collected after approval by the Ethics Committee of the Albert-Ludwigs University Freiburg, Germany (protocol no. 453/14) following written informed consent.
Whole-exome sequencing, RNA sequencing, HLA typing, and neoepitope prediction
DNA and RNA were extracted from formalin-fixed, paraffin-embedded tumor sections and PBMCs. Sequencing was performed by Personalis (Menlo Park, USA), and gene expression profiles and potential neoepitope lists were generated. HLA typing was also performed by next-generation sequencing. The neoepitopes were chosen by the rank of the predicted affinity using NetMHCpan 4.0, the ratio of wild-type- and tumor-binding rank, and based on source protein expression. Transcripts per million base pairs were used to compare gene expression levels between the two patients.
Immunohistochemistry (IHC)
Tumor sections were stained with hematoxylin and eosin (H&E) or with primary antibodies to CD8 (clone C8/144B from Dako-Agilent) and PD-L1 (Ventana clone SP142 from Roche). IHC staining was conducted using K8020 Envision Flex (for brown stainings) or K5005 alkaline phosphatase detection kits (for red stainings) in a DAKO Plus Austostainer.
Detection of differentiation antigen- and neoepitope-specific CD8+ T cells
PBMCs were incubated with 13 potential neoepitope peptides or HLA-A2-restricted peptides derived from differentiation antigens (each 10 μg/ml) plus anti-CD28 (0.5 μg/ml) (CD28.2). The following differentiation antigen-derived epitopes were used: gp100209–217 (IMDQVPFSV), MART-126–35 (ELAGIGILTV), gp100280–288 (YLEPGPVTV), and tyrosinase369–377 (YMDGTMSQV). From day 2 onwards, IL-2 (20–100 IU/ml) was added. Every 14 days, the cultures were restimulated with peptide-pulsed, 40-Gy-irradiated autologous PBMCs. Epitope-specific T cell assays were performed after a 6-h restimulation with the respective peptide pools or with individual peptides. After 1 h, brefeldin A was added. Cells were then stained for surface CD8+ and intracellular IFNγ and analyzed by flow cytometry.
Flow cytometry
Cells were stained with Zombie Red and then with antibodies against CD3 (OK3), CD8 (SK1), PD-1 (EH12.2H7), CD45RA (HI100), and CCR7 (G043H7). For nuclear staining of Ki67 (Ki67), we used the eBioscience Fixation/Permeabilisation kit. For IFNγ (clone 4S.B3) staining, the Fixation and Permeabilization buffer from Invitrogen was used. Cells were analyzed on a CytoflexS cytometer (Beckman Coulter).
Discussion
Clinical attempts to induce abscopal effects in metastatic patients have so far been based on the irradiation of one or two tumor nodules, with limited success. Our data suggest that patients with limited or oligometastatic disease, where a large proportion of the tumor mass can be irradiated, are good candidates to increase ICB response rates by RT, even in case of an unfavorable pretreatment immune signature, after progression on long-term anti-PD-1, and despite advanced age. We also show that deep abscopal effects can be achieved through a repeated irradiation, but long-term outcomes may be worse in patients with unfavorable immune signature. However, even in the patient with the unfavorable pretreatment immune signature, tumor epitope-specific T cells could be detected in the blood after the first (ineffective) attempt to induce an abscopal effect by RT, and this could be the basis for additional epitope-based immunotherapies.
Patient 1 had favorable pretreatment biomarkers. However, in sum, her tumor lesions showed a diameter of less than 5 cm and PD-L1 positivity, and therefore she had an approx. 40% chance of achieving a CR upon anti-PD-1 monotherapy [
25]. It is therefore unclear if the SBRT contributed to the CR in this patient. In contrast, patient 2 showed a clear RT-induced abscopal response after a second SBRT following progression on long-term anti-PD-1 and the first SBRT. This strong abscopal response occurred despite low CD8+ T cell infiltration, the absence of a T cell exhaustion and cytotoxicity signature, and the presence of immunosuppressive cells in the pretreatment tumor tissue.
Preclinical experiments suggest that non-ablative hypofractionated RT may work best to induce abscopal responses [
5,
6]. In our study, both patients initially received ablative SBRT, resulting in CR in patient 1 but no systemic regression in patient 2. The second (effective) SBRT in patient 2 had a curative total dose but non-ablative fraction doses. Moreover, based on the assumption that TILs are radiosensitive, it is usually assumed that repetitive tumor irradiations may be detrimental. However, the systemically effective (second) SBRT in patient 2 consisted of 8 fractions distributed over 2.5 weeks. This is consistent with our recent preclinical work showing that such extended hypofractionated RT schedules are not necessarily less effective than short schedules [
26]. It is also consistent with a recent report showing that tumor-resident T cells are quite radioresistant [
27].
RT-mediated abscopal effects depend on tumor-specific CD8+ T cells. Therefore, the direct RT-mediated reduction in the tumor mass (which is feasible in limited or oligometastatic disease [
24]) may be beneficial, because it likely facilitates T cell-mediated tumor rejection. In our patients, a relatively high proportion of the visible tumor mass (up to > 90%) was irradiated.
In accordance with the poor pretreatment immune signature, the radiologic CR in patient 2 was only transient. However, upon initiation of ICB and the first SBRT (August/September 2015), patient 2 showed an increase in Ki67+ (i.e., activated) CD8+ T cells and of Ki67+ PD-1+ CD8+ T cells in the blood, which was still detected several months later (May 2016) despite tumor progression. In accordance with these findings, differentiation antigen- and neoepitope-specific CD8+ T cells could clearly be detected during this period of post-treatment tumor progression (January 2016). The detection of tumor epitope-specific T cells could be the basis for additional, epitope-based immunotherapies (vaccination, adoptive T cell transfer) that could help to improve antitumor responses.
Easily detectable on-treatment changes in biomarkers are also of interest to identify effective treatment combinations. At the time point of successful systemic/abscopal response, both patients had high levels (approx. 40–50%) of Ki67+ (i.e., proliferating) PD-1+ CD8+ T cells; in patient 2, this increase was induced by RT despite resistance to anti-PD-1. Since the patient was still under PD-1 ICB, it is unclear whether (but possible that) the increase in Ki67+ T cells was induced by the tumor irradiation alone.
In both patients, a strong antitumor response was observed despite early ICB discontinuation and early commencement of systemic corticosteroids after CR. Whereas the bullous disease in patient 2 was clearly related to anti-PD-1 [
28], it is unclear to which extent anti-PD-1 and liver SBRT contributed to the transient grade 3 transaminitis in patient 1 [
29]. Our data also show that combined SBRT/ICB can be highly effective in elderly patients [
30].
Future clinical trials could investigate to which extent (early or late) SBRT enhances ICB response rates in patients with oligometastatic disease where all or large parts of the visible tumor mass can be irradiated. The effects of repeated irradiations could also be addressed in such trials. Our study furthermore suggests that patients with poor pretreatment CD8+ T cell infiltration should not be excluded from attempts to induce RT-mediated systemic/abscopal effects, at least in oligometastatic disease where all or large parts of the visible tumor mass can be irradiated. In addition, our data support the notion that old patients should not be excluded from attempts to induce RT-mediated abscopal effects.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.