Influenza vaccination of cancer patients during PD-1 blockade induces serological protection but may raise the risk for immune-related adverse events

The development of blocking antibodies that target inhibitory PD-1/PD-L1 or CTLA-4/CD80/CD86 pathways has led to significant improvements in the prognosis of patients suffering from various cancers including metastatic melanoma, non-small cell lung cancer (NSCLC), renal cell carcinoma (RCC), Hodgkin lymphoma, squamous carcinoma of the head and neck (SCCHN) and bladder cancer [1–6]. Checkpoint inhibition has revolutionized cancer therapy of patients with advanced disease by induction of durable remissions and potential cures in some patients [7–9]. PD-1 interactions with its ligands PD-L1 or PD-L2 is an immune checkpoint that is importantly involved in immune homeostasis and prevents extensive tissue destruction by T cells e.g. during viral infections [10], but can also be involved in T cell dysfunction and relapses of viral infections [11, 12]. Checkpoint inhibition with blocking antibodies against PD-1 or PD-L1 augments T-cell immunity [10] – thereby increasing cancer-specific immunity. However, also virus-specific immunity is increased due to blockade of the PD-1 signalling cascade [13, 14]. Treatments with agents targeting the PD-1/PD-L1 axis usually show a good safety profile with a low risk for grade 3 to 5 immune-related adverse events (irAEs) [15–18]. While severe irAEs are an uncommon complication of anti-PD-1/PD-L1 monotherapy, irAEs can be devastating for patients that are affected.

In patients with cancer, infection with influenza viruses is associated with significant morbidity and mortality [19, 20]. Therefore, vaccination as prevention for influenza virus infection is recommended for patients with cancer and in particular for patients that undergo anti-neoplastic therapy [19, 20]. Patients with NSCLC have an additional risk for complications due to concomitant pre-existing lung disorders such as chronic obstructive pulmonary disease (COPD) [21]. Several analyses of the vaccine-induced humoral immune response in patients undergoing classical cytotoxic chemotherapy have been performed [22–26]. In general, studies have shown that concomitant vaccination against seasonal influenza strains is safe in patients undergoing cytotoxic chemotherapy. However, most of these studies showed a reduced efficacy to mount seroprotective post-vaccine antibody titers [22–24]. While the humoral immune response in patients receiving cytotoxic chemotherapy is reduced, the response in patients undergoing checkpoint blockade for cancer is unknown.

This study aimed to determine the quantity and quality of influenza-specific immune responses and the frequency, type and severity of irAEs in cancer patients undergoing immunotherapy with antibodies targeting the PD-1/PD-L1 pathway.

Here we report on the humoral immune response and safety of a trivalent, inactivated, non-adjuvanted influenza vaccine in patients that were treated with a PD-1/PD-L1 blocking agent. The cohort of patients received a seasonal vaccination for the prevention of influenza during the season 2015/2016 in Switzerland. Most of the patients were treated for metastatic NSCLC. In our cohort, the overall seroprotective levels at day 30 were very similar between cancer patients undergoing checkpoint blockade and healthy age-matched controls. However, the seroconversion rate was significantly higher in patients under immune checkpoint blockade, indicating a much more potent immune stimulation in cancer patients compared to healthy individuals and reflecting the relatively low baseline levels in cancer patients. Some patients showed a rapid and massive increase of antibody titers (Fig. 1). The rapid increase and sufficient generation of antibody titers in patients undergoing immunotherapy with PD-1 blocking agents is in clear contrast to previously reported lower antibody titers in cancer patients undergoing cytotoxic chemotherapy [22–26]. These results raise interesting questions regarding the use of PD-1 blockade in a non-systemic application as a vaccine adjuvant. In addition, PD-1 seems to play a role in defective immune responses during viral respiratory infections [11, 12, 28]. In addition, a preclinical analysis in rhesus macaques has shown an enhanced frequency of antiviral T cells upon simian immunodeficiency virus (SIV) vaccination after PD-1 blockade [29] and immune response to herpes virus infection was enhanced upon PD-L1 inhibition in mice [30]. Finally, mice lacking PD-L1 on hematopoietic cells have an increased immune response to an infection with the lymphocytic choriomeningitis virus (LCMV) [31]. Thus, improved responses against viral antigens in patients undergoing PD-1 inhibition are not unexpected. However, in the recently presented retrospective INVIDIa study, a higher incidence of seasonal influenza was reported in patients undergoing immune checkpoint inhibitor therapy [32]. Interestingly, patients receiving vaccination and/or developing influenza infection showed a better overall survival. This finding is in accordance with our NSCLC cohort of 16 patients, in which the median OS is not reached after a follow-up period of more than 3 years (37.5 months).

We observed a significant rate of irAEs following vaccination in the long-term clinical course. The observed frequency was significantly higher than those published as safety data in PD-1 checkpoint blockade trials [15–17]. Patients included in immune checkpoint blockade trials were carefully screened and those at elevated risk for autoimmune disease were excluded. However, safety data from the Italian expanded access program with less stringent inclusion criteria than prospective landmark trials and being similar to daily practice showed comparable rates of irAEs (all grades 29%, grade 3/4 6%) as reported in phase III trials [33]. In an unselected non-study population of 40 metastatic NSCLC patients undergoing immune checkpoint inhibition at our center and not being vaccinated we observed a similar frequency of irAEs compared to a selected trial population (all grades 25.5%, grade 3/4 at 9.8%) and significantly different from the rates observed in vaccinated patients in this study. Severe irAEs were found at a low rate with an average risk in a recent meta-analysis for severe colitis at 1.5%, severe hepatitis/transaminitis at 1.5%, severe dermatitis at 1.1%, hypothyroidism at 0.3% and severe pneumonitis at 1.1% [4]. Other studies and case series demonstrated similar frequencies [15]. Although being a small study, our finding that 52.2% of previously vaccinated patients developed any grade of irAEs and 26.1% had a severe complication of PD-1 blockade raises important concerns about the safety of applying the seasonal influenza vaccination to patients undergoing cancer immunotherapy. It is important to acknowledge that patients responding to immunotherapy were likely overrepresented in our analysis due to a selection bias for patients that were treated for relatively longer times with a PD-1 inhibitor. This bias could also potentially select for patients that have an increased propensity for auto-immune side effects. Combination immunotherapy with blockade of CTLA-4 and PD-1 is approved for metastatic melanoma and is under investigation in several other indications [34]. This combination immunotherapy induces irAEs of any grade in a vast majority of treated patients and Grade 3–4 irAEs in over 50% of patients [34]. It is conceivable that combination immunotherapy has even a higher risk for side effects when combined with vaccinations and safety should be investigated in this patient population. For prophylactic vaccination in patients undergoing immune checkpoint inhibition, safety profiles for different vaccines has to be elucidated. In a retrospective analysis 30 of 108 patients (mainly melanoma) treated with immune checkpoint inhibitors received a total of 53 prophylactic vaccinations (influenza, pneumococcal and others) [35]. The authors did not find a higher rate of all grade irAEs in the vaccinated cohort; G3/4 irAEs were not reported separately.

The exact pathomechanism of irAEs after checkpoint blockade and how the breakdown of tolerance towards self-antigens exactly works in patients with irAEs is not completely understood [36, 37]. Most data are derived from preclinical models and correlative human studies. How the combination of prophylactic vaccination and PD-1 blockade could increase irAEs also remains speculative. The physiological role of the PD-1/PD-L1 pathway is to mediate peripheral tolerance of T cells and inhibition of immune checkpoints could break such tolerance [36, 38]. A recent report in a mouse model has provided evidence that PD-1 blockade together with a viral-based vaccination mediated infiltration of central memory T cells into the tissues, which could also induce auto-reactive immune responses [39]. Studies in patients treated with checkpoint inhibitors show an expansion of auto-reactive T-cell clones upon treatment with checkpoint inhibitors that can also be found in the peripheral blood in patients with irAEs [40, 41]. Identification of T-cell clones by sequencing of the complementarity-determining regions 3 (CDR3) of the T-cell receptor (TCR) beta chain has also shown similar clones to be present in auto-immune lesions in cases of myocarditis compared to those found in the primary lesion or pneumonitis [42, 43]. These findings support a hypothesis that shared antigens in the tumor and the irAE-affected organ can lead to auto-immune disorders by cross-presentation of such shared antigens [36]. Another potential mechanism is the exacerbation of previously subclinical auto-immune syndromes [44, 45]. We have described a case, in which anti-endothelial antibodies were already present before the initiation of PD-1 blockade and upon treatment the patient developed a cerebral vasculitis with necrosis of brain tissue [45]. An additional postulated mechanism of irAE induction is via epitope spreading during checkpoint blockade [36]. It could be speculated that PD-1 blockade together with vaccination – in particular in conjunction with a strong vaccine adjuvant – could boost the breakage of tolerance by enhancing one or several of above mentioned mechanisms associated with irAEs in patients. Moreover, since T cells show cross-reactivity to different antigen-MHC complexes, auto-immunity and irAEs could also be the result of TCR binding degeneracy [46] and cross-reactivity of T cells stimulated by the protein contained in the influenza vaccine to self-peptide-MHC complexes.

Therapeutic vaccination for cancer is currently tested in many clinical trials together with immune checkpoint inhibitors [9, 47]. This is based on preclinical models that have shown clear synergy between checkpoint blockade and vaccination [48–51]. Current strategies involve therapeutic vaccination with tumor epitopes – most often neoantigens – together with PD-1 or PD-L1 blocking antibodies [9, 52–54]. Our findings suggest that combination of therapeutic vaccination with checkpoint blockade could not only increase anti-tumor efficacy but also the rate of irAEs. Ongoing trials will provide more information on the toxicity of vaccine combinations with immune checkpoint blockade.

This study has clear limitations and further investigations are warranted. The small number of patients analyzed precludes a definitive statement on the safety of influenza vaccination in patients undergoing cancer immunotherapy. A larger cohort needs to be analyzed to advise for or against vaccination of patients that recently received therapy targeting the PD-1/PD-L1 axis. Moreover, predictions for newer therapeutic strategies that include immunotherapy cannot be made on the basis of this analysis. In particular, patients receiving combination immunotherapy including the combination of CTLA-4 and PD-1 inhibitors were not analyzed and the risk for adverse events should be investigated separately in this patient population. Although the observed rate of irAEs in our cohort is of concern, we believe that there is particular concern for patients with lung cancer under immunotherapy for severe complications of an influenza infection including pneumonia and respiratory failure because of concomitant structural lung disorders [55]. Some of these patients had prior resection of lung lobes or even a pneumonectomy and have therefore limited reserves due to impaired lung capacity. Moreover, the unexpectedly long survival of NSCLC patients in this cohort warrants further investigations in prospective clinical trials to understand if prophylactic vaccination may improve the outcome of cancer patients undergoing immune checkpoint blockade. When weighting benefit and potential risk of seasonal influenza vaccination for patients undergoing single-agent PD-1 or PD-L1 blockade – in particular those with lung cancer – we currently advice to make an individual decision against or for an influenza vaccination until results from larger cohorts are available.

This is the first analysis demonstrating adequate humoral immune response of a trivalent, inactivated, non-adjuvanted influenza vaccine in patients that were treated with a PD-1/PD-L1 blocking agent. However, there might be the potential of a higher rate of irAEs induced by immune checkpoint inhibitors in patients undergoing influenza vaccination.