Clinical effectiveness of influenza vaccination and infectionoutcome in cancer patients treated with checkpoint inhibitors

(1)

Örebro University

School of Medical Sciences Degree project, 15 ECTS January 2020

Clinical effectiveness of influenza vaccination and infection

outcome in cancer patients treated with checkpoint inhibitors

Version 2

Author: Alice Gustavsson Supervisor: Antonis Valachis MD, PhD, Department of Oncology Örebro University Hospital, Region Örebro county, Sweden

(2)

Abstract

BACKGROUND: In patients with cancer, influenza vaccine (flu-V) is recommended in an effort to prevent complications related to influenza syndrome (IS) and is regarded as effective in cancer patients during chemotherapy. However, the effectiveness of vaccination in patients treated with checkpoint inhibitors (CKI), has not been adequately studied.

AIM: To investigate the clinical effectiveness of influenza vaccination in patients treated with CKI and the clinical outcome of infection.

METHOD: A retrospective cohort study was conducted as a review of electronic medical records of 130 patients treated with anti-PD-1/PD-L1 due to advanced cancer at the University Hospital of Örebro (USÖ) during the time frame 2016-2019.

RESULTS: In the total cohort of 130 patients treated with CKI, 32 patients (24.6%) received flu-V. The duration of hospitalization due to influenza was 2.5 days as a median value for the vaccinated group, while the unvaccinated population had a median value of 11.0 days (p-value= 0.006). We did not observe significant differences in frequency of IS or rate of influenza-related hospitalizations comparing the two patient groups, with the vaccinated subgroup’s occurrence of IS and hospitalization being 18.8% and 12.5%, versus the non-vaccinated group’s occurrence of 11.2% and 7.1% (p-value= 0.273 and 0.344, respectively). CONCLUSION: Flu-V might reduce duration of hospitalization due to influenza disease in patients receiving CKI treatment, although no differences in frequency of IS and rate of hospitalization were observed. Our results are hypothesis-generating and additional studies are needed to re-evaluate the current vaccination protocol for this patient population.

(3)

Abbreviations

CKI: checkpoint inhibitors

CTLA-4: cytotoxic T-lymphocyte-associated protein 4 flu-V: influenza vaccine

IS: influenza syndrome

NSCLC: non-small cell lung cancer OS: overall survival

PD-1: programmed cell death protein 1

PD-L1: programmed cell death protein-ligand 1 USÖ: University Hospital of Örebro

(4)

The importance of infection prevention in cancer patients due to tumor- and treatment-induced immunosuppression has resulted in recommendations of vaccines for this population [1,2]. Viral infections often cause high mortality and morbidity rates, with previous estimated mortality rate of 9% from influenza-related infections in cancer patients hospitalized due to influenza, and avoidance of such is crucial in patients with impaired immunity [3].

The recent development of CKI (antibodies targeting the inhibitory CTLA-4/CD80/CD86 or PD-1/PD-L1 pathways) has led to prolonged survival and generated durable responses in patients with advanced cancer disease, such as metastatic melanoma, renal carcinoma, non-small cell lung cancer (NSCLC), urothelial cancer, head and neck carcinoma and Hodgkin lymphoma. The PD-1 or PD-L1 signaling cascade is an important mechanism in immune homeostasis and is critical in terminating immune responses, to safeguard against severe tissue destruction by T-cells [4–9]. By blocking the PD-1/PD-L1 axis with CKI, the T-cell activation will be enhanced, increasing the desired cancer-specific immunity. Notably, a potential effect of blocking the PD-1/PD-L1 axis will also be an increase in virus-specific immunity by decreasing the PD-L1 mediated inhibition of T-cell response normally seen in influenza virus infections [5,6].

Cancer patients with advanced disease or those treated with anti-neoplastic therapy are recommended seasonal flu-V, due to their vulnerable and immunosuppressed state [7]. Lung cancer patients are especially susceptible to complications owing to their concurrent lung disease [8]. Nevertheless, patients with a greater degree of immunosuppression might be less likely to respond to a vaccine [9]. Several studies have been conducted to evaluate the immunologic response after flu-V for patients with classical cytotoxic therapy, proving that the flu-V is safe, although showing a reduced efficacy of mounting a humoral immune response [10–13]. However, there is limited evidence on the clinical efficacy of flu-V for patients undergoing CKI, and the question has been raised if flu-V in advanced cancer patient receiving CKI may be clinically ineffective [10–12].

In fact, a recent prospective study suggests that flu-V, in patients during CKI, induces adequate serological protection, but with a higher rate of immunological toxicity [14]. In accordance to this observation, Gopalakrishnan et al found a reduced rate of hospital admissions due to influenza after flu-V in 534 patients treated with CKI [15]. However, the

(6)

INVIDIa study revealed a higher rate of IS for the vaccinated group compared to the unvaccinated one [16]. The higher incidence of IS in CKI-treated patients with flu-V administered could be explained by an intensified atypical T-cell reaction, after being activated by the initial influenza infection, causing excessive tissue damage and the development of flu symptoms, [17,18] and even more so in vaccinated patients due to the accumulation of antigenic stimulus [16].

Despite the conflicting evidence on the clinical effectiveness of flu-V in CKI-treated patients, the clinical course of IS, namely the severity, the risk for complications or death due to influenza, seems to be similar regardless of flu-V-status, although only investigated by one earlier study [16]. It is, therefore, essential to further investigate the clinical effectiveness of flu-V in CKI-treated patients as well as the clinical outcome of IS.

2. AIM

To investigate the clinical effectiveness of influenza vaccination in terms of frequency of influenza-like symptoms and hospital admissions due to influenza in patients treated with CKI, as well as the clinical outcome of infection (duration of hospitalization, mortality due to influenza infection).

3. MATERIAL AND METHODS 3.1 Study design and study cohort

This was a retrospective cohort study including patients treated with anti-PD-1/PD-L1, during the time frame 2016-2019 at the USÖ. Inclusion criteria were: (1) patients with advanced cancer treated with any anti-PD-l/PD-L1, irrespective of treatment line; (2) patients should have received at least one infusion with anti-PD-1/PD-L1. Patients receiving combined immunotherapy (Ipilimumab and Nivolumab), patients receiving CKI as a part of a clinical trial, patients treated with Ipilimumab (anti-CTLA-4) and prior exposure to CKI were all excluded.

(7)

3.2 Data collection and definitions

The eligible patients were collected in a pseudonymized database, with pre-collected data including start- and end date of treatment with CKI and vaccination status.

The following data were collected from the electronic medical records: documented IS, need of hospitalization, length of hospitalization, outcome of infection during hospitalization (cure, need of ICU-treatment or death). The time frame for identification of potential IS-episodes was start date of treatment with anti-PD-1/PD-L1 until end date of treatment plus six months to cover potential remaining effects of antibodies.

IS was defined either as a laboratory-confirmed influenza (using PCR), acute respiratory infection or influenza-like illness (cough or sore throat, and measured fever ³ 38°C or self-reported feverishness/chills) without any other documented infectious etiology.

3.3 Outcomes

The primary outcome was the frequency of IS (overall and those needed hospitalization) comparing the vaccinated and non-vaccinated patients treated with anti-PD-1/PD-L1. Secondary aims were the length of hospitalization and the IS-related mortality between the two patient groups.

3.4 Statistical analyses

Categorical variables were summarized with number and percentage whereas continuous variables with median value and interquartile range (IQR).

For bivariate analyses, chi-square test or Fisher´s exact test was used when comparing

categorical variables whereas Mann-Whitney test (all continuous variables were non-normally distributed) was used when comparing continuous and categorical variables. Two-sided p-value with a statistically significance when p-p-value < 0.05 was used in all analyses. The statistical analyses was performed with SPSS v 20.

(8)

3.5 Ethical considerations

The permission of the electronic medical record review was granted 2019.06.04 by the Head of the Department of Oncology USÖ. This study is a part of a larger project that is approved by the Swedish Ethical Review Authority (dnr 2019-02469).

Since the project was based on retrospective review of electronic medical records of patients already treated due to cancer, there was no need for informed consent from the patients whose journals have been reviewed.

All collected data were included in a pseudonymised database to protect patients’ identity, and with coding key kept where ineligible access was prevented.

A potential risk would be a leakage of the coded, collected data, which eventually could lead to identifying individuals and sensitive information in the hands of unauthorized people. We judge the risk to be minimal due to the fact in how we kept all data.

4. RESULTS

4.1 Patient characteristics

This study cohort enrolled 130 advanced cancer patients with ongoing or terminated treatment with CKI, at the department of Oncology, USÖ. Figure 1 shows the procedure of selecting the study population satisfying inclusion criteria.

(9)

Patient characteristics are reported in Table 1. Thirty-two patients (24.6%) received flu-V whereas the remaining group of 98 patients (75.4%) did not receive the injection, constituting the control group.

The follow-up after initiation of CKI treatment was significantly longer for the vaccinated subgroup with a median of 17 months compared to 11 months for the unvaccinated group (p-value= 0.001). The median age was significantly lower for the flu-V group (60 years) versus the control group with 70 years as median (p-value= 0.002).

Table 1 -Patient characteristics

Variables _{(100%), n= 130}Total cohort Vaccinated group _{(%), n= 32} Unvaccinated group _{(%), n= 98} P-value

Age in years, median 69.0 (28-89) 60.0 (28-73) 70.0 (30-89) 0.002

Gender 0.066 Male 67 (51.5%) 21 (65.6%) 46 (46.9%) Female 63 (48.5%) 11 (34.4%) 52 (53.1%) Type of malignancy 0.749 Melanoma 47 (36.2%) 12 (37.5%) 35 (35.7%) NSCLC 60 (46.2%) 13 (40.6%) 47 (48.0%) RCC 11 (8.5%) 4 (12.5%) 7 (7.1%) Urothelial cancer 10 (7.7%) 2 (6.3%) 8 (8.2%) HNC 2 (1.5%) 1 (3.1%) 1 (1.0%) De Novo metastatic 0.094 No 77 (59.2%) 23 (71.9%) 54 (55.1%) Yes 53 (40.8%) 9 (28.1%) 44 (44.9%) CKI 0.150 Nivolumab 79 (60.8%) 23 (71.9%) 56 (57.1%) Pembrolizumab 43 (33.1%) 9 (28.1%) 34 (34.7%) Atezolizumab 8 (6.2%) 0 (0.0%) 8 (8.2%) PS 0.857 0-1 _{107 (82.3%)} _{26 (81.3%)} _{81 (82.7%)} 2-3 _{23 (17.7%)} _{6 (18.8%)} _{17 (17.3%)} CCI 0.124 0 11 (8.5%) 2 (6.3%) 9 (9.2%) 1 11 (8.5%) 5 (15.6%) 6 (6.1%) 2 _{27 (20.8%)} _{3 (9.4%)} _{24 (24.5%)} > / = 3 _{81 (62.3%)} _{22 (68.8%)} _{59 (60.2%)}

Values are presented as n (%), median or range (min-max). RCC, renal cell carcinoma; HNC, head and neck cancer; PS, performance status; CCI, Charlson comorbidity index

(10)

4.2 The clinical effectiveness of influenza vaccination

We found no statistically significant difference in IS frequency between the two groups (p-value= 0.273) (Figure 2).

Likewise, no statistically significant difference could be seen in the influenza-related hospitalizations comparing the two groups (p-value= 0.344) (Figure 2).

4.3 The clinical outcome of IS

No cases of mortality due to influenza infection were documented during the follow-up. The median duration of hospitalization due to IS differed between the two groups, with a median value of 2.5 days for the vaccinated group and 11.0 days for the unvaccinated one (p-value= 0.006) (Figure 3). 18.8% 11.2% 12.5% 7.1% 0% 2% 4% 6% 8% 10% 12% 14% 16% 18% 20% Flu-V + FluV -Va lu e ( % )

Occurrence of IS and hospitalization for patients vaccinated and not vaccinated with flu-V

Influenza disease Hospitalization

Figure 2. Occurrence of IS and hospitalization for patients vaccinated and

unvaccinated with flu-V, value in %. Flu-V+ = vaccinated cohort Flu-V-= unvaccinated cohort

p= 0.273 p= 0.344 20.0 18.0 16.0 14.0 12.0 10.0 8.0 6.0 4.0 2.0 0.0 IS

(11)

5. DISCUSSION

In our study cohort of 130 CKI-treated patients, we investigated the clinical effectiveness of flu-V in terms of frequency of influenza-like symptoms and hospital admissions due to

influenza, and the clinical outcome in the event of infection. We found that flu-V may shorten hospital length of stay due to IS, but did not significantly reduce frequency of IS and need for hospital admissions. The only difference found was therefore in length of inpatient care, and could potentially been caused by an increased amount of complications for the non-vaccinated group.

Only one prior study investigated the clinical course of IS with and without flu-V for CKI-treated patients. In fact, the INVIDIa study [16] included 300 patients CKI-treated with CKI, and found that flu-V in this patient cohort did not improve the clinical outcome, with similar duration of symptoms and duration of hospitalization between vaccinated and unvaccinated patients observed. An explanation for these partially contradictory results could be the significant difference between the median age of vaccinated (60 years) and unvaccinated (70 years) patients in our study cohort, where the younger subgroup might have a better physical condition initially, managing a potential IS and the recovery from it better and therefore improving the clinical outcome of infection in terms of duration of hospitalization. Nevertheless, the improved clinical outcome after vaccination in our study cohort, might

Da ys o f ho sp it al iz at io n 0 5 10 15 20 25

Lenght of hospitalization for patients vaccinated and not vaccinated with flu-V

Flu-V+

Flu-V-Figure 3. Length of hospitalization for patients vaccinated and unvaccinated

with flu-V, median value in days. Flu-V+ = vaccinated cohort Flu-V-= unvaccinated cohort MV: median value (days)

MV: 11.0

MV: 2.5

(12)

actually be an indication that the administration of flu-V can accelerate the viral clearance and therefore reduce the length of inpatient care. With CKI hypothetically being responsible for a greater immunocompetence against viral infections [16], the addition of flu-V could

potentially result in an even stronger immune response against viral antigen.

In accordance to our results, no case of influenza-related death was reported in the previous study [16], and therefore flu-V’s effect on mortality rate cannot be evaluated. Interestingly, a positive correlation between flu-V during CKI-treatment and overall survival (OS) has been previously reported [14–16,19]. This prolonged survival could be explained with previously presented models demonstrating clear synergistic anti-tumor activity between CKI and vaccination, [20–23] acting as analogous stimuli and therefore resulting in an overall better treatment outcome [16]. The investigation of flu-V and OS was beyond the scope of this study, but we are planning to increase our study cohort by including patients from other study sites, enabling the analysis of OS in the near future.

The key strengths of this study were the inclusion of consecutive patients treated only with PD-1/PD-L1 inhibitors to avoid the different mechanism of actions with other

immunotherapies as anti-CTLA-4, and the uniformity and continuity on the electronic medical records due to the inclusion of patients from the same hospital. This also composes one of the limitations of our study, since the study cohort reflects a single-center treatment practice of CKI-treatment and vaccine administration.

There are several limitations to this study that should be taken into account when interpreting the results. First, its retrospective observational design is prone to well-described biases. Second, influenza infection could not be confirmed by PCR in most of the cases, representing a source of information bias considering the subjectivity of symptoms in our inclusion

criteria. In addition, there is a risk for missing unrecorded or undiagnosed cases of IS during the study period, which may lead to an underestimation of frequency of IS. Furthermore, our study results were based on bivariate analyses rather than multivariate models; as a result, the potential effect of confounding factors as age between vaccinated and unvaccinated patients has not been investigated. Lastly, the number of patients in our study cohort as well as the number of events are relatively small, hence the statistical power might be low. This could carry a risk of type II error, a limitation that implicitly could be avoided by increasing the

(13)

Considering the risk for potential life-threatening viral infections in patients with impaired immunity [3], patients undergoing the relatively new CKI-treatment constitutes an important population for further research to enable adjustment and development of flu-V administration guidelines for this patient group. Several aspects on the potential interaction between flu-V and CKI treatment including the flu-V’s effect on IS, rate and duration of hospitalization, the effect on anti-neoplastic therapy of CKI and the frequency of immune-related adverse events should be further investigated and considered for a more individualized vaccine program.

6. CONCLUSION

Flu-V might reduce duration of hospitalization due to influenza disease in patients receiving CKI treatment, although no differences in frequency of IS and rate of hospitalization were observed. Our results on duration of inpatient care are hypothesis-generating and additional studies are needed to further investigate the clinical effectiveness of flu-V and the clinical outcome of influenza disease to advice for or against vaccination for this patient population.

7. ACKNOWLEDGEMENT

I would like to extend my thanks to my supervisor Antonis Valachis for his support and encouragement during this project. With a never-ending patience, he has continuously been giving me feedback, sharing his knowledge and helping me interpreting our results.

8. REFERENCES

1. Pedrazzoli P, Baldanti F, Donatelli I, Castrucci MR, Puglisi F, Silvestris N, et al. Vaccination for seasonal influenza in patients with cancer: recommendations of the Italian Society of Medical Oncology (AIOM). Ann Oncol Off J Eur Soc Med Oncol 2014; 25:1243–7.

2. Grohskopf LA. Prevention and Control of Seasonal Influenza with Vaccines: Recommendations of the Advisory Committee on Immunization Practices — United States, 2019–20 Influenza Season. MMWR Recomm Rep [Internet] 2019 [cited 2019 Nov 22]; 68. Available from: https://www.cdc.gov/mmwr/volumes/68/rr/rr6803a1.htm

3. Cooksley CD, Avritscher EBC, Bekele BN, Rolston KV, Geraci JM, Elting LS. Epidemiology and outcomes of serious influenza-related infections in the cancer population. Cancer 2005; 104:618–28.

4. Chen L, Flies DB. Molecular mechanisms of T cell co-stimulation and co-inhibition. Nat Rev Immunol 2013; 13:227.

(14)

5. McNally B, Ye F, Willette M, Flaño E. Local Blockade of Epithelial PDL-1 in the Airways Enhances T Cell Function and Viral Clearance during Influenza Virus Infection. J Virol 2013; 87:12916–24.

6. Dirks J, Egli A, Sester U, Sester M, Hirsch HH. Blockade of programmed death receptor-1 signaling restores expression of mostly proinflammatory cytokines in anergic cytomegalovirus-specific T cells. Transpl Infect Dis Off J Transplant Soc 2013; 15:79–89. 7. Robin C, Beckerich F, Cordonnier C. Immunization in cancer patients: Where we stand.

Pharmacol Res 2015; 92:23–30.

8. Kohno S, Koga H, Oka M, Kadota J, Kaku M, Soda H, et al. The pattern of respiratory infection in patients with lung cancer. Tohoku J Exp Med 1994; 173:405–11.

9. Hibberd PL, Rubin RH. Approach to immunization in the immunosuppressed host. Infect Dis Clin North Am 1990; 4:123–42.

10. Choi DK, Fuleihan RL, Walterhouse DO. Serologic response and clinical efficacy of influenza vaccination in children and young adults on chemotherapy for cancer. Pediatr Blood Cancer 2016; 63:2011–8.

11. Nakashima K, Aoshima M, Ohfuji S, Suzuki K, Katsurada M, Katsurada N, et al. Immunogenicity of trivalent influenza vaccine in patients with lung cancer undergoing anticancer chemotherapy. Hum Vaccines Immunother 2017; 13:543.

12. Waqar SN, Boehmer L, Morgensztern D, Wang-Gillam A, Sorscher S, Lawrence S, et al. Immunogenicity of Influenza Vaccination in Patients With Cancer. Am J Clin Oncol 2018; 41:248–53.

13. Hakim H, Allison KJ, Van de Velde L-A, Tang L, Sun Y, Flynn PM, et al. Immunogenicity and Safety of High-Dose Trivalent Inactivated Influenza Vaccine Compared to Standard-Dose Vaccine in Children and Young Adults with Cancer or HIV Infection. Vaccine 2016; 34:3141–8.

14. Läubli H, Balmelli C, Kaufmann L, Stanczak M, Syedbasha M, Vogt D, et al. Influenza vaccination of cancer patients during PD-1 blockade induces serological protection but may raise the risk for immune-related adverse events. J Immunother Cancer 2018; 6:40. 15. Gopalakrishnan R, Johnson DB, York S, Neuss MN, Osterman TJ, Chism DD, et al. Impact

of the influenza vaccination on cancer patients undergoing therapy with immune checkpoint inhibitors (ICI). J Clin Oncol 2018; 36:3053–3053.

16. Bersanelli M, Giannarelli D, Castrignanò P, Fornarini G, Panni S, Mazzoni F, et al. INfluenza Vaccine Indication During therapy with Immune checkpoint inhibitors: a transversal challenge. The INVIDIa study. Immunotherapy 2018; 10:1229–39.

17. Hillaire MLB, Rimmelzwaan GF, Kreijtz JHCM. Clearance of influenza virus infections by T cells: risk of collateral damage? Curr Opin Virol 2013; 3:430–7.

(15)

19. Schenk EL. Clinical outcomes of patients on check point inhibitor therapy who receive routine vaccinations. J Clin Oncol 2017; 35:e14597–e14597.

20. Karyampudi L, Lamichhane P, Scheid AD, Kalli KR, Shreeder B, Krempski JW, et al. Accumulation of Memory Precursor CD8 T Cells in Regressing Tumors Following Combination Therapy with Vaccine and Anti-PD-1 Antibody. Cancer Res 2014; 74:2974– 85.

21. Duraiswamy J, Kaluza KM, Freeman GJ, Coukos G. Dual Blockade of PD-1 and CTLA-4 Combined with Tumor Vaccine Effectively Restores T Cell Rejection Function in Tumors. Cancer Res 2013; 73:3591–603.

22. Ha S-J, Mueller SN, Wherry EJ, Barber DL, Aubert RD, Sharpe AH, et al. Enhancing therapeutic vaccination by blocking PD-1–mediated inhibitory signals during chronic infection. J Exp Med 2008; 205:543–55.

23. Kamphorst AO, Ahmed R. Manipulating the PD-1 pathway to improve immunity. Curr Opin Immunol 2013; 25:381–8.

Clinical effectiveness of influenza vaccination and infectionoutcome in cancer patients treated with checkpoint inhibitors