Effect of antimicrobial therapy on progression-free survival of patients with non-small cell lung cancer receiving checkpoint inhibitor- and chemotherapy

Checkpoint inhibitor therapy (CPI) has significantly changed therapy in non-small cell lung cancer (NSCLC) in recent years. There are some data that the effect of CPI therapy is influenced by the microbiome. Little is known about the influence and timing of antimicrobial therapy (AMT) on the microbiome-mediated effect on CPI therapy. We retrospectively analysed 70 patients (age 68 ± 9.2 years) with NSCLC stage IV. Patients were treated according to the guidelines with either CPI alone (pembrolizumab, nivolumab, atezolizumab) or chemotherapy (platin doublet or docetaxel/nintedanib or pemetrexed). We registered patient’s characteristics including presence and timing of AMT. Group 1 consisted of 27 patients with AMT in the month before CPI- or chemotherapy, group 2 was 30 patients with AMT during CPI- or chemotherapy, and group 3 was 43 patients without AMT. Groups 1–3 showed comparable patient characteristics. Using cox-regression analysis, we found that AMT in the month before CPI resulted in a decreased progression-free survival (PFS) compared to patients with CPI and no AMT (14 ± 1.56 vs. 5 ± 0.99, p = 0.005, 95% CI: 0.13–0.67). In patients, who were treated with chemotherapy alone, there was no difference in PFS in those with or without AMT in the month before therapy (5 ± 0.99 vs. 6 ± 0.81 months, p = 0.3). Interestingly, AMT during chemotherapy or CPI therapy showed no effect on PFS. In a real-life setting, we found that AMT reduces PFS when given in the month before CPI therapy. AMT before chemotherapy and during CPI and chemotherapy seems not to influence PFS. The best PFS was seen in patients without AMT before CPI therapy. This implies the need for an even more restrictive use of AMT in the context of patients with NSCLC stage IV disease.


Background
Checkpoint inhibitor therapy (CPI) with PD-L1 or PD-1 inhibitors has significantly changed therapy in non-small cell lung cancer (NSCLC) in recent years. There are some data that the effect of CPI therapy is influenced by the microbiome. A gut dysbiosis due to antimicrobial therapy (AMT) is associated with a dysfunctional immune system (Pflug et al. 2016). In a mouse model, it has been shown that a dysbiosis in gut is associated with a worse outcome using antineoplastic therapy (Vetizou et al. 2015). Some studies showed negative effects of AMT on outcome in patients with cancer therapy. This was more pronounced in patients with CPI therapy, but also present under chemotherapy (Hekmatshoar et al. 2019).
In one large study with 1512 patients with NSCLC stage IV, Chalabi et al. (2020) saw a significantly shorter PFS and overall survival (OS) in patients, who were treated with AMT. In this study, there was no discrimination with respect to the timing of AMT. AMT could be given in the month before, during or the month after start of CPI-or chemotherapy. However, little is known about the influence and especially exact timing of AMT on the probable microbiome-mediated effect on CPI therapy. Therefore, the aim of our study was to analyse, if the effect of AMT is different, when given in the month before CPI therapy compared to AMT during CPI therapy.

Study design and participants
In a retrospective study design, we analysed 70 patients (age 68 ± 9.2 years) with NSCLC stage IV. All patients were treated in our lung cancer centre in the Florence-Nightingale hospital in Düsseldorf/Germany. We collected data from 06/2016 to 01/20.
We registered patient's characteristics including presence and timing of AMT. We accepted prior therapy lines according to guidelines. Some patients are listed twice due to their prior therapy lines.
We chose variables, based on their prognostic impact in NSCLC and risk factors for infections. This included age, sex, BMI, ECOG performance status, lung cancer pathology (squamous and non-squamous NSCLC), comorbidities (COPD, diabetes mellitus, and hypertension), PD-L1 status, AMT use and timing of AMT.
Group 1 consisted of 27 patients with AMT in the month before CPI-or chemotherapy, group 2 was 30 patients with AMT during CPI-or chemotherapy, and group 3 was 43 patients without AMT. Patient characteristics are shown in Table 1.
Information about timing of AMT and PFS were taken from our electronic patient database. Only patients with all available data about AMT and PFS were included in the analysis.
We did not include patients who were treated with a combination of CPI therapy and chemotherapy. The number of patients who matched criteria were too small in our time frame.

Statistics
Continuous variables are expressed as mean ± SD or median and compared using t test unless stated otherwise. Statistical analysis was performed using SPSS (Version 27, IBM, Armonk, NY, USA). Cox proportional-hazards regression was used to analyse the effect of several factors on progression-free survival in uni-and multivariable analyses.
Cox regression survival curves were generated to visualise the distribution of times from baseline to disease progression. All statistical tests were 2-tailed and a p value < 0.05 was considered statistically significant.

Results
We included 70 patients. We accepted prior therapy lines according to guidelines.
Some patients are listed twice due to their prior therapy lines.
57 were treated with CPI while 43 got chemotherapy. 10 patients with chemotherapy got AMT in the month before therapy and 10 patients during therapy. 17 patients with CPI were treated with AMT one month before therapy and 20 patients during therapy. At the time of final analysis, 60 patients died and 10 patients have still been treated with CPI.
Median PFS in group 1 (patients with AMT in the month before CPI-or chemotherapy) was comparable in those with CPI-and chemotherapy (5 ± 0.99 vs. 4 ± 0.63 months, p = 0.27).
Thus, CPI patients without AMT compared to CPI patients with AMT in the month before start of therapy showed significantly higher PFS (14 ± 1.56 vs. 5 ± 0.99, p = 0.005, 95% CI: 0.13-0.67). The Cox regression survival curve for PFS of patients with CPI and chemotherapy is shown in Fig. 1.
Patients with AMT before therapy suffered from a worse ECOG performance status.
Cox proportional-hazards regression showed no significant effect on PFS.
In addition to AMT in the month before CPI therapy, PFS was significantly negatively influenced by the presence of type II diabetes mellitus and might be positively influenced by PD-L1 status with a TPS > 50%.

Conclusions
The main finding of our study is that AMT before the start of CPI decreases PFS. Previous studies have shown the possible negative effect of AMT in patients with NSCLC and CPI. However, in these previous studies, there was no focus on the timing of AMT with respect to its effect. To the best of our knowledge, we are first to demonstrate, that especially AMT in the month before CPI results in a lower PFS, whereas AMT during CPI showed no influence on PFS.
AMT given in the month before CPI resulted in a median PFS of 5 months, which is comparable to the median PFS of chemotherapy. Interestingly, CPI patients without AMT had a remarkably higher median PFS of 14 months.
The hypothesis for the explanation of our findings is as follows: AMT leads to relevant short-term changes in the microbiome, which might negatively influence the immunological answer of CPI. Once the immune system und the influence of CPI has started to produce specific T-lymphocytes to attack tumour cells, it might be not so vulnerable anymore. This could explain, why AMT given during CPI therapy, which also causes dysbiosis, showed no significant effect in our study. The negative effect of diabetes, that we found, might also be linked to dysbiosis (Moreno-Indias et al. 2014) and could be associated with altered PFS under CPI therapy.
Our study has several limitations. First, it is a retrospective study with a moderate number of patients. Moreover, prior therapy lines were accepted. We did not analyse the microbiome and its possible changes under AMT during the study.
Patients with antimicrobial therapy have an increased risk of dying due to infection.
On the other hand, patients are only admitted to CPI therapy or chemotherapy, who have recovered from severe infection. Median PFS with chemotherapy was not altered by AMT as well. It is unlikely that treated infections cause such a decreased PFS of CPI therapy alone.
In conclusion, our data are hypothesis generating and need validation in a larger study collective.
Funding There was no funding.

Declarations
Conflict of interest All the authors have read the manuscript and agreed to publication. No author has got conflicts of interest.
Ethical approval According to an ethics committee, approval is not needed.