Abstract
Aim: To investigate potential association between administration of corticosteroids, antibiotics, probiotics, proton pump inhibitors, non-steroidal anti-inflammatory drugs (NSAID), statins and metformin and outcome in patients with non-small cell lung cancer (NSCLC) treated with nivolumab. Patients and Methods: A total of 224 patients with advanced NSCLC treated at nine comprehensive cancer centers were analyzed in this national retrospective study. Survival statistics were evaluated using Kaplan–Meier method and Cox analysis. Results: Only corticosteroid use had a significant negative effect on the objective response rate. In the univariate analysis, there was no significant effect of the studied concomitant medications on the efficacy of nivolumab. In a subsequent multifactorial analysis, a possible positive effect of the concomitant use of NSAID at the initiation of nivolumab treatment was revealed. Conclusion: The results of the present retrospective exploratory analysis underscore the importance of knowing the exact type of concomitant medication, the route of administration, the dose of medication, and the region of the ongoing study. The present data indicated a significantly higher rate of progression in patients treated with corticosteroids and the possible positive effect of NSAID use at the initiation of nivolumab treatment.
Nivolumab is a human monoclonal antibody to programmed cell death protein 1 (PD1) that represents a new therapeutic option in the second-line treatment of advanced non-small-cell lung cancer (NSCLC). Improved efficacy with a more favorable adverse event profile has been documented for nivolumab compared to docetaxel in phase III trials. However, the objective response rate of nivolumab monotherapy is only about 20%, with the disease control rate reaching approximately 50% (1, 2). Therefore, many patients do not benefit from nivolumab treatment and, considering the cost/efficacy ratio, identification of predictive parameters that would aid in patient selection for immunotherapy remains a topic of unmet medical need. Predictive biomarkers currently used include expression of programmed death-ligand 1 (PD-L1) and tumor mutation burden (3, 4). However, the predictive value of these biomarkers is far from being perfect, and many other, often unknown, parameters affect the efficacy of nivolumab, including biomarkers of inflammatory response and concomitant medications, as demonstrated in recent studies (5-8).
Corticosteroids are commonly used in patients with NSCLC to treat a range of conditions, e.g. symptomatic brain metastases and anorexia (9, 10). However, the immunosuppressive activity of corticosteroids obviously inhibits T-cell function (11). Corticosteroids can also modify the microbiome (12). Therefore, corticosteroid use may potentially impair the efficacy of immunotherapy. This hypothesis was first examined with regard to corticosteroid use during the treatment of immune-related adverse events, and was not confirmed (13), possibly due to the fact that the effect of corticosteroids on T-cell function is determined not only by the dose, but also by the timing of administration (14). Therefore, after T-cell activation induced by immunotherapy, effector cells may be protected from steroid-induced apoptosis (8). On the other hand, there are data indicating a deleterious effect of corticosteroid administered at the time of immunotherapy initiation on the efficacy of PD1/PD-L1 inhibitors (8, 15). However, a systematic review did not indicate a negative effect of corticosteroids on the outcome of treatment with PD1/PD-L1 inhibitors (16), and this hypothesis needs further investigation.
Antibiotics, probiotics and proton pump inhibitors (PPI) are also among the drugs that can affect the microbiome (17). Experimental work in mice highlighted the role of the gut microbiome in response to immunotherapy (18, 19). A subsequent study in humans also indicated a possible negative impact of antibiotics on the efficacy of immunotherapy (20). However, Kaderbhai et al. published discordant results (7). Similarly, the study reported by Hakozaki et al. did not demonstrate a significant effect of antibiotics on overall survival (OS) after nivolumab treatment in a multivariate analysis (17). Ahmed et al. hypothesized the possible existence of a difference between using broad-spectrum and narrow-spectrum antibiotics (21). Thus, the general effect of antibiotics on the efficacy of immunotherapy remains controversial.
PD1/PD-L1 blockade has also interactions with metabolic pathways (22-24). A possible effect of metformin (an antidiabetic agent) on potentiation of PD-L1 blockade was observed in a preclinical study (25). Metformin may also have a direct antitumor activity that has also been described for statins (26, 27). An off-target effect of the combination of metformin with statins resulting in longer survival in patients with NSCLC treated with chemotherapy was noted by Lee et al. (15). Similarly, a retrospective analysis reported by Omori et al. indicated a potential benefit of administering statins with nivolumab (8), with significantly improved overall response rate (ORR), but only a trend for improved OS. Therefore, these observations need further confirmation.
Non-steroidal anti-inflammatory drugs (NSAID) act, among other pathways, as cyclooxygenase 2 (COX2) inhibitors (28-30). COX2 is not detectable in most normal tissues, but is rapidly induced by inflammatory stimuli (31). COX2 is expressed across a range of solid tumor types and COX2 activity correlates with invasiveness and prognosis, reflecting an important role of COX2 in carcinogenesis and tumor progression (32). Prima et al. observed that COX2 and the prostaglandin E2 (PGE2) pathway regulate PD-L1 expression in tumor-associated macrophages and myeloid-derived suppressor cells (33). Moreover, an association between COX2 and PD-L1 expression was described in breast cancer and melanoma (34-36). A similar association was subsequently observed in colorectal cancer, where aspirin use was associated with significantly improved survival in patients with low PD-L1 expression (36). In addition, an association between COX2 expression and PD-L1 was also described in NSCLC, although the treatment of lung cancer cell lines with a COX2 inhibitor alone had no impact on PD-L1 expression in this study (37). Therefore, it is difficult to predict whether the concomitant use of NSAID (COX2 inhibitor) with nivolumab (PD-L1 inhibitor) might result in a synergistic effect.
Given this background, the aim of the present study was to investigate potential associations between corticosteroids, antibiotics, probiotics, PPIs, NSAIDs, statins and metformin and outcome in patients with NSCLC treated with nivolumab.
Patients and Methods
Study design and treatment. Clinical data of patients with cytologically or histologically confirmed advanced NSCLC treated with nivolumab were retrospectively analyzed. The patients were treated in the first- or higher line of treatment at nine oncology and pneumo-oncology Departments in the Czech Republic between 2015 and 2019. Nivolumab was administered intravenously at the approved doses of 3 mg/kg or a flat dose of 240 mg every 2 weeks. The treatment was administered until progression or unacceptable toxicity for a maximum of 2 years. In the case of treatment-related toxicity, administration of corticosteroids with/without interruption of nivolumab were recommended. Clinical follow-up including physical examination, chest X-ray and routine laboratory tests were performed at least every 4 weeks. Computed tomography (CT) or positron-emission tomography (PET)/CT were performed at regular intervals according to the local standards or when progression was suspected based on clinical or chest X-ray examination. Concomitant medications investigated in the present study included corticosteroids, antibiotics, probiotics, PPIs, NSAIDs, statins and metformin. These concomitant medications were followed 1 month before and 1 month after the initiation of nivolumab treatment. Patients were considered treated with the above-mentioned agents if the treatment occurred any time during this 2-month period, and treatment beyond this period was considered only if started within this period to avoid potential bias in patients treated with nivolumab for prolonged duration. A national register TULUNG, a non-interventional post-registration database of epidemiological and clinical data of patients with advanced-stage NSCLC treated with targeted or biological therapies in the Czech Republic, served as the data source. The patients gave their informed consent to be included in this database and for use of these data for scientific purposes.
Statistical methods. Standard frequency tables and descriptive statistics were used to characterize the sample data set. The ORR was defined as the best response according to the Response Evaluation Criteria in Solid Tumors (RECIST 1.1) (38). ORR was estimated using Pearson chi-square test and, in the case where the assumptions of this test were not met, Fisher's exact test. Progression-free survival (PFS) and OS were estimated using Kaplan–Meier method and all point estimates were accompanied by 95% confidence intervals. PFS was determined from the date of nivolumab treatment initiation until the date of first documented progression (as per RECIST 1.1) or death. OS was determined from the date of nivolumab treatment initiation until the date of death due to any cause. Patients whose disease had not progressed or who had died were censored at the date of last follow-up. Statistical significance of the differences in Kaplan–Meier estimates was assessed using the log-rank test. Finally, multivariate Cox proportional hazards model was used to evaluate the effect of all potential prognostic factors on the survival measures. For decision on statistical significance, α=0.05 was used.
Results
Patient characteristics. In total, 224 patients, 133 male and 91 females, with a median age of 67 years, were included in the present retrospective analysis. Among 22 patients treated with corticosteroids, 11 were treated for longer than 2 months followed-up in this study, while the duration of corticosteroid therapy in the remaining 11 patients was shorter (median of 4 days). The median duration of antibiotic therapy among patients treated with antibiotics was 7 days. The principal reason for antibiotic treatment was respiratory infection in 21 patients (77.8%). Other indications for antibiotic use included phlegmon (two patients), prophylaxis (two patients), urinary tract infection (one patient) and cholangitis (one patient). The median length of probiotic treatment among patients using probiotics was 7.5 days. Among patients using PPIs, most (59; 92.2%) were treated for more than 2 months in this retrospective analysis. Among the patients using NSAIDs, the majority (40; 87.0%) used NSAIDs for more than 2 months and all patients used statins or metformin more than 2 months. The baseline patient characteristics, and characteristics of the use of studied concomitant medications, are summarized in Tables I and II. ORR. Only 181 patients with a known response were included in this analysis. Only the association between ORR and corticosteroids was significant (p=0.019): more progressive disease was found in corticoid users. Among the patients treated with corticosteroids (n=16), partial response was observed in four (25%), stable disease in five (31%), and progressive disease in seven patients (44%). Among patients not treated with corticosteroids (n=165), partial response was seen in 27 (16.4%), stable disease in 106 (64.2%) and progressive disease in 32 (19.4%). There was no relation between ORR and administration of antibiotics (p=0.359), probiotics (p=0.515), PPIs (p=0.655), NSAIDs (p=0.504), metformin (p>0.999) or statins (p=0.875).
Univariate analysis of PFS and OS. There were no significant differences in OS or PFS between the patients taking or not taking the studied concomitant medications (Table III).
Multivariate Cox proportional hazards model. The Cox model only included patients with known value of C-reactive protein (CRP) (210 patients). Age, Eastern Cooperative Oncology Group performance score, histology and CRP concentration were found to be significant predictors of OS. An increase in patient age by 1 year, reduced the risk of death by 1.038-fold. Patients with ECOG PS 1 had a 1.980-fold higher risk of death than those with ECOG PS 0. Patients with squamous cell carcinoma had 1.733-fold lower risk of death than those with adenocarcinoma. Each CRP increase of 1 mg/dl increased the risk of death by 1.016-fold.
In the Cox model for PFS, age, histology, CRP concentration and treatment with NSAID were significant predictors. An increase in patient age by 1 year, reduced the risk of death by 1.028-fold. Patients with squamous cell carcinoma had a 1.689-fold lower risk of progression than those with adenocarcinoma. CRP increase of 1 mg/dl increased the risk of progression by 1.005-fold. Patients not taking NSAIDs had a 1.596-fold higher risk of progression than those taking an NSAID (Table IV).
Discussion
The data from the present retrospective analysis indicate an effect of corticosteroids on ORR and possible impact of NSAID on PFS in patients with NSCLC treated with nivolumab. Some of these results are in conflict with prior reports. We believe that there are more factors that can explain these differences as discussed below.
A number of studies has indicated a possible lower efficacy of immunotherapy when antibiotics are co-administered with nivolumab (17, 20, 21, 39-46). On the other hand, the present results are in accordance with Kaderbhai et al., who showed no significant effect of antibiotics on PFS or OS (7). However, that study included 53.3% of patients treated with antibiotics more than 1 month prior to starting treatment with nivolumab, in contrast to the present study and many other similar studies. This may have adversely affected the results as it is known that the gut microbiome normalizes after some time subsequently to antibiotic administration (17, 42). A number of factors may explain contradictory results regarding the interaction between antibiotic treatment and cancer immunotherapy. Firstly, the individual studies analyzed patients from different regions and different results may reflect regional differences in the gut microbiome (42). There were also differences with regard to antibiotic use. For example, in the present study, aminopenicillins and macrolides were the most commonly used antibiotics, while in the study reported by Hakozaki et al., trimethoprim/sulfamethoxazole was used more frequently, and quinolones were more extensively used in some other studies, specifically in the United States (17, 21, 39, 44). Individual antibiotics are characterized by a spectrum of antimicrobial activity, and only some antibiotics negatively affect the microbiome in terms of immunotherapy efficacy (21). The route of antibiotic administration may also play a role, with oral administration (predominant in the present cases) being potentially associated with improved efficacy of immunotherapy (47). Moreover, a bias cannot be excluded as patients using antibiotics usually have higher concentrations of CRP, which is a negative predictive and prognostic factor. Last but not least, a possible effect of other concomitant drugs that may affect the microbiome (e.g. corticosteroids, antacids, or NSAIDs) should not be forgotten (45). All these factors may not have been adequately taken into account in many reports.
Baseline patient characteristics.
Characteristics of concomitant medication.
Univariate analysis of progression-free (PFS) and overall (OS) survival.
Multivariate Cox proportional hazards model for progression-free (PFS) and overall (OS) survival.
The present study demonstrated the effect of corticotherapy on ORR treatment with nivolumab, while other studies showed also it to have an impact on PFS and OS (6, 15). Arbour et al. showed a decrease in both PFS and OS in patients treated with corticosteroids at a dose equivalent to ≥10 mg prednisone in comparison with other patients, including patients taking lower doses of corticosteroids (15). Fuca et al. also considered only patients with higher corticosteroid doses in their study showing worse outcome in patients treated with corticosteroid together with immunotherapy (8). In the present retrospective analysis, on the other hand, patients taking a low dose of corticosteroid were also considered, which may have affected the results. It is possible that lower doses of corticosteroids administered at the time of initiation of nivolumab therapy have no impact on the outcome. A meta-analysis on this topic also points to other factors that may have a role here (16). In particular, there may be a varying proportion of indications for corticotherapy in individual studies, such as intracranial metastases. The indication for corticotherapy was not analyzed in the present retrospective study, which may be one of its limitations. Finally, the negative results with regard to PFS and OS might easily be explained by the limited number of patients taking corticosteroids.
The results regarding the use of PPIs are consistent with the study by Routy et al. that indicated no effect of PPI on the outcome of immunotherapy (20). Similar results were reported by Hakozaki et al. in a cohort with a much higher use of PPIs (17), as well as by Zhao et al. (44). In the present study, no significant effect of probiotics was evident, but only few patients were taking probiotics. Probiotics are over-the-counter medications and their use might have not been reported by all patients, and, therefore, no definitive conclusions can be drawn.
As already mentioned, statins and metformin may potentiate the effect of anticancer therapy, including immunotherapy (26, 27). In univariate analysis, Omori et al. demonstrated significant improvement of the time to treatment failure in patients treated with nivolumab and statins (6), but only 10 patients in the cohort were treated with statins and multivariate analysis was not performed. The present study, which included a higher number of patients taking statins or metformin, did not indicate any effect of these drugs on the efficacy of nivolumab. Interactions between drugs administered simultaneously might also explain differences in the results.
NSAIDs have, among other mechanisms of action, an impact on inflammation by inhibiting COX2. COX2 is associated with PD-L1 expression in NSCLC (37). In a previous study, we demonstrated an association of parameters of inflammatory response with inferior outcome of nivolumab treatment in patients with NSCLC (5). A correlation between the concentrations of CRP, a biomarker of inflammatory response, and PFS and OS was evident in multivariate analysis of patients with NSCLC treated with nivolumab. Taking NSAIDs in the present study was associated with improved PFS in the multivariate analysis. Therefore, a positive immunomodulatory effect of NSAID on immunotherapy cannot be excluded. No such effect was evident in the univariate analyses; the effect might have been masked by other factors, including concomitant medications. Further larger prospective studies, as well as preclinical research on this topic, is therefore necessary.
The present study has several limitations. Firstly, this was a retrospective study of patients. Secondly, PFS was not confirmed by an independent review. Thirdly, the majority of patients had no PD-L1 status reported as this test was not necessary for the administration of nivolumab but PD-L1 status is associated with nivolumab efficacy in patients with non-squamous histology (2). Lastly, but not least, many analyses lacked sufficient statistical power as the number of patients taking some of concomitant medications was limited. Thus, the present report should be regarded as exploratory and the results should be confirmed in a larger prospective study.
In conclusion, the results of the present retrospective exploratory analysis underscore the importance of detailed knowledge of concomitant medications, including the route of administration and dose, in evaluating the effect on the outcome of nivolumab therapy. In the present study, a significantly higher progression rate was evident in patients treated with corticosteroids, and a positive effect of NSAID use at the time initiation of nivolumab treatment was observed.
Acknowledgements
This work was supported by AZV grant No. 17-30748A.
Footnotes
Authors' Contributions
MS and JB conceived the presented idea. MS, MZ, PZ, JK, OF, JS, LJ, MC, MH, MJ, JK, DK, OB and BM conceived and planned the experiments and collected the data. KH and MB analyzed the data. MS wrote the article with support from JB and BM. BM helped supervise the project.
Conflicts of Interest
In connection with this article, the Authors declare consulting services for BMS in the past.
- Received February 6, 2020.
- Revision received February 24, 2020.
- Accepted February 29, 2020.
- Copyright© 2020, International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved
