Article Text

Original research
Integrated analysis of concomitant medications and oncological outcomes from PD-1/PD-L1 checkpoint inhibitors in clinical practice
  1. Alessio Cortellini1,2,
  2. Marco Tucci3,4,
  3. Vincenzo Adamo5,
  4. Luigia Stefania Stucci3,
  5. Alessandro Russo5,
  6. Enrica Teresa Tanda6,
  7. Francesco Spagnolo6,
  8. Francesca Rastelli7,
  9. Renato Bisonni7,
  10. Daniele Santini8,
  11. Marco Russano8,
  12. Cecilia Anesi8,
  13. Raffaele Giusti9,
  14. Marco Filetti9,
  15. Paolo Marchetti9,10,11,
  16. Andrea Botticelli10,
  17. Alain Gelibter11,
  18. Mario Alberto Occhipinti11,
  19. Riccardo Marconcini12,
  20. Maria Giuseppa Vitale13,
  21. Linda Nicolardi14,
  22. Rita Chiari14,
  23. Claudia Bareggi15,
  24. Olga Nigro16,
  25. Alessandro Tuzi16,
  26. Michele De Tursi17,
  27. Nicola Petragnani18,
  28. Laura Pala19,
  29. Sergio Bracarda20,
  30. Serena Macrini20,
  31. Alessandro Inno21,
  32. Federica Zoratto22,
  33. Enzo Veltri22,
  34. Barbara Di Cocco22,
  35. Domenico Mallardo23,
  36. Maria Grazia Vitale23,
  37. David James Pinato24,
  38. Giampiero Porzio2,
  39. Corrado Ficorella1,2 and
  40. Paolo Antonio Ascierto23
  1. 1Department of Biotechnology and Applied Clinical Sciences, University of L'Aquila, L'Aquila, Italy
  2. 2Medical Oncology, St. Salvatore Hospital, L'Aquila, Italy
  3. 3Medical Oncology Unit, Department of Biomedical Sciences and Human Oncology, University of Bari, Bary, Italy
  4. 4National Cancer Research Center, Tumori Institute IRCCS Giovanni Paolo II, Bari, Italy
  5. 5Medical Oncology, Department of Human Pathology, A.O. Papardo, University of Messina, Messina, Italy
  6. 6IRCCS Ospedale Policlinico San Martino, Genova, Italy
  7. 7Medical Oncology, ASUR District Area 4 Fermo, Fermo, Italy
  8. 8Medical Oncology, Campus Bio-Medico University, Rome, Italy
  9. 9Medical Oncology Unit, Sant'Andrea Hospital of Rome, Rome, Italy
  10. 10Department of Clinical and Molecular Medicine, Sapienza University of Rome, Rome, Italy
  11. 11Medical Oncology Unit B, Policlinico Umberto I, Sapienza University of Rome, Roma, Italy
  12. 12Medical Oncology, Azienda Ospedaliero-Universitaria Pisana, Pisa, Italy
  13. 13Medical Oncology, University Hospital Modena, Modena, Italy
  14. 14UOC Oncologia Padova Sud, Azienda ULSS 6 Euganea, Padova, Italy
  15. 15Medical Oncology Unit, Fondazione IRCCS Ca' Granda Ospedale Maggiore di Milano Policlinico, Milano, Italy
  16. 16Medical Oncology, ASST Sette Laghi, Ospedale di Circolo e Fondazione Macchi, Varese, Italy
  17. 17Department of Medical, Oral and Biotechnological Sciences, Gabriele d'Annunzio University of Chieti and Pescara, Chieti, Italy
  18. 18Department of Psychological, Health and Territorial Sciences, University G. D'Annunzio of Chieti and Pescara, Chieti, Italy
  19. 19Division of Medical Oncology for Melanoma, Sarcoma and Rare Tumors, IEO European Institute of Oncology IRCCS, Milan, Italy
  20. 20Medical Oncology, Azienda Ospedaliera S. Maria, Terni, Italy
  21. 21Oncology Unit, IRCCS Ospedale Sacro Cuore Don Calabria, Negrar, Italy
  22. 22Medical Oncology, Santa Maria Goretti Hospital, Latina, Italy
  23. 23Melanoma, Cancer Immunotherapy and Development Therapeutics Unit, Istituto Nazionale Tumori IRCCS Fondazione "G. Pascale", Napoli, Italy
  24. 24Division of Cancer, Department of Surgery and Cancer, Imperial College London, Hammersmith Hospital, London, UK
  1. Correspondence to Dr Alessio Cortellini; alessiocortellini{at}gmail.com

Abstract

Background Concomitant medications, such as steroids, proton pump inhibitors (PPI) and antibiotics, might affect clinical outcomes with immune checkpoint inhibitors.

Methods We conducted a multicenter observational retrospective study aimed at evaluating the impact of concomitant medications on clinical outcomes, by weighing their associations with baseline clinical characteristics (including performance status, burden of disease and body mass index) and the underlying causes for their prescription. This analysis included consecutive stage IV patients with cancer, who underwent treatment with single agent antiprogrammed death-1/programmed death ligand-1 (PD-1/PD-L1) with standard doses and schedules at the medical oncology departments of 20 Italian institutions. Each medication taken at the immunotherapy initiation was screened and collected into key categories as follows: corticosteroids, antibiotics, gastric acid suppressants (including proton pump inhibitors - PPIs), statins and other lipid-lowering agents, aspirin, anticoagulants, non-steroidal anti-inflammatory drugs (NSAIDs), ACE inhibitors/Angiotensin II receptor blockers, calcium antagonists, β-blockers, metformin and other oral antidiabetics, opioids.

Results From June 2014 to March 2020, 1012 patients were included in the analysis. Primary tumors were: non-small cell lung cancer (52.2%), melanoma (26%), renal cell carcinoma (18.3%) and others (3.6%). Baseline statins (HR 1.60 (95% CI 1.14 to 2.25), p=0.0064), aspirin (HR 1.47 (95% CI 1.04 to 2.08, p=0.0267) and β-blockers (HR 1.76 (95% CI 1.16 to 2.69), p=0.0080) were confirmed to be independently related to an increased objective response rate. Patients receiving cancer-related steroids (HR 1.72 (95% CI 1.43 to 2.07), p<0.0001), prophylactic systemic antibiotics (HR 1.85 (95% CI 1.23 to 2.78), p=0.0030), prophylactic gastric acid suppressants (HR 1.29 (95% CI 1.09 to 1.53), p=0.0021), PPIs (HR 1.26 (95% CI 1.07 to 1.48), p=0.0050), anticoagulants (HR 1.43 (95% CI: 1.16 to 1.77), p=0.0007) and opioids (HR 1.71 (95% CI 1.28 to 2.28), p=0.0002) were confirmed to have a significantly higher risk of disease progression. Patients receiving cancer-related steroids (HR 2.16 (95% CI 1.76 to 2.65), p<0.0001), prophylactic systemic antibiotics (HR 1.93 (95% CI 1.25 to 2.98), p=0.0030), prophylactic gastric acid suppressants (HR 1.29 (95% CI 1.06 to 1.57), p=0.0091), PPI (HR 1.26 (95% CI 1.04 to 1.52), p=0.0172), anticoagulants (HR 1.45 (95% CI 1.14 to 1.84), p=0.0024) and opioids (HR 1.53 (95% CI 1.11 to 2.11), p=0.0098) were confirmed to have a significantly higher risk of death.

Conclusion We confirmed the association between baseline steroids administered for cancer-related indication, systemic antibiotics, PPIs and worse clinical outcomes with PD-1/PD-L1 checkpoint inhibitors, which can be assumed to have immune-modulating detrimental effects.

  • immunotherapy
http://creativecommons.org/licenses/by-nc/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See http://creativecommons.org/licenses/by-nc/4.0/.

View Full Text

Statistics from Altmetric.com

Introduction

Drug–drug interactions (DDIs) have traditionally played an important role in the safe and effective delivery of systemic anticancer therapy.1 Concomitant medications can alter efficacy and worsen toxicity from systemic therapies through pharmacodynamic (PK) and pharmacokinetic (PD) interactions, particularly due to interference with absorption, distribution, metabolism and elimination of drugs.1 The advent of immune checkpoint inhibitors (ICIs) has reignited the interest toward DDIs beyond traditional PK/PD considerations.2 ICIs exert their action mainly relying on the restoration/activation of T-cell responses against cancer, and therefore, might be altered by those factors which particularly affect the immune balance prior to the ICIs administration, such as disruption of the homeostatic balance within the gut microbiome3 and drug-induced immune suppression.4

Concomitant medications including steroids, proton pump inhibitors and systemic antibiotics have been postulated to exert immune-modulatory effects within the tumor microenviroment, thus affecting clinical outcomes from ICI therapy.2

However, while some degree of biological plausibility exists to justify an immune-mediated basis to the detrimental effect observed on response and survival from ICIs, the strength and reliability of the association has been largely derived from retrospective/post hoc analyzes and the dispute between causative instead of associative relationship has not been fully resolved.2 Given their immunosuppressive action, steroids were the first class of medications which was significantly related to worse clinical outcomes with cancer immunotherapy.5 Nevertheless, a significant association with worse outcome was later confirmed for baseline steroids administered for palliation of cancer-rleated symptoms but not for other indications including treatment of immune-related adverse events.6 7

In the case of systemic antibiotics, the evidence for a causative effect seems stronger and more plausible in view of their capacity to perturbate the gut microbiome, a renown determinant of response to ICIs.8–10 Nevertheless, the risk of collinearity with the underlying cause for the antibiotics prescription (eg, infections which might subtend to poorer clinical condition), has yet to be fully discriminated.

Proton pump inhibitors were associated to decreased progression-free survival (PFS) and overall survival (OS) in non-small-cell-lung-cancer (NSCLC) and melanoma patients receiving programmed death-1 (PD-1)/programmed death ligand-1 (PD-L1) checkpoint inhibitors,9 11 while some studies investigated the impact of other concomitant medication, such as non-steroidal anti-inflammatory drugs (NSAIDs), metformin, aspirin, β-blockers and statins, without conclusive results.12 13

While a growing body of evidence underscores the importance of concomitant medications in affecting outcome from ICI, a key limitation affecting most of the published evidence is the lack of an integrated analysis of multiple classes of concomitant therapies. This is of particular importance to determine whether the influence on clinical outcomes might be driven by associative rather than causative links, especially given the high prevalence of polypharmacy in patients with cancer.14

Recently, we created a large multicenter, observational study of patients receiving PD-1/PD-L1 checkpoint inhibitors in clinical practice, already subject of several analyzes,15–20 and we now gathered the baseline concomitant medication information for the same population, in order to evaluate their impact on clinical outcomes.

Materials and methods

Study design

We conducted a real-world, multicenter, retrospective observational data collection aimed at evaluating the impact of concomitant medications at immunotherapy initiation on clinical outcomes, by weighing their associations with baseline clinical characteristics (including performance status, burden of disease and body mass index (BMI)) and the underlying indication for steroids, antibiotics and gastric acid suppressants prescription. This study included consecutive patients with confirmed diagnosis of stage IV solid cancer, who underwent treatment with single agent anti-PD-1/PD-L1 as first or subsequent line, with data availability regarding baseline concomitant medication. The data collection was further implemented and updated involving patients treated at the medical oncology departments of 20 Italian institutions (online supplemental table 1), between June 2014 and March 2020. Patients were treated according to the tumor type indication with pembrolizumab, nivolumab, atezolizumab and other PD-1/PD-L1 prescribed at doses and schedules indicated in the respective product SPCs.

Supplemental material

Clinical outcomes of interest included objective response rate (ORR), PFS and OS. Patients were assessed with radiological imaging in clinical practice, with a frequency ranging from 12 to 16 weeks, according to the monitoring requirements for high-cost drugs of the respective national drug regulatory agencies (the on-line monitoring dashboard of the ‘Agenzia Italiana del Farmaco’ requires a disease assessment at least every 16 weeks; available at: https://servizionline.aifa.gov.it/). RECIST (V. 1.1) criteria were used21 and a subsequent confirming imaging was recommended. However, treatment beyond disease progression was allowed when clinically indicated. ORR was defined as the portion of patients experiencing an objective response (complete or partial response) as best response to immunotherapy. PFS was defined as the time from treatment initiation to disease progression or death, whichever occurred first. OS was defined as the time from treatment initiation to death. For PFS as well as for OS, patients without events were considered as censored at the time of the last follow-up. Data cut-off period was May 2020.

Fixed multivariable regression models were used to estimate clinical outcomes according to each concomitant medication category following adjustment for preplanned adjusting covariates that might represent confounders.22–24 The key covariates were: primary tumor type (NSCLC, melanoma, renal cell carcinoma and others), age (<70 vs ≥70 years),25–28 sex (male vs female), Eastern Cooperative Oncology Group-Performance Status (ECOG-PS) (0–1 vs ≥2), burden of disease (number of metastatic sites≤2 vs >2), treatment line (first vs non-first) and BMI. BMI was used given to its alleged role in affecting immunotherapy clinical outcomes15 16 and as a surrogate of cardiovascular/metabolic conditions which might have influenced the prescription of certain concomitant medications. Weight and height were obtained from patients’ medical records at the time of immunotherapy initiation. BMI was calculated using the formula of weight/height2 (kilograms per square meter) and categorized according to WHO categories: underweight, BMI <18.5 kg/m2; normal-weight, 18.5 kg/m2≤ BMI ≤24.9 kg/m2; overweight, 25 kg/m2≤ BMI ≤29.9 kg/m2; obese, BMI ≥30 kg/m2. In order to properly weighing the role of baseline concomitant medication, their association with ECOG-PS, burden of disease and with BMI were evaluated.

Concomitant medications

Information on prescribing of concomitant medications was gathered from patients’ clinical records. Each medication prescribed at the time of immunotherapy initiation was screened and categorized as follows:

  • Corticosteroids administration (dose ≥10 mg prednisone equivalent per day, with a minimum 24 hours of dosing) within the 30 days before immunotherapy initiation, classified according to their indication as: no (including those patients receiving <10 mg prednisone equivalent) versus cancer indications (administration for symptoms palliation, radiation therapy, central nervous system metastases) versus non-cancer indications (eg, other inflammation processes non related to cancer).

  • Systemic antibiotics within the 30 days before immunotherapy initiation, classified according to their indication as: no versus prophylaxis (eg, to prevent COPD exacerbation or diverticulitis prevention) versus infection (in case of a diagnosed infective disease).

  • Baseline gastric acid suppressant, classified according to their indication as: no vs gastritis/gastroesophageal reflux disease (GERD) versus prophylaxis (eg, to prevent gastritis due to other concomitant medication); no versus H2 Antagonists (such as ranitidine) vs proton pump inhibitors.

  • Baseline statins (yes vs no).

  • Other baseline lipid-lowering agents (fibrates, ezetimibe and similar) (yes vs no).

  • Baseline aspirin (considered as low-dose daily assumption of aspirin for cardiovascular prevention) (yes vs no).

  • Baseline anticoagulants (including new oral anticoagulants, low molecular weight heparin and cumarinic anticoagulant drugs) (yes vs no).

  • NSAIDs within the 30 days before treatment initiation, including COX-2 inhibitors (including both chronic and PRN administration) (yes vs no).

  • Baseline ACE inhibitors/angiotensin II receptor blockers (ARBs) (yes vs no), calcium antagonists (yes vs no), β-blockers (yes vs no).

  • Baseline metformin (yes vs no) and other oral antidiabetics (yes vs no).

  • Baseline opioids (yes vs no).

Statistical analysis

Baseline patient characteristics were reported with descriptive statistics. χ2 test was used for the univariate analysis of ORR. Logistic regression was used for the multivariate analysis of ORR and to compute the ORs with 95% CIs. Median PFS and median OS were evaluated using the Kaplan-Meier method. Median period of follow-up was calculated according to the reverse Kaplan-Meier method. Cox proportional hazards regression was used for the univariate analysis, for the fixed multivariate analysis of PFS and OS and to compute the HRs for disease progression and death with 95% CIs. The alpha level for all analyzes was set to p<0.05. χ2 test was also used to evaluate the associations between baseline concomitant medication and ECOG-PS (0–1 vs ≥2), burden of disease (number of metastatic sites≤2 vs>2) and BMI (underweight, normal-weight, overweight and obese). In order to properly evaluate the role of some baseline medications, a further analysis using the BMI as a continuous covariate was performed, through the one-way analysis of variance (ANOVA). All statistical analyzes were performed using MedCalc Statistical Software V.19.3.1 (MedCalc Software, Ostend, Belgium; https://www.medcalc.org; 2020).

Results

Patients’ characteristics

In total, 1012 consecutive advanced cancer patients were evaluated. Patients characteristics are and baseline medication are summarized in table 1. The median age was 68.5 years (range: 21–92), male/female ratio was 647/365. Primary tumors were: NSCLC (52.2%), melanoma (26%), renal cell carcinoma (18.3%) and others (3.6%).

Table 1

Patients characteristics

Efficacy analysis

The median follow-up was 24.2 months (95% CI 23.3 to 67.2); in the study population ORR was 37.6% (95% CI 33.8%% to 41.7) (361 responses out of 960 evaluable patients), while median PFS and median OS were 10.2 months (95% CI 9.2 to 11.4; 681 progression events) and 19.7 months (95% CI 17.5 to 24.6; 520 censored patients), respectively. Table 2 reports the univariate and multivariate analyzes of ORR. Compared with patients who did not received baseline steroids, patients receiving them for cancer-related symptoms were confirmed to have a significantly lower ORR compared with patients who did not receive baseline steroids (HR 0.55 (95% CI 0.38 to 0.81), p=0.0020), while not patients who received steroids for non-cancer indications. Also baseline statins (HR 1.60 (95% CI 1.14 to 2.25), p=0.0064), aspirin (HR 1.47 (95% CI 1.04 to 2.08), p=0.0267) and β-blockers (HR 1.76 (95% CI 1.16 to 2.69), p=0.0080) were confirmed to be independently related to an increased ORR. Table 3 summarizes the univariate and multivariate analyzes of PFS. Patients receiving cancer-related steroids (HR 1.72 (95% CI 1.43 to 2.07), p<0.0001), prophylactic systemic antibiotics (HR 1.85 (95% CI 1.23 to 2.78), p=0.0030), prophylactic gastric acid suppressants (HR 1.29 (95% CI 1.09 to 1.53), p=0.0021), proton pump inhibitors (HR 1.26 (95% CI 1.07 to 1.48), p=0.0050), anticoagulants (HR 1.43 (95% CI 1.15 to 1.76), p=0.0009) and opioids (HR 1.54 (95% CI 1.11 to 2.12), p=0.0083), were confirmed to have a significantly higher risk of disease progression. On the contrary, patients who assumed aspirin were confirmed to have a significantly lower risk of disease progression (HR 0.79 (95% CI 0.64 to 0.98), p=0.0318). Table 4 summarizes the univariate and multivariate analyzes of OS. Patients receiving cancer-related steroids (HR 2.16 (95% CI 1.76 to 2.65), p<0.0001), prophylactic systemic antibiotics (HR 1.93 (95% CI 1.25 to 2.98), p=0.0030), prophylactic gastric acid suppressants (HR 1.29 (95% CI 1.06 to 1.57), p=0.0091), proton pump inhibitors (HR 1.26 (95% CI 1.04 to 1.52), p=0.0172), anticoagulants (HR 1.45 (95% CI 1.14 to 1.84), p=0.0024) and opioids (HR 1.53 (95% CI 1.11 to 2.11), p=0.0098) were confirmed to have a significantly higher risk of death. Figures 1 and 2 report the Kaplan-Meier survival curves for PFS and OS according to baseline steroids, systemic antibiotics, gastric acid suppressants, anticoagulants and opioids.

Figure 1

Kaplan-Meier survival estimates. Progression-free survival; (A) Steroids. No: 13.5 months (95% CI 10.8 to 15.4; 472 events); non-cancer indications: 10.0 months (95% CI 7.2 to 18.3; 36 events); cancer indications: 4.9 months (95% CI 3.6 to 6.5; 247 events); (B) Systemic antibiotics. No: 10.5 months (95% CI 9.2 to 11.9, 622 events); prophylaxis: 2.8 months (95% CI 2.1 to 6.7, 25 events); infections: 10.9 months (95% CI 6.4 to 37.5, 34 events); (C) Gastric acid suppressants. No: 13.5 months (95% CI 10.5 to 18.2, 288 events); gastritis/GERD: 11.2 months (95% CI 7.9 to 17.3, 60 events); prophylaxis: 8.2 months (95% CI 6.9 to 9.9, 333 events). Overall survival; (D) Steroids. No: 30.8 months (95% CI 24.4 to 36.3; 432 censored); non-cancer indications: 44.6 months (95% CI 12.0 to 44.6; 30 censored); cancer indications: 7.8 months (95% CI 5.4 to 9.8; 58 censored); (E) Systemic antibiotics. No: 22.8 months (95% CI 18.9 to 27.4, 494 censored); prophylaxis: 4.9 months (95% CI 3.5 to 11.0, 8 censored); infections: 15.2 months (95% CI 9.8 to 18.1, 18 censored); (F) Gastric acid suppressants. No: 29.4 months (95% CI 22.8 to 39.8, 266 censored); gastritis/GERD: 23.2 months (95% CI 13.4 to 30.8, 59 censored); prophylaxis: 14.8 months (95% CI 12.3 to 52.3, 195 censored). GERD, gastroesophageal reflux disease.

Figure 2

Kaplan-Meier survival estimates. Progression-free survival; (A) Gastric acid suppressants. No: 13.5 months (95% CI 10.5 to 18.2, 288 events); H2 antagonists: 10.3 months (95% CI 3.8 to 13.9; 40 events); proton pump inhibitors: 8.4 months (95% CI 7.5 to 10.0; 353 events); (B) Anticoagulants. No: 10.9 months (95% CI 9.9 to 13.0, 573 events); yes: 6.3 months (95% CI 3.9 to 9.2, 108 events); (C) Opioids. No: 11.0 months (95% CI 10.0 to 13.5, 564 events); yes: 3.8 months (95% CI 2.9 to 6.4, 56 events). Overall survival (D) Gastric acid suppressants. No: 29.4 months (95% CI 22.8 to 39.8, 266 censored); H2 antagonists: 21.1 months (95% CI 6.1 to 25.0; 28 censored); proton pump inhibitors: 15.4 months (95% CI 12.5 to 18.1; 226 censored); (E) Anticoagulants. No: 23.9 months (95% CI 18.9 to 28.6, 460 censored); yes: 12.4 months (95% CI 7.8 to 15.1; 60 censored); (F) Opioids No: 23.2 months (95% CI 18.9 to 28.8, 452 censored); yes: 8.6 months (95% CI 4.7 to 12.7; 22 censored).

Table 2

Univariate and multivariate analyzes of ORR

Table 3

Univariate and multivariate analyzes of PFS

Table 4

Univariate and multivariate analyzes of OS

Baseline associations

All the baseline associations are summarized in online supplemental table 5; the administration of baseline steroids (p<0.0001), systemic antibiotics (p=0.0001), gastric acid suppressant (both according to their indication (p<0.0001) and drug class (p=0.0002)), anticoagulants (p=0.0011), antidepressants (p=0.0002) and opioids (p=0.0123) was significantly associated to a poorer ECOG-PS. Similarly, the administration of baseline steroids (p=0.0014), gastric acid suppressant (both according to their indication (p<0.0001) and drug class (p<0.0001)), β-blockers (p=0.0166), and opioids (p=0.0014) was significantly associated to a higher burden of disease.

The administration of statins (p=0.005), anticoagulants (p=0.001), ACE inhibitors/ARBs (p=0.002), calcium antagonists (p=0.008), β-blockers (p=0.008), and other oral antidiabetics (p=0.036) was significantly associated to a higher BMI, while the administration of NSAIDs (p=0.003), and opioids (p=0.004) to a lower BMI at the ANOVA analysis. Using WHO categories for BMI, we confirmed the association with anticoagulants (p=0.0438), NSAIDs (0.0069) and opioids (p=0.0153).

Discussion

Identification of factors that prelude to immune-refractoriness is an area of high unmet need in cancer immunotherapy. A number of non-oncological medical therapies have been postulated to render the tumor microenviroment more tolerogenic, therefore exerting detrimental effects on depth, duration of response and survival of patients treated with ICI.2 Our purpose was to provide a more comprehensive analysis with a large population of patients with different malignancies receiving PD-1/PD-L1 inhibitors, in order to gain reliable results about the putative immune-modulating effects of concomitant medication most usually taken by patients with cancer.

We produce important confirmatory evidence regarding the association between exposure to steroids, systemic antibiotics and proton pump inhibitors and worse outcomes from ICI. In addition, we provide novel evidence for a shorter survival in patients on anticoagulants and opioids at ICIs initiation, a finding that was not previously reported in large populations. Similarly, a significant association between improved ORR/PFS and baseline aspirin, and between improved ORR and statins and β-blockers, had never been reported in the context of cancer patients receiving PD-1/PD-L1 inhibitors.

Intriguingly, among the baseline medication which resulted to be significantly related to clinical outcomes in our study population, the common thread might be somehow considered the immune modulating effects, particularly exerted through the modifying pressure on the gut-microbiome.

Steroids were the only baseline medication concordantly related to ORR, PFS and OS in our study population. Glucocorticoids can affect the gut microbiome, the intestinal mucosa and synthesis/secretion of mucins.29–31 Nevertheless, we have to consider the possible associative (and not causative) effect played by the significant relation between steroids assumption and poorer PS/higher burden of disease. In fact, patients receiving baseline steroids for symptoms palliation were confirmed to have significantly worse ORR, PFS and OS, compared with patients who did not received steroids, while not patients who received steroids for non-cancer indications, similarly to what reported by Ricciuti et al.6

It is also well known that antibiotics might affect immunity by inducing gut microbiome alterations.32 In our study, only systemic antibiotics administered for prophylaxis were confirmed to be significantly related to shortened PFS and OS at the multivariate analysis, while not antibiotics administered to treat active infections. Interestingly, it was further revealed that antibiotics administered prior of the immunotherapy initiation was confirmed to be related to worse outcomes, while not those administered concurrently,10 supporting the hypothesis that the underlying modulating effects on the gut microbiome can affect the immunotherapy clinical outcomes only when the modifying pressure is exerted on the prior immune-balance, and not during the treatment. From this perspective, antibiotics administered for prophylactic indications might exert the same negative effect of those administered to treat active infections. However, we have to consider that patients receiving antibiotics have poorer clinical conditions overall and looking at the table 5 we can noticed that those on prophylactic antibiotics had the highest percentage of ECOG-PS ≥2 patients.

Table 5

Summary of the associations between each drug category and ECOG-PS, burden of disease and BMI

Previous studies investigated the role of proton pump inhibitors exclusively,9 11 while this is the first analysis which evaluated the role of gastric acid suppressants overall. Proton pump inhibitors could negatively affect the gut microbiome due to both the changes of the gastric pH and to bacterial species selections,33 34 but also H2 antagonists are known to have modifying gut microbiome functions and to induce intestinal barrier dysfunctions.35 36 Curiously, proton pump inhibitors administration was confirmed to be associated to shortened PFS and OS, but not H2 antagonists and patients receiving gastric acid suppressants for prophylactic purpose experienced significantly shorter PFS and OS, while patients who received these agents to treat gastritis/GERD achieved similar outcomes to patients who did not receive them. In this case, the highest percentage of patients with ECOG-PS ≥2 is among the patients with gastritis/GERD and among the patients on H2 antagonists, but to proper weigh our results, we must take into account the significant association between baseline gastric acid suppressants and burden of disease (online supplemental table 3). Therefore, we are not able to recommend H2 antagonists prescription instead of proton pump inhibitors for patients with cancer who are in need of a gastric acid suppressant treatment and are going to receive a PD-1/PD-L1 checkpoint inhibitor, even more considering the recent alerts from drug regulatory agencies regarding the possible contamination with N-nitrosodimethylamine of some of these agents.37 38

Anticoagulants have been assumed to modulate the immune balance, affecting the antibacterial innate immune response,39 while chronic opioid dosing has been already associated to shift of the gut microbiome and intestinal barrier dysfunction.40–43 Nevertheless, it should be considered that patients requiring anticoagulation therapy and opioids are often frailer than patients who do not: a point that should be emphasized when evaluating PFS and OS where poorer PS and higher disease burden may confound the analyzes.

The relationship between aspirin and cancer prevention/progression have been historically known,44 45 but in the setting of immunotherapy of cancer, few studies have been published. Wang et al12 evaluated a cohort of 330 melanoma patients receiving PD-1 inhibitors, without reporting any association between ORR, PFS, OS and NSAIDs use (including aspirin). Even if (cyclooxygenase) COX-2 expression was known to be positively associated with PD-L1 tumor expression,46 we did not find associations between baseline NSAIDs (excluding aspirin) and immunotherapy clinical outcomes, but the significant association between improved ORR and baseline aspirin, allows to speculate about the possible synergistic effects of COX inhibition in antitumor immunity.47 To our knowledge, the association between statins administration and improved clinical outcomes of patients with cancer receiving ICIs have never been described, however, it is well known that cholesterol metabolism plays a role in CD8+T cell function and might be modulated in order to enhance antitumor immunity.48–51 β-blockers have already been known to improve recurrence-free survival in patients with radically resected melanoma and to have synergistic effects with immunotherapy in mice models.52 53 In our cohort baseline β-blockers are significantly associated to improved ORR, while in the study of Wang et al no significant associations were found.12 Intriguingly, the inhibition of β-adrenoceptors in the intestinal mucosa and gut lymphatic tissue has been linked with changes in type and virulence of the intestinal microbiome and to reduced bacterial translocation trough the intestinal barrier.54 Finally, to properly weighing the ORR analysis results, we have to consider the significant association between β-blockers and low burden of disease and between β-blockers, aspirin, lipid-lowering agents and higher baseline BMI. However, contrary to what we previously reported,15 16 BMI was not significantly associated to improved outcomes in this population, even though a trend toward better ORR, PFS and OS for increased BMI levels was found. Considering that the most robust evidence of an association between improved outcomes and obesity came from NSCLC,55 this finding might be related to the internal distribution of the study population, which after the update and the addition of data from some new institutions passed form 65.1% and 18.7% of NSCLC and melanoma patients to 52.2% and 26%, respectively.

Despite the suggestion that metformin administration might exert a synergistic antitumor role with ICIs,2 56 we did not find any significant association between ORR, PFS, OS and baseline metformin, in keeping with previously published evidence.12

Beyond the dispute between association and causation, we have to consider that there are some other potential mechanisms by which concomitant medications could affect clinical outcomes during immunotherapy, in addition to gut microbiome alteration. It is well known that corticosteroids can exert immune-suppressive effects through several mechanisms, such as activation of glucocorticoid response elements with the inhibition of interleukin 1 (IL-1) and IL-6 transcription,57 58 induction of T-cell suppression and diminishing naïve T cell proliferation.59 Gastric acid suppressants can cause immune-suppressive effects through the inhibition of adhesion molecules of inflammatory cells and affecting cytokines secretion.60 Aspirin can exert several effects on both innate and adaptive immune responses. It can modulate proliferation/maturation of immune cells, regulate the cytokine production, and induce the lipoxin-driven immune counter-regulation. Nevertheless, aspirin can also have the immune suppressive ability of inducing tolerogenic dendritic cells, therefore expanding Treg cells.61

Our study acknowledges a number of limitations, including the retrospective design and the lack of central radiology review. The heterogeneity of tumor types evaluated might had affected the analysis even if we included the primary tumor in the preplanned fixed multivariate model. We have to also consider the small sample size of some subgroups as patients receiving steroids for non-cancer indication, gastric acid suppressants to treat gastritis/GERD and receiving H2 antagonists. Moreover, we are planning to investigate the possible detrimental effect on immunotherapy clinical outcomes of specific polypharmacy patterns. To confirm our results, interactions between concomitant baseline medications and immunotherapy clinical outcomes should be assessed prospectively.

Conclusion

This is the largest study to provide a broad, integrated analysis of multiple concomitant medications as determinants of response and survival to immunotherapy in patients with solid tumors. While unable to discriminate between a mechanistic and an associative effect, our study strengthens the knowledge around the association between baseline steroids administered for cancer-related indications, systemic antibiotics, proton pump inhibitors and worse clinical outcomes with PD-1/PD-L1 checkpoint inhibitors, which can be assumed to have immune-modulating detrimental effects. To correctly weight the association between anticoagulants/opioids and worse PFS/OS we must consider their statistical association with poorer PS/higher burden of disease, while the significant association between the administration of aspirin, β-blockers, statins and improved ORR deserves further investigations.

References

View Abstract

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

Footnotes

  • Twitter @Al3ssandroRusso

  • Contributors All authors contributed to the publication according to the ICMJE guidelines for the authorship. All authors read and approved the manuscript and agree to be accountable for all aspects of the research in ensuring that the accuracy or integrity of any part of the work are appropriately investigated and resolved.

  • Funding This work was supported by the Consorzio Interuniversitario Nazionale per la Bio-Oncologia (CINBO). DP is supported by grant funding from the Wellcome Trust Strategic Fund (PS3416) and from the NIHR Imperial Biomedical Research Center (BRC) ITMAT Push for Impact Scheme 2019 and acknowledges infrastructural support by the Cancer Research UK Imperial Center and the Imperial Experimental Cancer Medicine Center (ECMC).

  • Competing interests AC received speaker fees and grant consultancies from Roche, MSD, BMS, AstraZeneca, Novartis, Astellas. RG received speaker fees and grant consultancies from AstraZeneca and Roche. MGV received speaker fees, grant consultancies and travel support from BMS, Ipsen, Novartis, Pfizer, Astellas, Jansen and Pierre-Fabre. AR received grant consultancies from AstraZeneca and MSD. RM received grant consultancies from Pierre-Fabre, MSD, Incyte, BMS, and Roche. FS received speaker fees and grant consultancies from Roche, Novartis, BMS, MSD, Pierre-Fabre, Sanofi, Merck and Sunpharma. DP received lecture fees from ViiV Healthcare, Bayer Healthcare and travel expenses from BMS and Bayer Healthcare; consulting fees for Mina Therapeutics, EISAI, Roche, Astra Zeneca; received research funding (to institution) from MSD, BMS. PAA received speaker fees and grant consultancies from BMS, Roche-Genentech, MSD, Dohme, Array, Novartis, Merck-Serono, Pierre-Fabre, Incyte, New Link Genetics, Genmab, Medimmune, AstraZeneca, Syndax, SunPharma, Sanofi, Idera, Ultimovacs, Sandoz, Immunocore, 4SC, Alkermes, Italfarmaco, Nektar, Boehringer-Ingelheim; he also received research funds from BMS, Roche-Genentech, Array.

  • Patient consent for publication Not required.

  • Ethics approval All patients alive at the time of data collection provided an informed consent for the present retrospective analysis. The procedures followed were in accordance with the precepts of Good Clinical Practice and the declaration of Helsinki. The study was approved by the respective local ethical committees on human experimentation of each institution, after previous approval by the coordinating center (University of L’Aquila, Internal Review Board protocol number 32865, approved on July 24 2018).

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data availability statement The datasets used during the present study are available from the corresponding author on reasonable request.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.