Advertisement

Efficacy of chemotherapy and atezolizumab in patients with non-small-cell lung cancer receiving antibiotics and proton pump inhibitors: pooled post hoc analyses of the OAK and POPLAR trials

Open AccessPublished:January 16, 2020DOI:https://doi.org/10.1016/j.annonc.2020.01.006

      Highlights

      • Use of antibiotics or proton pump inhibitors in patients with non-small cell lung cancer is associated with poor outcome.
      • The effect of antibiotics and proton pump inhibitors on outcome after immunotherapy warrants further investigation.
      • Treating physicians should carefully evaluate the need for co-medications such as antibiotics or proton pump inhibitors.

      Background

      Preclinical data have shown that proton pump inhibitors (PPI) can modulate the microbiome, and single-arm studies suggested that antibiotics (ATB) may decrease the efficacy of immune checkpoint inhibitors (ICI), but randomized controlled trial data are lacking. This pooled analysis evaluated the effect of ATB and PPI on outcome in patients randomized between ICI and chemotherapy.

      Patients and methods

      This retrospective analysis used pooled data from the phase II POPLAR (NCT01903993) and phase III OAK (NCT02008227) trials, which included 1512 patients with previously treated non-small-cell lung cancer (NSCLC) randomly assigned to receive atezolizumab (n = 757) or docetaxel (n = 755). The main objective of this analysis was to assess the impact of ATB and PPI use on overall survival (OS) and progression-free survival (PFS).

      Results

      A total of 169 (22.3%) patients in the atezolizumab group and 202 (26.8%) in the docetaxel group received ATB, and 234 (30.9%) and 260 (34.4%), respectively, received PPI. Multivariate analysis in all patients revealed that ATB were associated with shorter OS [hazard ratio (HR) 1.20, 95% confidence interval (CI) 1.04–1.39], as was PPI (HR 1.26, 95% CI 1.10–1.44). Within the atezolizumab population, OS was significantly shorter in patients who received ATB (8.5 versus 14.1 months, HR 1.32, 95% CI 1.06–1.63, P = 0.01) or PPI (9.6 versus 14.5 months, HR 1.45, 95% CI 1.20–1.75, P = 0.0001). PPI use was associated with shorter PFS in the atezolizumab population (1.9 versus 2.8 months, HR 1.30, 95% CI 1.10−1.53, P = 0.001). There was no association between ATB and PPI use and PFS or OS within the docetaxel population.

      Conclusion

      In this unplanned analysis from two randomized trials, data suggest that ATB or PPI use in patients with metastatic NSCLC is associated with poor outcome and may influence the efficacy of ICI.

      Key words

      Introduction

      Only a minority of patients across tumor types respond to immune checkpoint inhibitors (ICI) that target the programmed death ligand 1 (PD-L1) or programmed death 1 receptor.
      • Brahmer J.R.
      • Tykodi S.S.
      • Chow L.Q.
      • et al.
      Safety and activity of anti-PD-L1 antibody in patients with advanced cancer.
      • Borghaei H.
      • Paz-Ares L.
      • Horn L.
      • et al.
      Nivolumab versus docetaxel in advanced nonsquamous non-small-cell lung cancer.
      • Ribas A.
      • Hamid O.
      • Daud A.
      • et al.
      Association of pembrolizumab with tumor response and survival among patients with advanced melanoma.
      Variable response rates have been attributed to several factors, including lack of preexisting infiltrating lymphocytes,
      • Tumeh P.C.
      • Harview C.L.
      • Yearley J.H.
      • et al.
      PD-1 blockade induces responses by inhibiting adaptive immune resistance.
      low tumor mutational burden,
      • Rizvi N.A.
      • Hellmann M.D.
      • Snyder A.
      • et al.
      Cancer immunology. Mutational landscape determines sensitivity to PD-1 blockade in non-small cell lung cancer.
      defective antigen presentation,
      • Zaretsky J.M.
      • Garcia-Diaz A.
      • Shin D.S.
      • et al.
      Mutations associated with acquired resistance to PD-1 blockade in melanoma.
      and functionally exhausted tumor infiltrating lymphocytes.
      • Huang A.C.
      • Postow M.A.
      • Orlowski R.J.
      • et al.
      T-cell invigoration to tumour burden ratio associated with anti-PD-1 response.
      Recent work in preclinical models has highlighted the importance of the gut microbiota in modifying tumor responses to chemotherapeutic agents
      • Iida N.
      • Dzutsev A.
      • Stewart C.A.
      • et al.
      Commensal bacteria control cancer response to therapy by modulating the tumor microenvironment.
      ,
      • Viaud S.
      • Saccheri F.
      • Mignot G.
      • et al.
      The intestinal microbiota modulates the anticancer immune effects of cyclophosphamide.
      and immunotherapies.
      • Gopalakrishnan V.
      • Spencer C.N.
      • Nezi L.
      • et al.
      Gut microbiome modulates response to anti-PD-1 immunotherapy in melanoma patients.
      • Matson V.
      • Fessler J.
      • Bao R.
      • et al.
      The commensal microbiome is associated with anti-PD-1 efficacy in metastatic melanoma patients.
      • Routy B.
      • Le Chatelier E.
      • Derosa L.
      • et al.
      Gut microbiome influences efficacy of PD-1-based immunotherapy against epithelial tumors.
      Specifically, co-medications such as antibiotics (ATB) can affect the integrity of the intestinal microbiota, which plays a key role in regulating the host innate and acquired immune response.
      • Gopalakrishnan V.
      • Spencer C.N.
      • Nezi L.
      • et al.
      Gut microbiome modulates response to anti-PD-1 immunotherapy in melanoma patients.
      • Matson V.
      • Fessler J.
      • Bao R.
      • et al.
      The commensal microbiome is associated with anti-PD-1 efficacy in metastatic melanoma patients.
      • Routy B.
      • Le Chatelier E.
      • Derosa L.
      • et al.
      Gut microbiome influences efficacy of PD-1-based immunotherapy against epithelial tumors.
      Several retrospective analyses of small clinical studies involving patients with metastatic lung cancer, kidney cancer, melanoma, and bladder cancer, suggested a reduced clinical benefit of ICI in patients who received ATB around the initiation of ICI.
      • Gopalakrishnan V.
      • Spencer C.N.
      • Nezi L.
      • et al.
      Gut microbiome modulates response to anti-PD-1 immunotherapy in melanoma patients.
      • Matson V.
      • Fessler J.
      • Bao R.
      • et al.
      The commensal microbiome is associated with anti-PD-1 efficacy in metastatic melanoma patients.
      • Routy B.
      • Le Chatelier E.
      • Derosa L.
      • et al.
      Gut microbiome influences efficacy of PD-1-based immunotherapy against epithelial tumors.
      • Derosa L.
      • Hellmann M.D.
      • Spaziano M.
      • et al.
      Negative association of antibiotics on clinical activity of immune checkpoint inhibitors in patients with advanced renal cell and non-small-cell lung cancer.
      • Sen S.
      • Carmagnani Pestana R.
      • Hess K.
      • et al.
      Impact of antibiotic use on survival in patients with advanced cancers treated on immune checkpoint inhibitor phase I clinical trials.
      Proton pump inhibitors (PPI) have also been associated with gut dysbiosis, decreased bacterial richness, and promotion of T-cell tolerance.
      • Imhann F.
      • Bonder M.J.
      • Vich Vila A.
      • et al.
      Proton pump inhibitors affect the gut microbiome.
      ,
      • Jackson M.A.
      • Goodrich J.K.
      • Maxan M.E.
      • et al.
      Proton pump inhibitors alter the composition of the gut microbiota.
      However, less information is available on the effect of PPI on ICI efficacy.
      • Routy B.
      • Le Chatelier E.
      • Derosa L.
      • et al.
      Gut microbiome influences efficacy of PD-1-based immunotherapy against epithelial tumors.
      All previous retrospective studies analyzing the effects of ATB and PPI on outcome after ICI have been based on either single-arm studies or cohorts of patients treated within a single institution, without control groups. Considering the possibility of bias and that ATB and PPI may even have a prognostic impact regardless of the given treatment, we investigated the effect of ATB and PPI in patients with non-small-cell lung cancer (NSCLC) in randomized, controlled trials that included a control group of patients not receiving ICI. In NSCLC, ICI have demonstrated a significant overall survival (OS) benefit irrespective of PD-L1 expression.
      • Borghaei H.
      • Paz-Ares L.
      • Horn L.
      • et al.
      Nivolumab versus docetaxel in advanced nonsquamous non-small-cell lung cancer.
      ,
      • Carbone D.P.
      • Reck M.
      • Paz-Ares L.
      • et al.
      First-line nivolumab in stage IV or recurrent non-small-cell lung cancer.
      • Brahmer J.
      • Reckamp K.L.
      • Baas P.
      • et al.
      Nivolumab versus docetaxel in advanced squamous-cell non-small-cell lung cancer.
      • Fehrenbacher L.
      • Spira A.
      • Ballinger M.
      • et al.
      Atezolizumab versus docetaxel for patients with previously treated non-small-cell lung cancer (POPLAR): a multicentre, open-label, phase 2 randomised controlled trial.
      • Reck M.
      • Rodriguez-Abreu D.
      • Robinson A.G.
      • et al.
      Pembrolizumab versus chemotherapy for PD-L1-positive non-small-cell lung cancer.
      • Rittmeyer A.
      • Barlesi F.
      • Waterkamp D.
      • et al.
      Atezolizumab versus docetaxel in patients with previously treated non-small-cell lung cancer (OAK): a phase 3, open-label, multicentre randomised controlled trial.
      The primary analysis of the randomized, phase III OAK trial demonstrated a median OS of 13.8 months in the atezolizumab arm versus 9.6 months in the docetaxel arm [hazard ratio (HR) 0.73, 95% confidence interval (CI) 0.62–0.87, P = 0.0003].
      • Rittmeyer A.
      • Barlesi F.
      • Waterkamp D.
      • et al.
      Atezolizumab versus docetaxel in patients with previously treated non-small-cell lung cancer (OAK): a phase 3, open-label, multicentre randomised controlled trial.
      Similarly, the randomized phase II POPLAR trial also reported a significant improvement in median OS for atezolizumab versus docetaxel (12.6 months versus 9.7 months, respectively, HR 0.73, 95% CI 0.53–0.99, P = 0.04).
      • Fehrenbacher L.
      • Spira A.
      • Ballinger M.
      • et al.
      Atezolizumab versus docetaxel for patients with previously treated non-small-cell lung cancer (POPLAR): a multicentre, open-label, phase 2 randomised controlled trial.

      Patients and methods

       Study design and participants

      This retrospective analysis used pooled data from the phase II POPLAR (NCT01903993)
      • Fehrenbacher L.
      • Spira A.
      • Ballinger M.
      • et al.
      Atezolizumab versus docetaxel for patients with previously treated non-small-cell lung cancer (POPLAR): a multicentre, open-label, phase 2 randomised controlled trial.
      and phase III OAK (NCT02008227)
      • Rittmeyer A.
      • Barlesi F.
      • Waterkamp D.
      • et al.
      Atezolizumab versus docetaxel in patients with previously treated non-small-cell lung cancer (OAK): a phase 3, open-label, multicentre randomised controlled trial.
      randomized trials. Similarities in the design of the OAK and POPLAR trials justified pooling of the data. Both trials included patients with measurable, previously treated NSCLC, who were randomly assigned to receive either atezolizumab or docetaxel (no crossover) and used the same stratification factors and schedule of assessments. In both trials, patients were randomized 1 : 1 to intravenous atezolizumab or docetaxel; the primary end point was OS, and progression-free survival (PFS) was one of the secondary end points. Detailed descriptions of the eligibility criteria and recruitment methods for both trials have been published previously.
      • Fehrenbacher L.
      • Spira A.
      • Ballinger M.
      • et al.
      Atezolizumab versus docetaxel for patients with previously treated non-small-cell lung cancer (POPLAR): a multicentre, open-label, phase 2 randomised controlled trial.
      ,
      • Rittmeyer A.
      • Barlesi F.
      • Waterkamp D.
      • et al.
      Atezolizumab versus docetaxel in patients with previously treated non-small-cell lung cancer (OAK): a phase 3, open-label, multicentre randomised controlled trial.
      ,
      • Fehrenbacher L.
      • von Pawel J.
      • Park K.
      • et al.
      Updated efficacy analysis including secondary population results for OAK: a randomized phase III study of atezolizumab versus docetaxel in patients with previously treated advanced non-small cell lung cancer.
      The studies were done in accordance with the Declaration of Helsinki and approval was obtained from local ethics committees.
      The OAK trial included 850 patients in the primary intention-to-treat population and 1225 patients in the secondary intention-to-treat population (the additional patients were enrolled to power an OS comparison in the subgroup of patients with high PD-L1 expression).
      • Fehrenbacher L.
      • von Pawel J.
      • Park K.
      • et al.
      Updated efficacy analysis including secondary population results for OAK: a randomized phase III study of atezolizumab versus docetaxel in patients with previously treated advanced non-small cell lung cancer.
      The use of ATB and PPI was extracted from information about concomitant medications recorded in the case report forms. All ATB classes were considered, including quinolones, penicillins, and cephalosporins (supplementary Table S1, available at Annals of Oncology online). PPI included omeprazole, pantoprazole, lansoprazole, rabeprazole, esomeprazole, and dexlansoprazole. For both ATB and PPI, any use within a window of 30 days before and 30 days after the start of first study treatment was chosen based on an assessment of the criteria used in previous publications.
      • Routy B.
      • Le Chatelier E.
      • Derosa L.
      • et al.
      Gut microbiome influences efficacy of PD-1-based immunotherapy against epithelial tumors.
      ,
      • Derosa L.
      • Hellmann M.D.
      • Spaziano M.
      • et al.
      Negative association of antibiotics on clinical activity of immune checkpoint inhibitors in patients with advanced renal cell and non-small-cell lung cancer.
      Data were extracted at the time of primary data cutoff (OAK trial: 7 July 2016; POPLAR trial: 8 May 2015).

       Statistical considerations

      In this retrospective analysis, baseline patient characteristics of the pooled population were recorded according to use of ATB or PPI within the treatment window. Variables for the statistical model were selected based on their prognostic values in NSCLC, and included age, sex, race, Eastern Cooperative Oncology Group (ECOG) performance status (PS 0−1), tobacco use, histology (non-squamous, squamous), immune cell score, and number of prior therapies.
      Possible factors associated with OS and PFS were first identified by a screening process via univariate Cox models. Variables with a P value ≤0.15 were then tested in separate multivariate Cox models. The Akaike information criterion was used to determine the best multivariate model. An initial model without interactions was used to identify the prognostic impact of ATB and PPI use. Separate models for ATB and PPI were then used to test for interactions. Additionally, HRs for subgroup analyses and within treatment effects were evaluated with an unstratified Cox model. Survival curves were estimated using the Kaplan–Meier method.

      Results

       Study population

      The pooled analysis reported here included the primary population from POPLAR (n = 287), and the total population from OAK (n = 1225 patients), giving a total sample size of 1512 patients (Figure 1). At the termination of the combined trials, 966 (63.9%) participants had died and 124 (8.2%) had withdrawn or were lost to follow-up. The median duration of follow-up was 19.2 months.
      Figure thumbnail gr1
      Figure 1CONSORT diagram showing trial profile for POPLAR and OAK trials combined.
      In the pooled analysis, median OS was significantly longer in the atezolizumab group than in the docetaxel group [13.2 months (95% CI 11.3−14.5) versus 9.7 months (95% CI 9.0−11.1), respectively], corresponding to a HR of 0.79 (95% CI 0.69−0.91) in the stratified analysis.
      A total of 169 (22.3%) patients in the atezolizumab group and 202 (26.8%) in the docetaxel group received ATB within the treatment window; 234 (30.9%) and 260 (34.4%), respectively, received PPI; and 74 (9.8%) and 82 (10.9%), respectively, received ATB plus PPI. The median durations of use were 8 days for ATB and 13 days for PPI. As shown in Table 1, demographic and baseline characteristics were balanced both between and within treatment arms, except for higher proportions of patients with an ECOG PS of 1 in the atezolizumab groups who had received ATB versus those who had not received ATB (74.6% versus 61.6%, respectively) or had received PPI versus those who had not received PPI (73.5% versus 60.4%, respectively). Baseline characteristics of patients who received ATB plus PPI within the treatment window are shown in supplementary Table S2, available at Annals of Oncology online.
      Table 1Demographic and baseline characteristics of the pooled population according to antibiotic or PPI use within the treatment window
      VariableAtezolizumab (N = 757)Docetaxel (N = 755)Atezolizumab (N = 757)Docetaxel (N = 755)
      ATB+ n (%)ATB− n (%)ATB+ n (%)ATB− n (%)PPI+ n (%)PPI− n (%)PPI+ n (%)PPI− n (%)
      Number of patients169 (22)588 (78)202 (27)553 (73)234 (31)523 (69)260 (34)495 (66)
      Age<65 y94 (55.6)328 (55.8)102 (50.5)311 (56.2)130 (55.6)292 (55.8)146 (56.2)267 (53.9)
      SexMale103 (60.9)368 (62.6)126 (62.4)329 (59.5)146 (62.4)325 (62.1)178 (68.5)277 (56)
      Tobacco useNever

      Previous/current
      39 (23.1)

      130 (76.9)
      100 (17)

      488 (83)
      34 (16.8)

      168 (83.2)
      91 (16.5)

      462 (83.5)
      37 (15.8)

      197 (84.2)
      102 (19.5)

      421 (80.5)
      28 (10.8)

      232 (89.2)
      97 (19.6)

      398 (80.4)
      Number of prior therapies1

      2
      125 (74)

      44 (26)
      432 (73.5)

      156 (26.5)
      157 (77.7)

      45 (22.3)
      404 (73.1)

      149 (26.9)
      177 (75.6)

      57 (24.4)
      380 (72.7)

      143 (27.3)
      186 (71.5)

      74 (28.5)
      375 (75.8)

      120 (24.2)
      HistologyNon-squamous

      Squamous
      124 (73.4)

      45 (26.6)
      423 (71.9)

      165 (28.1)
      133 (65.8)

      69 (34.2)
      414 (74.9)

      139 (25.1)
      169 (72.2)

      65 (27.8)
      378 (72.3)

      145 (27.7)
      179 (68.8)

      81 (31.2)
      368 (74.3)

      127 (25.7)
      PD-L1TC0/IC0

      TC1/2/3 or IC1/2/3

      Unknown
      63 (37.3)

      103 (60.9)

      2 (1.2)
      248 (42.2)

      337 (57.3)

      1 (0.2)
      87 (43.1)

      115 (56.9)

      0
      225 (40.7)

      324 (58.6)

      3 (0.5)
      90 (38.5)

      142 (60.7)

      1 (0.4)
      221 (42.3)

      298 (57.0)

      2 (0.4)
      108 (41.5)

      151 (58.1)

      1 (0.4)
      204 (41.2)

      288 (58.2)

      2 (0.4)
      ECOG PS0

      1
      43 (25.4)

      126 (74.6)
      224 (38.1)

      362 (61.6)
      68 (33.7)

      134 (66.3)
      211 (38.2)

      341 (61.7)
      61 (26.1)

      172 (73.5)
      206 (39.4)

      316 (60.4)
      97 (37.3)

      163 (62.7)
      182 (36.8)

      312 (63.0)
      Use of ATB or PPI within the treatment window is indicated by + (received) and − (did not receive).
      ATB, antibiotics; ECOG, Eastern Cooperative Oncology Group; PD-L1, programmed death ligand 1; PPI, proton pump inhibitors; PS, performance status.

       Effect of ATB and PPI on survival

      After selecting the variables that were significantly associated with OS in the univariate analyses (Table 2 and supplementary Table S3, available at Annals of Oncology online), the multivariate models for OS showed that ATB and PPI were associated with unfavorable prognosis (Table 3). The HRs were 1.20 (95% CI 1.04−1.39, P = 0.01) for ATB use and 1.26 (95% CI 1.10−1.44, P < 0.01) for PPI use.
      Table 2Overall survival by ATB and PPI use within window in the pooled randomized population
      Docetaxel (N = 755)Atezolizumab (N = 757)
      ATB+n

      Events, n (%)

      Median (95% CI) (months)

      Hazard ratio (95% CI)

      P value (log-rank)
      202

      150 (74.3)

      9.10 (8.28–10.91)











      0.90 (0.70–1.15)

      0.3960
      169

      109 (64.5)

      8.54 (6.80–12.39)





      ATB−n

      Events, n (%)

      Median (95% CI) (months)

      Hazard ratio (95% CI)

      P value (log-rank)
      553

      354 (64.0)

      10.28 (9.13–11.70)











      0.75 (0.65–0.87)

      0.0001
      588

      353 (60.0)

      14.06 (12.62–15.61)





      PPI+n

      Events, n (%)

      Median (95% CI) (months)

      Hazard ratio (95% CI)

      P value (log-rank)
      260

      189 (72.7)

      9.13 (8.34–10.25)











      0.92 (0.75–1.14)

      0.4384
      234

      162 (69.2)

      9.63 (7.79–12.62)





      PPI−n

      Events, n (%)

      Median (95% CI) (months)

      Hazard ratio (95% CI)

      P value (log-rank)
      495

      315 (63.6)

      10.97 (9.23–12.39)











      0.73 (0.62–0.85)

      0.0001
      523

      300 (57.4)

      14.52 (12.94–16.00)





      ATB+ and PPI+
      Subjects received both ATB and PPI.
      n

      Events, n (%)

      Median (95% CI) (months)

      Hazard ratio (95% CI)

      P value (log-rank)
      82

      60 (73.2)

      8.97 (6.24–10.55)











      1.26 (0.87–1.82)

      0.2289
      74

      54 (73.0)

      6.64 (3.55–8.54)





      ATB− and PPI−
      Subjects received neither ATB nor PPI.
      n

      Events, n (%)

      Median (95% CI) (months)

      Hazard ratio (95% CI)

      P value (log-rank)
      673

      444 (66.0)

      10.22 (9.10–11.53)











      0.74 (0.65–0.85)

      <0.0001
      683

      408 (59.7)

      14.06 (12.62–15.47)





      Summaries of time-to-event (median) are Kaplan–Meier estimates. 95% CI for median was computed using the method of Brookmeyer and Crowley. Hazard ratios were estimated by unstratified Cox regression. P values were from the unstratified log-rank test (two-sided). Use of ATB, PPI or antibiotic class within the treatment window is indicated by + (received) and − (did not receive). Hazard ratio <1 favors atezolizumab.
      ATB, antibiotics; CI, confidence interval; ECOG, Eastern Cooperative Oncology Group; PPI, proton pump inhibitors.
      a Subjects received both ATB and PPI.
      b Subjects received neither ATB nor PPI.
      Table 3Prognostic effect of variables in the overall pooled population
      A. Multivariate model without interaction
      Some variables were not included in the multivariate model after selection from the univariate models or Akaike information criterion (AIC).
      PFSOS
      Multivariate Cox HR (95% CI)Multivariate Cox HR (95% CI)
      Treatment arm (docetaxel or atezolizumab)0.96 (0.86–1.08); P = 0.520.79 (0.70–0.90); P < 0.01
      ATB (no or yes)1.10 (0.97–1.25); P = 0.141.20 (1.04–1.39); P = 0.01
      PPI (no or yes)1.14 (1.02–1.28); P = 0.031.26 (1.10–1.44); P < 0.01
      Histology (non-squamous or squamous)1.15 (1.02–1.29); P = 0.031.38 (1.20–1.59); P < 0.01
      Number of prior therapies (1 versus 2)0.88 (0.77–0.99); P = 0.04
      IC score
      IC score = tumor-infiltrating immune cells (as percentage of tumor area: IC3 ≥ 10%, IC2 ≥ 5% and <10%, IC1 ≥ 1% and <5%, and IC0 < 1%).
       10.87 (0.75–1.00)
       20.92 (0.75–1.14)
       30.76 (0.60–0.96); P = 0.05
      Sex (female or male)1.13 (0.99–1.29); P = 0.08
      Metastatic sites (<2 versus ≥2 at enrolment)1.29 (1.1–1.52); P < 0.011.84 (1.48–2.28); P < 0.01
      ECOG PS (0 or 1)1.29 (1.15–1.45); P < 0.011.71 (1.49–1.97); P < 0.01
      B. Multivariate model with treatment interactionsInteraction P valueInteraction P value
      ATB/treatment interactionP = 0.57P = 0.31
      PPI/treatment interactionP = 0.11P = 0.402
      ATB, antibiotics; CI, confidence interval; HR, hazard ratio; PPI, proton pump inhibitors.
      a Some variables were not included in the multivariate model after selection from the univariate models or Akaike information criterion (AIC).
      b IC score = tumor-infiltrating immune cells (as percentage of tumor area: IC3 ≥ 10%, IC2 ≥ 5% and <10%, IC1 ≥ 1% and <5%, and IC0 < 1%).
      OS was significantly shorter for patients in the atezolizumab group who had received ATB compared with those who had not (8.5 versus 14.1 months, HR 1.32, 95% CI 1.06−1.63, P = 0.01) in the unstratified Cox model (Figure 2A). Similarly, Figure 2B shows that OS was significantly shorter for patients who used PPI in the atezolizumab group compared with those who had not (9.6 versus 14.5 months, HR 1.45, 95% CI 1.20−1.75, P = 0.0001). Although no significant differences in OS were found for ATB or PPI use within the docetaxel group, all HRs were >1 (Figure 2). Tests for interaction in separate multivariate models were not statistically significant (Table 3B). Exploratory analysis suggests that the effect observed on OS was seen across all classes of ATB used (supplementary Table S4, available at Annals of Oncology online).
      Figure thumbnail gr2
      Figure 2Kaplan-Meier curves showing overall survival with atezolizumab and docetaxel according to use of ATB (A) or PPI (B).
      Hazard ratios are from an unstratified Cox model. Use of ATB or PPI within the treatment window is indicated by + (received) and − (did not receive). HR > 1 indicates worse OS for ATB or PPI use within the specific treatment.
      ATB, antibiotics; CI, confidence interval; HR, hazard ratio; OS, overall survival; PPI, proton pump inhibitors.
      While the prognostic effect of PPI use on PFS was shown in the multivariate model (HR 1.14, 95% CI 1.02−1.28, P = 0.03), the interaction between PPI use and treatment was not significant (P = 0.11; Table 3B).
      PPI use was associated with a significantly shorter PFS in the atezolizumab group (1.9 versus 2.8 months, HR 1.30, 95% CI 1.10−1.53, P = 0.001); supplementary Figure S1, available at Annals of Oncology online. No differences in PFS were seen in patients who had received ATB (supplementary Figure S1, available at Annals of Oncology online).
      Additional data on the effect of the combination of ATB plus PPI on outcomes can be found in Table 2 and supplementary Table S5 and Figure S2, available at Annals of Oncology online.

      Discussion

      Using data from a pooled analysis involving 1512 patients with previously treated NSCLC, we showed that ATB or PPI use may be associated with an adverse prognostic effect regardless of treatment. These data also suggest that the efficacy of ICI may be reduced by ATB or PPI. Compared with previous reports, our analysis is the first to include more than 1500 patients from randomized, controlled trials and the first published report suggesting a detrimental effect of PPI on ICI efficacy in patients with advanced NSCLC.
      In patients treated with atezolizumab, PPI use was associated with a greater risk of progression or death, while ATB use was associated with a greater risk of death. We also observed that ATB or PPI use was associated with reduced OS in the docetaxel group, but the magnitude of the difference was small and not statistically significant.
      Multivariate analyses confirmed the prognostic effect of PPI or ATB use, but showed no significant interaction by treatment arm, although an effect of these co-medications on the outcomes with ICI or chemotherapy cannot be ruled out.
      Our findings are consistent with previously published studies reporting a potentially adverse impact of ATB in patients treated with ICI; however, these were small single-arm studies.
      • Routy B.
      • Le Chatelier E.
      • Derosa L.
      • et al.
      Gut microbiome influences efficacy of PD-1-based immunotherapy against epithelial tumors.
      ,
      • Derosa L.
      • Hellmann M.D.
      • Spaziano M.
      • et al.
      Negative association of antibiotics on clinical activity of immune checkpoint inhibitors in patients with advanced renal cell and non-small-cell lung cancer.
      Derosa et al.
      • Derosa L.
      • Hellmann M.D.
      • Spaziano M.
      • et al.
      Negative association of antibiotics on clinical activity of immune checkpoint inhibitors in patients with advanced renal cell and non-small-cell lung cancer.
      reported that ATB use within 30 days of beginning ICI (n = 48 of 239 patients) was associated with decreased OS (HR 4.4, 95% CI 2.6−7.7), whereas a retrospective analysis showed that ATB use within a window 1 month before to 1 month after ICI (n = 38 of 96 patients) had no influence on median OS (HR 0.84, 95% CI 0.48−1.47) among patients with non-squamous, advanced NSCLC.
      • Huemer F.
      • Rinnerthaler G.
      • Lang D.
      • et al.
      Association between antibiotics use and outcome in patients with NSCLC treated with immunotherapeutics.
      Such conflicting findings are difficult to explain and may result from the inherent limitations of retrospective analyses, small sample sizes, different definitions for the co-medication use window, and imbalances in patient characteristics between groups.
      Studies analyzing the microbiome of patients receiving ICI have found that patients responding to the treatment have a more diverse gut microbiome, which in turn has been shown to correlate with T cell numbers in the peripheral blood and tumor microenvironment.
      • Gopalakrishnan V.
      • Spencer C.N.
      • Nezi L.
      • et al.
      Gut microbiome modulates response to anti-PD-1 immunotherapy in melanoma patients.
      • Matson V.
      • Fessler J.
      • Bao R.
      • et al.
      The commensal microbiome is associated with anti-PD-1 efficacy in metastatic melanoma patients.
      • Routy B.
      • Le Chatelier E.
      • Derosa L.
      • et al.
      Gut microbiome influences efficacy of PD-1-based immunotherapy against epithelial tumors.
      Together, these data suggest that the decrease in diversity of bacterial species following ATB may provide an explanation for the adverse effects on outcome in patients undergoing ICI treatment.
      Similarly, PPI have been shown to affect the gut microbiome, possibly by a direct effect of stomach acid, which normally provides the main defense system against bacterial influx from food and oral bacterial flora. In some studies, the changes induced by PPI were more prominent than those associated with ATB.
      • Imhann F.
      • Bonder M.J.
      • Vich Vila A.
      • et al.
      Proton pump inhibitors affect the gut microbiome.
      ,
      • Jackson M.A.
      • Goodrich J.K.
      • Maxan M.E.
      • et al.
      Proton pump inhibitors alter the composition of the gut microbiota.
      Other studies have suggested an association between PPI use and community pneumococcal pneumonia, supporting the hypothesis that PPI may affect the functionality of the immune system.
      • Laheij R.J.
      • Sturkenboom M.C.
      • Hassing R.J.
      • et al.
      Risk of community-acquired pneumonia and use of gastric acid-suppressive drugs.
      Future research should focus on elucidating the possible mechanisms for interactions of ICI with co-medications, and the role of the microbiome.
      This study has several limitations. Our findings are based on an exploratory retrospective analysis of subgroups that were not pre-specified. Furthermore, some differences in patient baseline characteristics, including a higher rate of ECOG PS 1 in patients treated with ATB, were also observed. In addition, although we observed some significant interactions between the use of co-medication and the benefit of either chemotherapy or ICI, the effect size was rather small and, therefore, validation in other independent datasets is required before our results can affect clinical decision making.
      Lastly, we did not characterize the biological mechanisms by which these medications impacted the efficacy of PD-L1 inhibition and assumed, based on historical data, that these medications induced dysbiosis consequently affecting the response to ICI.
      Whether ATB and PPI are purely prognostic or contributing to resistance to checkpoint inhibition remains a matter of debate. However, these data should encourage physicians to carefully evaluate the need for co-medications such as PPI and ATB in their patients. Validation of these results from other randomized, controlled trials is needed, considering that prospective testing of ATB and PPI on outcome after ICI may not be feasible.

      Acknowledgements

      The authors would like to thank Simon Fear and Michaela Sedova for their contributions in the analysis of these data as well as the patients participating in the POPLAR and OAK trials and their families, the investigators, and staff at participating centers. Medical writing assistance for this manuscript was provided by Tim Kelly, BSc (Medi-Kelsey Ltd, Ashbourne, UK), funded by F. Hoffmann-La Roche Ltd , Basel, Switzerland.

      Funding

      F Hoffmann-La Roche / Genentech funded the study (no grant number is applicable), provided study drugs, were involved in the study design, data collection, data analysis, data interpretation, and writing of the report, and gave approval to submit for publication. The corresponding authors had full access to all the data in the study and had final responsibility for the decision to submit for publication.

      Disclosure

      MC reports grants to the institute from Bristol Myers Squibb/II-ON and Roche, outside the submitted work. AC and ADS are employees of F. Hoffmann-La Roche Ltd. DRN is an employee of Genentech Inc. and reports ownership of stocks from F. Hoffmann-La Roche Ltd. DRG reports a consulting or advisory role for AstraZeneca, Celgene, CellMax Life, Genentech, Guardant Health, Lilly, Liquid Genomics, Inc., and research funding from AstraZeneca/MedImmune (Inst) and Genentech (Inst). AR reports a consultant or advisory role for AbbVie, AstraZeneca/MedImmune, Boehringer Ingelheim, Bristol-Myers Squibb, Lilly, MSD, Pfizer, and Roche/Genentech. MLA is an employee of Genentech Inc. and is now an employee of Insitro. TP reports honoraria received from Novartis, BMS, Merck, Pfizer, Roche, and AstraZeneca. MK reports funding and a speaker's fee to the institute from BMS, Roche and an unpaid advisory role for BMS, outside the submitted work. FGH reports research funding from BMS, Accuray Inc., Bioprotect, and Prostate Cancer Foundation.

      Data sharing statement

      Qualified researchers may request access to individual patient level data through the clinical study data request platform (www.clinicalstudydatarequest.com). Further details on Roche's criteria for eligible studies are available here (https://clinicalstudydatarequest.com/Study-Sponsors/Study-Sponsors-Roche.aspx). For further details on Roche's Global Policy on the Sharing of Clinical Information and how to request access to related clinical study documents, see here (https://www.roche.com/research_and_development/who_we_are_how_we_work/clinical_trials/our_commitment_to_data_sharing.htm).

      Supplementary data

      Figure thumbnail figs1
      Figure thumbnail figs2

      References

        • Brahmer J.R.
        • Tykodi S.S.
        • Chow L.Q.
        • et al.
        Safety and activity of anti-PD-L1 antibody in patients with advanced cancer.
        N Engl J Med. 2012; 366: 2455-2465
        • Borghaei H.
        • Paz-Ares L.
        • Horn L.
        • et al.
        Nivolumab versus docetaxel in advanced nonsquamous non-small-cell lung cancer.
        N Engl J Med. 2015; 373: 1627-1639
        • Ribas A.
        • Hamid O.
        • Daud A.
        • et al.
        Association of pembrolizumab with tumor response and survival among patients with advanced melanoma.
        JAMA. 2016; 315: 1600-1609
        • Tumeh P.C.
        • Harview C.L.
        • Yearley J.H.
        • et al.
        PD-1 blockade induces responses by inhibiting adaptive immune resistance.
        Nature. 2014; 515: 568-571
        • Rizvi N.A.
        • Hellmann M.D.
        • Snyder A.
        • et al.
        Cancer immunology. Mutational landscape determines sensitivity to PD-1 blockade in non-small cell lung cancer.
        Science. 2015; 348: 124-128
        • Zaretsky J.M.
        • Garcia-Diaz A.
        • Shin D.S.
        • et al.
        Mutations associated with acquired resistance to PD-1 blockade in melanoma.
        N Engl J Med. 2016; 375: 819-829
        • Huang A.C.
        • Postow M.A.
        • Orlowski R.J.
        • et al.
        T-cell invigoration to tumour burden ratio associated with anti-PD-1 response.
        Nature. 2017; 545: 60-65
        • Iida N.
        • Dzutsev A.
        • Stewart C.A.
        • et al.
        Commensal bacteria control cancer response to therapy by modulating the tumor microenvironment.
        Science. 2013; 342: 967-970
        • Viaud S.
        • Saccheri F.
        • Mignot G.
        • et al.
        The intestinal microbiota modulates the anticancer immune effects of cyclophosphamide.
        Science. 2013; 342: 971-976
        • Gopalakrishnan V.
        • Spencer C.N.
        • Nezi L.
        • et al.
        Gut microbiome modulates response to anti-PD-1 immunotherapy in melanoma patients.
        Science. 2018; 359: 97-103
        • Matson V.
        • Fessler J.
        • Bao R.
        • et al.
        The commensal microbiome is associated with anti-PD-1 efficacy in metastatic melanoma patients.
        Science. 2018; 359: 104-108
        • Routy B.
        • Le Chatelier E.
        • Derosa L.
        • et al.
        Gut microbiome influences efficacy of PD-1-based immunotherapy against epithelial tumors.
        Science. 2018; 359: 91-97
        • Derosa L.
        • Hellmann M.D.
        • Spaziano M.
        • et al.
        Negative association of antibiotics on clinical activity of immune checkpoint inhibitors in patients with advanced renal cell and non-small-cell lung cancer.
        Ann Oncol. 2018; 29: 1437-1444
        • Sen S.
        • Carmagnani Pestana R.
        • Hess K.
        • et al.
        Impact of antibiotic use on survival in patients with advanced cancers treated on immune checkpoint inhibitor phase I clinical trials.
        Ann Oncol. 2018; 29: 2396-2398
        • Imhann F.
        • Bonder M.J.
        • Vich Vila A.
        • et al.
        Proton pump inhibitors affect the gut microbiome.
        Gut. 2016; 65: 740-748
        • Jackson M.A.
        • Goodrich J.K.
        • Maxan M.E.
        • et al.
        Proton pump inhibitors alter the composition of the gut microbiota.
        Gut. 2016; 65: 749-756
        • Carbone D.P.
        • Reck M.
        • Paz-Ares L.
        • et al.
        First-line nivolumab in stage IV or recurrent non-small-cell lung cancer.
        N Engl J Med. 2017; 376: 2415-2426
        • Brahmer J.
        • Reckamp K.L.
        • Baas P.
        • et al.
        Nivolumab versus docetaxel in advanced squamous-cell non-small-cell lung cancer.
        N Engl J Med. 2015; 373: 123-135
        • Fehrenbacher L.
        • Spira A.
        • Ballinger M.
        • et al.
        Atezolizumab versus docetaxel for patients with previously treated non-small-cell lung cancer (POPLAR): a multicentre, open-label, phase 2 randomised controlled trial.
        Lancet. 2016; 387: 1837-1846
        • Reck M.
        • Rodriguez-Abreu D.
        • Robinson A.G.
        • et al.
        Pembrolizumab versus chemotherapy for PD-L1-positive non-small-cell lung cancer.
        N Engl J Med. 2016; 375: 1823-1833
        • Rittmeyer A.
        • Barlesi F.
        • Waterkamp D.
        • et al.
        Atezolizumab versus docetaxel in patients with previously treated non-small-cell lung cancer (OAK): a phase 3, open-label, multicentre randomised controlled trial.
        Lancet. 2017; 389: 255-265
        • Fehrenbacher L.
        • von Pawel J.
        • Park K.
        • et al.
        Updated efficacy analysis including secondary population results for OAK: a randomized phase III study of atezolizumab versus docetaxel in patients with previously treated advanced non-small cell lung cancer.
        J Thorac Oncol. 2018; 13: 1156-1170
        • Huemer F.
        • Rinnerthaler G.
        • Lang D.
        • et al.
        Association between antibiotics use and outcome in patients with NSCLC treated with immunotherapeutics.
        Ann Oncol. 2019; 30: 652-653
        • Laheij R.J.
        • Sturkenboom M.C.
        • Hassing R.J.
        • et al.
        Risk of community-acquired pneumonia and use of gastric acid-suppressive drugs.
        JAMA. 2004; 292: 1955-1960