Abstract
Background:
Antibiotic use alters commensal gut microbiota, which is a key regulator of immune homeostasis.
Objective:
To investigate the impact of antibiotic use on clinical outcomes in metastatic renal cell carcinoma (mRCC) patients treated with systemic agents.
Design, setting, and participants:
We analyzed two cohorts: an institutional cohort (n = 146) receiving programmed cell death protein 1 (PD-1)/programmed death-ligand 1 (PD-L1)-based immune checkpoint inhibitors (ICIs) and a trial-database cohort (n = 4144) receiving interferon-α (n = 510), mammalian target of rapamycin (mTOR) inhibitors (n = 660), and vascular endothelial growth factor targeted therapies (VEGF-TT; n = 2974) on phase II/III clinical trials.
Outcomes measurements and statistical analysis:
The association of antibiotic use (defined as use from 8 wk before to 4 wk after the initiation of anticancer therapy) with progression-free survival (PFS) and overall survival (OS) was evaluated using Cox regression, adjusted for known prognostic factors including International Metastatic RCC Database Consortium risk factors.
Results and limitations:
Most patients were male, had clear cell histology, and were at an intermediate risk. Overall, in the institutional cohort, objective response rate (ORR) was 30%, PFS was 7.2 mo, and 1-yr OS was 77%. Antibiotic users (n = 31, 21%) had a lower ORR (12.9% vs 34.8%, p = 0.026) and shorter PFS (adjusted hazard ratio [HR] = 1.96, 95% confidence interval [CI] 1.20–3.20, p = 0.007) than antibiotic nonusers. In the trial-database cohort, antibiotic use (n = 709, 17%) adversely impacted OS in patients treated with interferon (HR = 1.62, 95% CI 1.13–2.31, p = 0.008) or with VEGF-TT and prior cytokines (HR = 1.65, 95% CI 1.04–2.62, p = 0.033), but not patients treated with mTOR inhibitors or VEGF-TT without prior cytokines. Limitations include retrospective design, and limited details regarding concomitant medications and antibiotic indication/duration.
Conclusions:
Antibiotic use appears to reduce the efficacy of immunotherapy-based regimens in mRCC. The modulation of gut microbiota may play an important role in optimizing outcomes of patients treated with ICIs.
Patient summary:
We evaluated metastatic renal cell carcinoma patients and found that those who were treated with immunotherapy had worse outcomes if they also received antibiotics at the start of treatment. This study highlights the importance of judicious antibiotic use.
Keywords: Antibiotics, Checkpoint inhibitors, Immunotherapy, Programmed cell death protein1/, programmed death-ligand 1, Renal cell carcinoma
Introduction
The treatment landscape for metastatic renal cell carcinoma (mRCC) has evolved over the last 20 yr from an era of cytokine-based regimens to that of vascular endothelial growth factor targeted therapies (VEGF-TT) and, more recently, immune checkpoint inhibitors (ICIs) [1,2]. Interleukin-2 and interferon-α (IFN) had been the mainstay of treatment in select patients until the last decade, when multiple trials demonstrated improved outcomes and less toxicity with agents targeting the VEGF and mammalian target of rapamycin (mTOR) pathways [3–11]. More recently, the treatment paradigm has shifted focus to ICIs due to evidence of improved overall survival (OS) in both the pretreated and the untreated setting [12,13]. With a plethora of systemic treatments available in mRCC, questions remain about which patient factors may affect the clinical efficacy of these various agents.
The gut microbiome has emerged as a key component of the host immune system and a central player in anticancer immunosurveillance [14,15]. Antibiotic use disrupts the microbial ecosystem, resulting in or predisposing individuals to various multisystem illnesses [16–18]. There has been recent interest in how the use of antibiotics influences the host microbiome in patients with various cancers. In this study, we evaluated the association of antibiotic use and clinical outcomes in mRCC patients treated with the range of approved systemic therapies, including cytokines, VEGF-TT, mTOR inhibitors, and ICIs. We leveraged both an institutional cohort and a clinical trial database to investigate which established mRCC therapies may be influenced by the concomitant use of antimicrobial treatment.
Patients and methods
Patients and data collection
We performed a retrospective analysis using two separate datasets. The first cohort (institutional) included all mRCC patients treated with anti–programmed cell death protein 1 (anti–PD-1)/programmed death-ligand 1 (PD-L1)–based ICI treatments at the Dana-Farber Cancer Institute (DFCI) from 2009 to 2017. For patients receiving multiple ICI-based regimens, baseline clinical information was captured for the first utilized ICI-based regimen. Data were collected on demographic information, smoking status, histology, International Metastatic Renal Cell Carcinoma Database Consortium (IMDC) risk factors (hemoglobin < lower limit of normal, corrected calcium > upper limit of normal [ULN], platelets > ULN, absolute neutrophil count > ULN, Karnofsky performance status < 80%, and time from diagnosis to systemic treatment < 1 yr), neutrophil-to-lymphocyte ratio (NLR) at the start of an ICI, type and line of systemic therapy, duration of treatment, best response to PD-1/PD-L1, scan date of progression, and date of death or last follow-up. Response and progression were determined by Response Evaluation Criteria in Solid Tumor (RECIST) version 1.1 [19] and were investigator assessed for the institutional cohort.
The trial-database cohort included mRCC patients treated in phase II (NCT00267748, NCT00077974, NCT00137423, NCT00054886, NCT00338884, and NCT00835978) and phase III (NCT00678392, NCT00083889, NCT00474786, NCT00920816, NCT00065468, and NCT00631371) clinical trials sponsored by Pfizer. This pooled analysis included patients treated with IFN, VEGF-TT, or mTOR inhibitors from 2003 to 2013. Patient follow-up consisted of imaging assessment every 12 wk until disease progression or withdrawal from study. Response was determined utilizing RECIST version 1.1 and was investigator assessed in all studies except NCT00083889, which had central radiographic response assessment. Treatment-associated toxicities were defined and evaluated according to the Common Terminology Criteria for Adverse Events, version 3.0.
In both cohorts, antibiotic use was defined as oral or intravenous systemic antibiotic treatment at any time from 8 wk before to 4 wk after initiation of anticancer therapy. This is aligned with previous clinical data originally establishing this time window for evaluation [20]. Further, this definition also takes into consideration preclinical data that while some gut microbiota may return to preantibiotic treatment levels after 4 wk, a plethora of diverse taxa fail to recover to pretreatment levels for periods up to 60 d or more [21].
Statistical analysis
Patient and disease characteristics were described using frequencies (percentages) and medians (ranges). Comparisons between antibiotic-use groups (users vs nonusers) were conducted by Fisher’s exact test for categorical variables and Kruskal-Wallis test for continuous values. We investigated the impact of antibiotic use on the objective response rate (ORR: complete response or partial response), progression-free survival (PFS), and OS. PFS was defined as the time from ICI initiation (for the institution cohort) or the time from the first dose of protocol therapy or date of randomization (for the trial-database cohort) to the date of progression or death, whichever came first, censored at the last follow-up for patients who have not progressed. OS was defined similarly, but only deaths from any cause were considered events.
Distributions of PFS and OS were estimated using Kaplan-Meier methodology. Differences between antibiotic-use groups were examined with log-rank test. For the institutional cohort, multivariable Cox proportional hazard models estimate hazards ratios (HRs) for OS and PFS, adjusted for line of therapy, type of therapy (monotherapy vs combination therapy), number of IMDC risk factors, baseline NLR, and presence of any poor histological components (non–clear cell, rhabdoid, or sarcomatoid differentiation). These covariates were included to account for heterogeneous ICI treatment and known prognostic factors; no formal model selection was used. For the trial-database cohort, the multivariable Cox regression models were adjusted for age, sex, race, histology, sites of metastasis, previous nephrectomy, previous therapy, IMDC risk factors, body mass index, angiotensin system inhibitor use, statin use, baseline hypertension, and diabetes. Subgroup analyses by type of therapy were conducted for both cohorts; however, for the institutional cohort, subgroup analysis was unadjusted given the sample size.
SAS version v9.4 (SAS Institute, Cary, NC, USA) was used to carry out the above analysis. All statistical tests were two sided, and statistical significance was considered at p < 0.05.
Results
Baseline characteristics and treatments—institutional cohort
The institutional cohort analysis included 146 patients who received anti–PD-1/PD-L1–based treatments at the DFCI, of whom 21% (n = 31) were considered antibiotic users. Baseline patient and disease characteristics were balanced between both groups (p > 0.15; Table 1). The median age was 61 yr (range: 22–82), 71.2% were male, 84.9% (n = 124) of patients had clear cell histology, and 15.1% (n = 22) had sarcomatoid differentiation. Approximately 60% (n = 85) of the patients were at an intermediate risk, while 17.8% (n = 26) were at a favorable risk and 21.2% (n = 31) at a poor risk. Sixty-three patients (43.2%) received ICI treatment in the first-line setting, 39 (26.7%) in the second-line setting, and 44 (30.1%) in the third-line setting or later. The majority received ICI monotherapy (n = 80, 54.8%). Overall, 95 patients received PD-1–based therapy (n = 75 monotherapy and n = 20 combination therapy) and 51 received PD-L1–based therapy (n = 5 monotherapy and n = 46 combination therapy). Fifty patients received standard of care PD-1 monotherapy, and 96 were treated as part of a clinical trial.
Table 1 –
Patient characteristics — institutional cohort.
| Characteristics | Total cohort N = 146 | Antibiotic users N = 31 (21%) | Antibiotic nonusers N = 115 (79%) | p valuea |
|---|---|---|---|---|
|
| ||||
| Age at start of therapy, median | 61 (22–82) | 61 (40–82) | 61 (22–80) | 0.711 |
| Gender | 0.504 | |||
| Male | 104 (71.2) | 24 (77.4) | 80 (69.6) | |
| Female | 42 (28.8) | 7 (22.6) | 35 (30.4) | |
| Smoker | 0.686 | |||
| No | 72(49.7) | 14(45.2) | 58(50.9) | |
| Yes | 73(50.3) | 17(54.8) | 56(49.1) | |
| ECOG | 0.999 | |||
| 0–1 | 133 (91.1) | 28 (90.3) | 105 (91.3) | |
| 2 | 9 (6.2) | 2 (6.5) | 7 (6.1) | |
| Unknown | 4 (2.7) | 1 (3.2) | 3 (2.6) | |
| Pathology | 0.164 | |||
| Clear cell | 124 (84.9) | 29 (93.5) | 95 (82.6) | |
| Non–clear cell | 22 (15.1) | 2 (6.5) | 20 (17.4) | |
| Presence of sarcomatoid or rhabdoid component | 0.289 | |||
| Yes | 26 (17.8) | 3 (9.7) | 23 (20.0) | |
| IMDC risk group | 0.533 | |||
| Favorable | 26 (17.8) | 7 (22.6) | 19 (16.5) | |
| Intermediate | 85 (58.2) | 16 (51.6) | 69 (60.0) | |
| Poor | 31 (21.2) | 8 (25.8) | 23 (20.0) | |
| Unknown | 4 (2.7) | 0 (0) | 4 (3.5) | |
| Line of therapy | 0.182 | |||
| 1 | 63 (43.2) | 9 (29.0) | 54 (47.0) | |
| 2 | 39 (26.7) | 11 (35.5) | 28 (24.3) | |
| ≥3 | 44 (30.1) | 11 (35.5) | 33 (28.7) | |
| Type of anti–PD-1/PD-L1 therapy | 0.542 | |||
| Monotherapy | 80(54.8) | 19(61.3) | 61 (53.0) | |
| Anti–PD-1 | 75 (51.4) | 19 (61.2) | 56 (48.8) | |
| Anti–PD-L1 | 5 (3.4) | 0 (0) | 5 (4.3) | |
| Combination therapy | 66(45.2) | 12(38.7) | 54 (47.0) | |
| Anti–PD-1 + anti-VEGF | 9 (6.2) | 2 (6.5) | 7 (6.1) | |
| Anti–PD-L1 + anti-VEGF | 46 (31.5) | 8 (25.8) | 38 (33.0) | |
| Anti–PD-1 + anti–CTLA-4 | 7 (4.8) | 1 (3.2) | 6 (5.2) | |
| Anti–PD-1 + other | 4 (2.7) | 1 (3.2) | 3 (2.6) | |
| Antibiotic classes | ||||
| Beta-lactams | 11 (35.5) | – | ||
| Fluoroquinolones | 7 (22.6) | – | ||
| Macrolide | 3 (9.7) | – | ||
| Tetracycline | 3 (9.7) | – | ||
| Other | 4 (12.9) | – | ||
| Unknown | 3 (9.7) | – | ||
CTLA-4 = cytotoxic T-lymphocyte–associated protein 4; ECOG = Eastern Cooperative Oncology Group; IMDC = International Metastatic Renal Cell Carcinoma Database Consortium; PD-1 = programmed cell death protein 1; PD-L1 = programmed death-ligand 1; VEGF = vascular endothelial growth factor.
Excluded “unknown group” in the comparisons.
Of the 31 patients treated with antibiotics, the majority (87%) received antibiotics within 1 mo before or after ICI initiation. The most common classes of agents were beta-lactams/beta-lactamase inhibitor combinations (39.3%), fluoroquinolones (25%), macrolides (10.7%), and tetracyclines (10.7%).
Baseline characteristics and treatments—trial-database cohort
A total of 4144 patients were included in the trial-database cohort, of whom 17% (n = 709) were considered antibiotic users. Baseline patient and disease characteristics were overall balanced between both groups (Table 2). The median age was 60 yr (range: 19–91), 70.7% were male, and 89.6% (n = 3713) of patients had clear cell histology. Two-thirds of patients had a prior nephrectomy, and the most common sites of baseline metastasis were the lung (77%), bone (28.5%), and liver (26.6%). Approximately 41% (n = 1704) of the patients were at an intermediate risk, while 14.7% (n = 611) were at a favorable risk, and 25.7% (n = 1064) at a poor risk. Overall, 510 patients (12.3%) received IFN, 660 (15.9%) received mTOR inhibitors, and 2974 (71.8%) received VEGF-TT (including sunitinib, axitinib, sorafenib, and bevacizumab).
Table 2 –
Patient characteristics — trial-database cohort.
| Characteristics | Antibiotic users N = 709 (17%) | Antibiotic nonusers N = 3435 (83%) | Total cohort N = 4144 |
|---|---|---|---|
|
| |||
| Age at start of therapy, median | 60 (22–82) | 60 (19–91) | 60 (19–91) |
| Gender | |||
| Male | 471 (66.4) | 2459 (71.6) | 2930 (70.7) |
| Female | 238 (33.6) | 976 (28.4) | 1214 (29.3) |
| ECOG | |||
| 0–1 | 682 (96.2) | 3383 (98.5) | 4065 (98.1) |
| 2 | 19 (2.7) | 40 (1.2) | 59 (1.4) |
| Unknown | 8 (1.1) | 12 (0.3) | 20 (0.5) |
| Pathology | |||
| Clear cell | 632 (89.1) | 3081 (89.7) | 3713 (89.6) |
| Non–clear cell | 51 (7.2) | 280 (8.2) | 331 (8.0) |
| Unknown | 26 (3.7) | 84 (2.1) | 100 (2.4) |
| Baseline site of metastasis | |||
| Bone | 197 (27.8) | 982 (28.6) | 1179 (28.5) |
| Lung | 565 (79.7) | 2625 (76.4) | 3190 (77.0) |
| Liver | 216 (30.5) | 887 (25.8) | 1103 (26.6) |
| Other | 565 (79.7) | 2784 (81.1) | 3349 (80.8) |
| Previous nephrectomy | |||
| Yes | 456 (64.3) | 2293 (66.8) | 2749 (66.3) |
| No | 253 (35.7) | 1142 (33.2) | 1395 (33.7) |
| IMDC risk group | |||
| Favorable | 76 (10.7) | 535 (15.6) | 611 (14.7) |
| Intermediate | 281 (39.6) | 1423 (41.4) | 1704 (41.1) |
| Poor | 235 (33.2) | 829 (24.1) | 1064 (25.7) |
| Unknown | 117 (16.5) | 648 (18.9) | 765 (18.5) |
| Type of systemic therapy | |||
| Interferon | 71 (10.0) | 439 (12.8) | 510 (12.3) |
| VEGF targeted therapy | 481 (67.9) | 2493 (72.6) | 2974 (71.8) |
| mTOR inhibitor | 157 (22.1) | 503 (14.6) | 660 (15.9) |
| Antibiotic classes | |||
| Beta-lactams | 311 (43.9) | – | |
| Fluoroquinolones | 213 (30) | – | |
| Macrolide | 53 (7.5) | – | |
| Tetracycline | 14 (2) | – | |
| Other | 118 (16.6) | – | |
ECOG = Eastern Cooperative Oncology Group; IMDC = International Metastatic Renal Cell Carcinoma Database Consortium; mTOR = mammalian target of rapamycin; VEGF = vascular endothelial growth factor.
Of the 709 patients treated with antibiotics, the most common classes of agents were beta-lactams/beta-lactamase inhibitor combinations (43.9%), fluoroquinolones (30%), and macrolides (7.5%).
Associations of antibiotic use and outcomes—institutional cohort
Median follow-up since the initiation of ICI therapy was 16.6 mo (range: 0.7–67.8). Median duration on therapy was 5.1 mo (range: <1–61.4), and 110 patients had discontinued ICI treatments at the time of analysis. The ORR was 30% (44 out of 146, 95% confidence interval [CI] 23–38%). Median PFS was 7.2 mo (95% CI 3.5–8.8) and 1-yr OS rate was 77% (95% CI 68–83). At the time of analysis, 100 PFS events and 53 deaths have been observed.
Patients who were treated with ICI and also received systemic antibiotics had a significantly lower ORR (12.9% vs 34.8%, p = 0.026; Table 3), shorter PFS (adjusted HR = 1.96, 95% CI 1.20–3.20, p = 0.007), and numerically worse OS (adjusted HR = 1.44, 95% CI 0.75–2.77, p = 0.270) when compared with those who did not receive antibiotics (Table 3 and Fig. 1). The negative association of antibiotic use with PFS was consistent in subgroup analysis by the type of therapy (monotherapy vs combination therapy), line of therapy (1 vs > 1) and by IMDC risk group (all interaction p > 0.35). Regarding specific type of ICI, the negative association of antibiotic use appeared more evident in PD-L1–based therapies (HR = 3.45) compared with PD-1 therapies (HR = 1.41); however, this was not significantly different (interaction p value = 0.09). Results were also consistent in sensitivity analyses when we restricted the timeframe for antibiotics use as −30 to +30 d around ICI initiation (adjusted PFS HR = 2.03, 95% CI 1.21–3.41; adjusted OS HR = 1.59, 95% CI 0.80–3.15).
Table 3 –
Outcomes of antibiotic users compared with those of nonusers in institutional and trial-database cohorts.
| No antibiotic use | Antibiotic use | p value | |
|---|---|---|---|
|
| |||
| Institutional cohort—overall | N = 115 | N = 31 | |
| ORR, N (%) | 40 (34.8) | 4 (12.9) | 0.026 |
| PFS, adjusted a HR (95% CI) | Ref | 1.96 (1.20–3.20) | 0.007 |
| OS, adjusted a HR (95% CI) | Ref | 1.44 (0.75–2.77) | 0.270 |
| Trial-database cohort b—overall | N = 3435 | N = 709 | |
| ORR, N (%) | 833 (24.2) | 137 (19.3) | 0.005 |
| PFS, adjusted HR | Ref | 1.16 (1.04–1.30) | 0.008 |
| OS, adjusted HR | Ref | 1.25 (1.10–1.41) | <0.001 |
| Trial-database cohort b—by treatment | |||
| Interferon | N = 439 | N = 71 | |
| OS, adjusted HR | Ref | 1.62 (1.13–2.31) | 0.008 |
| VEGF with prior cytokines | N = 444 | N = 76 | |
| OS, adjusted HR | Ref | 1.65 (1.04–2.62) | 0.033 |
| VEGF without prior cytokines | N = 2049 | N = 405 | |
| OS, adjusted HR | Ref | 1.16 (0.97–1.38) | 0.109 |
| mTOR inhibitors | N = 503 | N = 157 | |
| OS, adjusted HR | Ref | 1.13 (0.89–1.44) | 0.313 |
CI = confidence interval; HR = hazard ratio; IMDC = International Metastatic Renal Cell Carcinoma Database Consortium; mTOR = mammalian target of rapamycin; NLR = neutrophil-to-lymphocyte ratio; PFS = progression-free survival; ORR = objective response rate; OS = overall survival; Ref = reference; VEGF = vascular endothelial growth factor
Bold typeface indicates statistically significant results.
Institutional cohort was adjusted for line of therapy, type of therapy (monotherapy vs combination), number of IMDC risk factors, baseline NLR, and presence of any poor histological components (non–clear cell, rhabdoid, or sarcomatoid differentiation).
Trial-database cohort was adjusted for age, sex, race, sites of metastasis, prior nephrectomy, prior therapy, IMDC risk factors, body mass index, angiotensin system inhibitor use, statin use, baseline hypertension, and diabetes.
Fig. 1 –
Kaplan–Meier curves for (A) PFS and (B) OS according to antibiotic use (yes vs no) in the institutional cohort. CI = confidence interval; OS = overall survival; PFS = progression-free survival.
Associations of antibiotic use and outcomes—trial-database cohort
Overall in the trial-database cohort (n = 4144), the ORR was 19.3% versus 24.2% (p = 0.005), and adjusted HR was 1.16 (95% CI 1.04–1.30) for PFS and 1.25 (95% CI 1.10–1.41) for OS when comparing antibiotics users versus nonusers (Table 3 and Fig. 2). Similar results were observed in subgroup analyses by line of therapy and by IMDC risk groups (data not shown).
Fig. 2 –
Kaplan–Meier curves for (A) PFS and (B) OS according to antibiotic use (yes vs no) in the trial-database cohort. OS = overall survival; PFS = progression-free survival.
Interestingly, a more granular analysis of OS by type of anticancer therapy revealed that the association of antibiotic use and worse OS was limited to patients who received immune-based therapy (Table 3 and Fig. 3). In the group of patients who were treated with IFN (n = 510), antibiotic users were found to have a significantly worse OS compared with nonusers (adjusted HR = 1.62, 95% CI 1.13–2.31, p = 0.008). Similarly, in patients who were treated with VEGF-TT and received prior cytokines (n = 520), antibiotic users were also found to have a significantly worse OS compared with nonusers (adjusted HR = 1.65, 95% CI 1.04–2.62, p = 0.033). However, in patients who were treated with VEGF-TT and did not receive prior cytokines (n = 2454), there was no OS difference between antibiotic users and nonusers (adjusted HR = 1.16, 95% CI 0.97–1.38, p = 0.109). Furthermore, there was also no difference in OS between antibiotic users and nonusers in patients who were treated with mTOR inhibitors (adjusted HR = 1.13, 95% CI 0.89–1.44, p = 0.313).
Fig. 3 –
Kaplan–Meier curves for OS by treatment type according to antibiotic use (yes vs no) in the trial-database cohort: (top left) interferon, (top right) VEGF with prior cytokines, (bottom left) VEGF without prior cytokines, (bottom right) mTOR inhibitors. mTOR = mammalian target of rapamycin; OS = overall survival; VEGF = vascular endothelial growth factor.
Overall, the incidence of common adverse events (any grade) was similar between antibiotic users and nonusers, including fatigue (46%), diarrhea (45%), nausea (38%), and reduced appetite (37%). Pyrexia (35% vs 22%) and urinary tract infections (15% vs 4%) were more commonly noted in antibiotic users than in nonusers.
Discussion
In this analysis, we demonstrate that for mRCC patients treated with contemporary PD-1/PD-L1 ICIs, concomitant use of antibiotics near initiation of anticancer treatment was associated with worse clinical outcomes in terms of reduced ORRs and shorter PFS. In addition, utilizing a large clinical trial database, we show that patients who received IFN, or VEGF-TT with prior cytokines, had significantly shorter OS when they were treated with antibiotics, whereas patients receiving mTOR inhibitors, or VEGF-TT without prior cytokines, displayed no difference in OS when comparing antibiotic users with nonusers. Taken together, these data suggest that antibiotic use appears to affect outcomes in mRCC patients treated with immunotherapy and warrants further prospective validation, particularly to elucidate and ultimately manipulate the complex interplay between the host microbial ecosystem and anticancer immunity.
Our findings are consistent with and build upon previous reports evaluating the association of antibiotic use and outcomes in patients treated with PD-1/PD-L1 ICIs. Routy et al [20] examined 249 patients with a variety of solid tumors who were treated with ICIs in the second-line or beyond. Overall, antibiotic use was associated with significantly reduced PFS and OS. Their cohort included 67 patients with mRCC, and therein PFS was significantly reduced in antibiotic users (n = 20, 29%) compared with nonusers (4.3 vs 7.4 mo, p = 0.012). Derosa et al [22] reported on 121 patients with advanced RCC treated with ICIs, of whom 88% received ICI monotherapy. Sixteen patients (13%) were considered antibiotic users (defined as antibiotic use within 30 d prior to the initiation of ICI treatment), and these patients were found to have significantly reduced PFS (1.9 vs 7.4 mo, p < 0.01) and OS (17.3 vs 30.6 mo, p = 0.03) compared with nonusers. Our defined time window of antibiotic use was similar to that employed by Routy et al [20], and the results of our sensitivity analysis using a shorter timeframe were also consistent with results demonstrated by Derosa et al [22], lending creditability to our overall findings. Our study leverages both an institutional and a trial-database cohort, which includes patients receiving single-agent or combination therapies that reflect the evolving treatment landscape of mRCC. The institutional cohort in our study comprised 146 patients with mRCC, of whom 43% received first-line ICI therapy and 45% received combination ICI-based therapy. Given the shift in standard of care to include combination treatment for the majority of mRCC patients [2,13], our data may better reflect the contemporary incorporation of ICIs in the evolving treatment landscape.
There are important clinical implications of our data, particularly in the context of the range of systemic therapies available for mRCC. The analysis of our trial-database cohort suggests the concept that intestinal dysbiosis may be associated with reduced therapeutic effect of immunotherapy, as evidenced by significantly worse OS in antibiotic users receiving cytokines at some point along their treatment trajectory (Table 3 and Fig. 3). These findings may partly be explained by reports linking the gut microbiome to inflammatory cytokine production capacity. Schirmer et al [23] highlight significant microbial taxonomic associations with stimulus-specific monocyte or lymphocyte-derived cytokine responses, and Belizario et al [24] point out the lingering, systemic inflammatory effects of cytokines and the resultant disruption to microbiome health and homeostasis. For example, certain commensal bacterial species, including Bacteroides, Streptococcus, Lactobacillus, Actinomyces, and Bifidobacterium spp., have been shown to have positive correlations with cytokine responses. Antibiotic use may, therefore, interfere with these positive microbiome-cytokine interactions by disrupting the supportive ecosystem provided by these species—many of which are targeted by and sensitive to common antimicrobial agents [23]. Regarding patients treated with oral targeted agents such as VEGF-TT or mTOR inhibitors, our data suggest that any potential gut dysbiosis from antibiotic use did not translate into significantly reduced survival (Table 3 and Fig. 3). Interestingly, a recent report suggests that mRCC patients treated with first-line VEGF-TT alone who received antibiotics targeting Bacteroides spp. (n = 17) had a trend toward improved PFS compared with antibiotic nonusers (n = 96, HR = 0.52, 95% CI 0.20–1.00, p = 0.052) [25]. This is somewhat contrasted with accumulating evidence outlining the beneficial effect of Bacteroides spp. at the level of the colonic lamina propria, through potentiation of host antitumor immune responses [26,27]. In our subset of VEGF-TT–treated patients without prior cytokines, PFS was numerically worse by 1.2 mo in antibiotic users (n = 405)–the majority (61%) of whom received antimicrobials that would cover Bacteroides spp.–compared with nonusers, although this did not translate into an OS difference. Taken together, these data support further study of the impact of individual microbial phyla and species on cancer evolution and response to treatment, particularly as front-line treatment for most mRCC patients shifts to combination strategies.
Preclinical data have shown reduced therapeutic effects of immunotherapy in antibiotic-treated or germ-free mice [26,28–30]. The damaging of ileal epithelial cells by intraepithelial lymphocytes during cytotoxic T-lymphocyte–associated protein 4 (CTLA-4) blockade purportedly leads to the accumulation of Bacteroides fragilis and Burkholderiales spp., which activate dendritic cells and T-helper 1 immune responses [26]. The efficacy of PD-1/PD-L1 blockade is thought to be associated with the presence of intestinal Bifidobacterium spp., which activate antigen-presenting cells [30]. Recent investigations in melanoma patients–classified in groups as extreme responders or nonresponders to ICIs–have highlighted the concept of a ratio of putative “favorable” and “unfavorable” bacteria, which may distinguish these clinical subgroups [31]. This has led to optimism about potential interventional strategies in cancer patients treated with immunotherapy, for example, by eliminating “unfavorable” bacteria, providing colonies with immune-potentiating effects, or manipulating the diet using prebiotics, probiotics, and synbiotics [27,32]. Ongoing studies are evaluating the impact of probiotics or microbial transplantation in cancer patients treated with immunotherapy (NCT03829111 and NCT03772899). It is important to note that only 15% of bacteria grown from feces are estimated to be detectable using the currently available methods of metagenomics and 16S rRNA sequencing [33]. Therefore, in order to truly render the microbiome as “druggable,” optimized techniques of analysis, cultivation, and manufacturing remain an important, ongoing need [34].
These data should be interpreted in the context of the study design. First, this retrospective analysis has the potential for selection bias and unmeasured confounders. We attempted to adjust for key prognostic variables in multivariable models. Of note, our trial-database cohort included pooled data from 4144 patients treated in randomized, phase II/III clinical trials. Second, our institutional cohort included patients who were treated with PD-1 or PD-L1 ICIs, and there may be subtle differences between these drug pathways, alone or when employed in combination strategies. While it is unlikely that these slight mechanistic differences would significantly affect our overall findings, a prospective study would be informative. Third, we could not control for concomitant medications or patient diet, and data from the trial-database cohort were lacking about the specific indication, dose, and duration of antimicrobial use. Fourth, granular data for potential immune-related toxicities were not available in this analysis. Finally, while not the objective of this particular work, efforts are ongoing to capture blood and stool samples at the time of PD-1/PD-L1–based ICI treatments to better understanding the host mechanisms underlying our findings.
Conclusions
In this combined analysis of 4290 patients with mRCC, we demonstrate that antibiotic use was associated with worse outcomes in patients treated with either contemporary PD-1/PD-L1–based ICIs or cytokines. Antibiotic use did not appear to impact survival in patients treated with VEGF-TT without prior cytokines or mTOR inhibitors. Collectively, these findings further suggest the role antimicrobial agents potentially play in intestinal dysbiosis. These data underscore the importance of antimicrobial stewardship since prescribing patterns may affect subsequent outcomes of ICI-based therapies in these patients. Ultimately, prospective, large-scale efforts to better capture and analyze on-treatment fecal samples, with different bacterial subspecies, may lead to a deeper understanding of the host microbiome and elucidate meaningful intervention strategies to optimize efficacy and/or reduce toxicity of contemporary ICIs.
Acknowledgments:
The authors thank the patients and investigators who participated in the clinical trials used for this analysis. The phase II and III clinical trials were sponsored by Pfizer, Inc.
Funding/Support and role of the sponsor:
This research was supported in part by the Dana-Farber/Harvard Cancer Center Kidney SPORE, and the Trust Family, Michael Brigham, and Loker Pinard Funds for Kidney Cancer Research at Dana-Farber Cancer Institute for Toni K. Choueiri.
Footnotes
Financial disclosures: Toni K. Choueiri certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: Aly-Khan A. Lalani: honoraria/consulting from Bristol-Myers Squibb, Eisai, Ipsen, Merck, Pfizer, Roche, TerSera, Abbvie, and Janssen. Dominick Bossé: honoraria/consulting from Pfizer. Xun Lin: employment with Pfizer; stocks/ownership with Pfizer. Rana R. McKay: institutional research funding from Pfizer and Bayer; honoraria/consulting from Novartis, Janssen, and Tempus. Ronit Simantov: employment with Pfizer at the time of study design/analysis; stocks/ownership with Pfizer. Bradley A. McGregor: honoraria/consulting from Astellas, Seattle Genetics, Bayer, Astra-Zeneca, Genentech, and Exelixis. Lauren C. Harshman: consulting from Genentech, Pfizer, Dendreon, Medivation/Astellas, Merck, and Novartis; institutional research funding/support from Bayer, Astellas, Pfizer, Dendreon, Sotio, Genentech, Merck, BMS, and Janssen. Toni K. Choueiri: honoraria/consulting from AstraZeneca, Alexion, Sanofi/Aventis, Bayer, BMS, Cerulean, Eisai, Foundation Medicine Inc., Exelixis, Genentech, Roche, GlaxoSmithKline, Merck, Novartis, Peloton, Pfizer, EMD Serono, Prometheus Labs, Corvus, Ipsen, Up-to-Date, NCCN, Analysis Group, NCCN, Michael J. Hennessy (MJH) Associates, Inc (Healthcare Communications Company with several brands such as OnClive and PER), L-path, Kidney Cancer Journal, Clinical Care Options, Platform Q, Navinata Healthcare, Harborside Press, American Society of Medical Oncology, NEJM, Lancet Oncology, Heron Therapeutics; Institutional research funding/support from AstraZeneca, Bayer, BMS, Cerulean, Eisai, Foundation Medicine Inc., Exelixis, Ipsen, Tracon, Genentech, Roche, Roche Products Limited, GlaxoSmithKline, Merck, Novartis, Peloton, Pfizer, Prometheus Labs, Corvus, Calithera, Analysis Group, and Takeda. The remaining authors report no conflicts of interest.
Study concept and design: Lalani, Xie, Simantov, Lin, Choueiri.
Acquisition of data: Lalani, Kaymakcalan, Steinharter, Martini, Simantov.
Analysis and interpretation of data: Lalani, Xie, Simantov, Lin, Choueiri.
Drafting of the manuscript: Lalani, Xie, Choueiri.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Xie, Lin.
Obtaining funding: None.
Administrative, technical, or material support: None.
Supervision: Choueiri.
Other: None.
References
- [1].Choueiri TK, Motzer RJ. Systemic therapy for metastatic renal-cell carcinoma. N Engl J Med 2017;376:354–66. [DOI] [PubMed] [Google Scholar]
- [2].Lalani AA, McGregor BA, Albiges L, et al. Systemic treatment of metastatic clear cell renal cell carcinoma in 2018: current paradigms, use of immunotherapy, and future directions. Eur Urol 2019;75:100–10. [DOI] [PubMed] [Google Scholar]
- [3].Escudier B, Eisen T, Stadler WM, et al. Sorafenib in advanced clear-cell renal-cell carcinoma. N Engl J Med 2007;356:125–34. [DOI] [PubMed] [Google Scholar]
- [4].Motzer RJ, Hutson TE, Tomczak P, et al. Sunitinib versus interferon alfa in metastatic renal-cell carcinoma. N Engl J Med 2007;356:115–24. [DOI] [PubMed] [Google Scholar]
- [5].Sternberg CN, Davis ID, Mardiak J, et al. Pazopanib in locally advanced or metastatic renal cell carcinoma: results of a randomized phase III trial. J Clin Oncol 2010;28:1061–8. [DOI] [PubMed] [Google Scholar]
- [6].Motzer RJ, Escudier B, Oudard S, et al. Phase 3 trial of everolimus for metastatic renal cell carcinoma final results and analysis of prognostic factors. Cancer 2010;116:4256–65. [DOI] [PubMed] [Google Scholar]
- [7].Rini BI, Escudier B, Tomczak P, et al. Comparative effectiveness of axitinib versus sorafenib in advanced renal cell carcinoma (AXIS): a randomised phase 3 trial. Lancet 2011;378:1931–9. [DOI] [PubMed] [Google Scholar]
- [8].Motzer RJ, Hutson TE, Glen H, et al. Lenvatinib, everolimus, and the combination in patients with metastatic renal cell carcinoma: a randomised, phase 2, open-label, multicentre trial. Lancet Oncol 2015;16:1473–82. [DOI] [PubMed] [Google Scholar]
- [9].Choueiri TK, Escudier B, Powles T, et al. Cabozantinib versus everolimus in advanced renal-cell carcinoma. N Engl J Med 2015;373:1814–23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [10].Choueiri TK, Halabi S, Sanford BL, et al. Cabozantinib versus sunitinib as initial targeted therapy for patients with metastatic renal cell carcinoma of poor or intermediate risk: the Alliance A031203 CABOSUN trial. J Clin Oncol 2017;35:591–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [11].Motzer RJ, Hutson TE, Ren M, Dutcus C, Larkin J. Independent assessment of lenvatinib plus everolimus in patients with metastatic renal cell carcinoma. Lancet Oncol 2016;17:e4–5. [DOI] [PubMed] [Google Scholar]
- [12].Motzer RJ, Escudier B, McDermott DF, et al. Nivolumab versus everolimus in advanced renal-cell carcinoma. N Engl J Med 2015;373:1803–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [13].Motzer RJ, Tannir NM, McDermott DF, et al. Nivolumab plus ipilimumab versus sunitinib in advanced renal-cell carcinoma. N Engl J Med 2018;378:1277–90. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [14].Zitvogel L, Ayyoub M, Routy B, Kroemer G. Microbiome and anticancer immunosurveillance. Cell 2016;165:276–87. [DOI] [PubMed] [Google Scholar]
- [15].Zitvogel L, Daillere R, Roberti MP, Routy B, Kroemer G. Anticancer effects of the microbiome and its products. Nat Rev Microbiol 2017;15:465–78. [DOI] [PubMed] [Google Scholar]
- [16].Lange K, Buerger M, Stallmach A, Bruns T. Effects of antibiotics on gut microbiota. Dig Dis 2016;34:260–8. [DOI] [PubMed] [Google Scholar]
- [17].Korpela K, de Vos WM. Antibiotic use in childhood alters the gut microbiota and predisposes to overweight. Microb Cell 2016;3:296–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [18].Arrieta MC, Stiemsma LT, Dimitriu PA, et al. Early infancy microbial and metabolic alterations affect risk of childhood asthma. Sci Transl Med 2015;7:307ra152. [DOI] [PubMed] [Google Scholar]
- [19].Eisenhauer EA, Therasse P, Bogaerts J, et al. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1). Eur J Cancer 2009;45:228–47. [DOI] [PubMed] [Google Scholar]
- [20].Routy B, Le Chatelier E, Derosa L, et al. Gut microbiome influences efficacy of PD-1-based immunotherapy against epithelial tumors. Science 2018;359:91–7. [DOI] [PubMed] [Google Scholar]
- [21].Jakobsson HE, Jernberg C, Andersson AF, Sjolund-Karlsson M, Jansson JK, Engstrand L. Short-term antibiotic treatment has differing long-term impacts on the human throat and gut microbiome. PLoS One 2010;5:e9836. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [22].Derosa L, Hellmann MD, 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;29:1437–44. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [23].Schirmer M, Smeekens SP, Vlamakis H, et al. Linking the human gut microbiome to inflammatory cytokine production capacity. Cell 2016;167:1125–36, e8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [24].Belizario JE, Faintuch J, Garay-Malpartida M. Gut microbiome dysbiosis and immunometabolism: new frontiers for treatment of metabolic diseases. Mediators Inflamm 2018;2018:2037838. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [25].Hahn AW, Froerer C, VanAlstine S, et al. Targeting Bacteroides in stool microbiome and response to treatment with first-line VEGF tyrosine kinase inhibitors in metastatic renal-cell carcinoma. Clin Genitourin Cancer 2018;16:365–8. [DOI] [PubMed] [Google Scholar]
- [26].Vetizou M, Pitt JM, Daillere R, et al. Anticancer immunotherapy by CTLA-4 blockade relies on the gut microbiota. Science 2015;350:1079–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [27].Zitvogel L, Ma Y, Raoult D, Kroemer G, Gajewski TF. The microbiome in cancer immunotherapy: Diagnostic tools and therapeutic strategies. Science 2018;359:1366–70. [DOI] [PubMed] [Google Scholar]
- [28].Viaud S, Saccheri F, Mignot G, et al. The intestinal microbiota modulates the anticancer immune effects of cyclophosphamide. Science 2013;342:971–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [29].Iida N, Dzutsev A, Stewart CA, et al. Commensal bacteria control cancer response to therapy by modulating the tumor microenvironment. Science 2013;342:967–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [30].Sivan A, Corrales L, Hubert N, et al. Commensal Bifidobacterium promotes antitumor immunity and facilitates anti-PD-L1 efficacy. Science 2015;350:1084–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [31].Matson V, Fessler J, Bao R, et al. The commensal microbiome is associated with anti-PD-1 efficacy in metastatic melanoma patients. Science 2018;359:104–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [32].Bultman SJ. The microbiome and its potential as a cancer preventive intervention. Semin Oncol 2016;43:97–106. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [33].Sender R, Fuchs S, Milo R. Revised estimates for the number of human and bacteria cells in the body. PLoS Biol 2016;14:e1002533. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [34].Lagier JC, Khelaifia S, Alou MT, et al. Culture of previously uncultured members of the human gut microbiota by culturomics. Nat Microbiol 2016;1:16203. [DOI] [PMC free article] [PubMed] [Google Scholar] [Retracted]