The Negative Impact of Antibiotics on Outcomes in Cancer Patients Treated With Immunotherapy

A New Independent Prognostic Factor?

A. Elkrief; L. Derosa; G. Kroemer; L. Zitvogel; B. Routy


Ann Oncol. 2019;30(10):1572-1579. 

In This Article

ATB-related Dysbiosis and Impact of ATB on Efficacy of Immune-check Point Inhibition: Clinical Data

The first retrospective study attempting to determine the impact of ATBs on the clinical efficacy of PD-1 or PD-L1 blockade (second line or higher) included 249 patients with NSCLC, RCC and urothelial carcinoma (UC)[10] (Table 1). Both univariate and multivariate analyses confirmed that ATB received in the critical timeframe was independently associated with reduced overall survival (OS).[10]

In addition to demonstrating worse outcomes in patients treated with ATB, another study of 168 patients with advanced NSCLC also elucidated that the route of ATB was also important.[45] However, this difference in outcome may be explained, at least in part, by the fact that patients receiving intravenous ATB were hospitalized and had a poor performance status, hence representing a high-risk group.

In these several reports and others including NSCLC patients treated with ICI, all but one study consistently demonstrated the negative impact of ATB on therapeutic outcome. In the sole negative study encompassing 74 NSCLC patients treated with nivolumab,[46] the ATB-treated arm included patients receiving ATBs until 3 months before ICI treatment (which is a relatively long period, longer than in any other study) and only included 20.3% of all patients (which is a surprisingly low percentage compared with other studies that usually only treated patients until 2 months or less in the ATB-treated arm).

The notion of gut microbiome repopulation was examined in a preclinical study in 2010 by Manichanh et al..[47] As expected, the investigators observed a 10-fold decrease in the bacterial load and reduced bacterial phylotype richness after 3 days of broad-spectrum ATB administration. One to four months after ATB discontinuation, the fecal samples regained a partial but incomplete regain in richness with shifts in the overall microbiome profile. In humans, Corbeil et al. showed that a 7-day exposure to cefprozil led to a loss of metagenome-defined species. Although recovery of the gut microbiota occurred after 3 months from ATB discontinuation, important and lasting changes in the genomic content were seen, especially in participants with a low baseline gut microbiota diversity.[48]

Based on these findings that suggest a time-dependent and partial repopulation of the gut microbiota after ATB discontinuation, it is important to consider which patients should be included in the ATB-treated arm. When should the cut-off relevant for posterior ICI treatment be set? At 1, 2 or 3 months pre-ICI therapy?

A study by Derosa et al. on 249 NSCLC and 121 RCC patients examined the impact of ATB within 1 month before the initiation of ICI in the first- or second-line settings, and in keeping with most of the aforementioned studies, worse clinical outcomes were observed.[49] However, the group set the cut-off to 2 months (rather than 1 month) before the initiation of ICI to include or exclude patients in the ATB or non-ATB groups, and found that the impact of ATBs on the clinically efficacy of ICI was less marked. This is in keeping with Kaderbhai's negative study described above (which set the cut-off at 3 months), suggesting that progressive recovery of the microbiota post-ATB treatment reverses the negative impact of ATB treatment on the efficacy of immunotherapy.

A similar study in 157 Italian patients with NSCLC reviewed the impact of ATB administered 1 month before the initiation of ICI to 3 months after.[50] In this study, the rate of ATB prescription was lower (17%) than in all previously mentioned studies. This study also shed light on the potential importance of duration of ATB, as shorter durations of ATB (6 days) did not impact patient prognosis compared with a longer course (9 days). Four other studies in the lung cancer population showed a similar trend. Despite various rates of prescription of ATB in these studies, all of them showed shorter OS and PFS in the ATB groups.[51–53]

Focusing on RCC alone, Lalani et al. conducted a large study on 4290 patients treated with PD-1/PD-L1 ICI, interferon and mTOR inhibitors or VEGF inhibitors. Among patients receiving ICI, ATB use was associated with a lower ORR and shorter PFS. However, no significant difference in OS was found in this study. It is important to note that only the patients receiving ICI experienced worse outcomes in the ATB group.

The first study to include melanoma patients showed that the duration of ATB was important, as cumulative ATB use, defined by ATB duration of over 10 days or two or more successive courses of ATB was associated with even lower PFS and OS.[54] When examining the timing of ATB treatment, it appeared that patients treated with ATB before ICI initiation had a shorter PFS and OS compared with those who were treated with ATB after the first administration of ICI. These findings were corroborated another study of melanoma patients, where the multivariate showed that ATB administration was associated with worse PFS.[55]

Focusing on an even shorter time frame of 14 days between ATB-use and first injection of ICI, Ahmed et al. conducted a retrospective study on 60 patients with a variety of cancer types receiving pembrolizumab, nivolumab, atezolizumab or a combination of ICI with chemotherapy. In spite of the heterogeneity of the disease types and treatment modalities, the authors observed an ATB-associated decrease in ORR as well as in PFS in particular when patients received broad-spectrum ATB. Interestingly, patients receiving narrow-spectrum ATB had a similar ORR and PFS as the ATB-free group.[56]

In order to better understand the link between pro-inflammatory conditions and gut health, Derikx et al. identified plasma citrulline as a potential marker.[57] Citrulline, a modified amino acid, is thought to be a surrogate marker for overall gut health as its plasma concentration tends to reflect the mass of functional enterocytes.[57] For example, patients with severe oral or GI mucositis secondary to allogenic hematopoietic stem cell transplantation exhibit a loss of small bowel epithelial cell mass correlating with low levels of circulating citrulline.[58] Ouaknine et al. showed that low levels of plasma citrulline were associated with worse PFS and OS (P = 0.0001) in NSCLC patients treated with nivolumab, and along the same vein, that ATB use was associated with lower levels of plasmatic citrulline.[51]

In sum, there is strong evidence indicating that ATB-induced dysbiosis is associated with poor therapeutic efficacy of ICI-based immunotherapy, suggesting a causal link between dysbiosis and poor therapeutic outcome (Figure 1). However, alternative explanations are possible. First, patients with a more advanced disease afflicted by an intrinsically poor prognosis might be more often in need of antibiotherapies. Arguing against this scenario, however, multivariate analyses correcting for ATB-independent markers of poor prognosis (such as age, tumor stage, histology, prior lines of treatment, recent hospitalization) consistently showed that ATB use was independently associated with worse outcome, suggesting that ATB-use indeed may be an independent factor. As an additional possibility, ATB-treated patients may represent a subgroup of individuals with an enfeebled immune system that are both prone to bacterial infection and refractory to anticancer immunotherapy. Thus, intrinsic or acquired immune defects (that predispose both to bacterial infection and poor immunosurveillance) might explain the correlation between ATB use and poor ICI efficacy without that the former would constitute a cause for the latter.

Figure 1.

Considerations when prescribing ATB in patients receiving immune-checkpoint inhibitors. ICI, immune-checkpoint inhibitors; NSCLC, non-small-cell lung cancer; RCC, renal cell carcinoma; ATB, antibiotics.