br In lung cancer di erent agents have been
In lung cancer, diﬀerent agents have been approved or are in ad-vanced phase of study for locally advanced and metastatic disease. In many trials ICIs have shown a benefit over traditional cytotoxic agents in terms of survival both in first and in subsequent line of therapy [3–14]. Despite the general validity of this observation, it is also well known that only a limited number of patients really benefit from IO [13,14,17,19–22]. While about one-third of cases show disease re-sponse and may expect a quite long survival, the remaining two-thirds never respond to ICIs maintaining a dismal prognosis [13,14,17,19–22].
Given the relevance of this topic, many studies have focused on potential variables able to predict the response to IO. Data obtained with diﬀerent agents and in diﬀerent settings have supported a role for PD-L1 expression, tumor mutation burden and microsatellite instability [13,14,17,38–40]. Nonetheless, it is likely that the majority of factors accounting for such a variability still remains unknown. In this field, research has recently been developed about gut microbiota and its imbalances due to antibiotics.
In this work, we performed a single Institution retrospective ana-lysis of a cohort of patients with metastatic NSCLC treated with IO. At first, we considered the administration of Lipo2000 in the EIOP, sig-nificant for stratification purposes. We did not identify any diﬀerences in RR, PFS and OS between antibiotic treated and non-treated patients, although a numerical tendency in favor of non-treated cases was evi-dent. Afterwards, when we evaluated the impact of AIER on PFS and OS in the WIOP, this continuous variable showed to have a significant detrimental eﬀect on outcome. Given this results, we performed a second analysis using the median value of AIER as a cut-oﬀ for patient stratification. When we stratified the population according to AIER, a significant advantage emerged for patients with an AIER inferior to the median value of 4.2%.
The AIER is a completely new empiric variable, that we proposed in order to explore whether the length of antibiotic therapy in relation to that of IO could be more influential that the timing of antibiotic ad-ministration. To our knowledge, this is the first time that a variable influencing the duration of treatment and not its relation to the be-ginning of IO shows a correlation to outcome. The potential rationale behind this observation may lay in the physiological capability of in-testinal flora to regain homeostasis after an alteration. In other words, each antibiotic administration induces an imbalance in bacterial species populating the intestinal tract. After the end of the antibiotic cycle, the commensal bacteria species progressively return to their primitive composition and the alteration induced by treatment is overcome. If the
Median 95% CI3 p value Median 95% CI p value
1 Progression Free Survival.
2 Overall Survival.
3 Confidence Interval.
4 Performance Status.
5 Immune Checkpoint Inhibitor.
7 Not Reached.
9 Early Immunotherapy Period.
10 Antibiotic-Immunotherapy Exposure Ratio.
11 Whole Immunotherapy Period.
process is limited in time, in particular, if IO is carried on meanwhile, it is likely that the eﬀects of antibiotics on intestinal flora and, conse-quently, on anti-tumor immune response, is negligible. On the contrary, if a patient receives multiple or prolonged antibiotic cycles, the re-peated hits on gut microbiota may interfere with bacterial reconstitu-tion, leading to a deeper impact on systemic immunity. The significant relation between AIER and outcome, irrespective of the eﬀects of PS and IO line, and in absence of a clear role for the simple antibiotic use at the beginning of IO, supports the potential role of AIER as a new in-dependent determinant.
As regards to the lack of association between antibiotic use in the
EIOP and outcome, our results diﬀer from most recent research data showing a significant negative impact of early antibiotics use on PFS. We can be made only hypotheses to explain this discrepancy, maybe due to the population characteristics and to the small sample size. In any case, a numerical trend towards superior PFS and OS in patients not receiving antibiotics was identified, in line with literature data. Nonetheless, it has to be underlined that also a previous retrospective work on 74 NSCLC cases treated with nivolumab did not find diﬀer-ences in RR and PFS according to antibiotic use in the 3 months before the beginning of the ICI treatment . This underlines the lack of definitive evidence in the field of microbiota and IO, in particular for