Our study findings suggest that there is a risk of developing FN in the cohort receiving atezolizumab. In the IMagyn050 trial, the only significant adverse event, regardless of investigator-assessed causality, occurring in 2% or more of patients in either group was FN (8% for atezolizumab and 4% for placebo) [19]. Atezolizumab, a monoclonal antibody, exerts its mechanism by binding to PD-L1, disrupting PD-L1/PD-1 interaction and consequently augmenting T-cell activity directed against neoplastic cells [20]. In a meta-analysis, Petrelli et al. examined patients treated with PD-L1 inhibitors across various tumour types. Among 9324 patients from 47 studies, they reported an FN incidence of 0.45% [21]. To address FN, primary prophylaxis with granulocyte colony stimulating factor (G-CSF) preparations based on the frequency of FN onset and empiric treatment of FN are necessary, and it is important to carry out evaluations by repeated monitoring of blood cultures and neutrophils 3–4 days after starting the empiric treatment [16, 17]. In addition, to reduce the progression of FN, it is recommended to take preventive measures such as conducting regular blood tests and carefully monitoring the patient’s condition while administering atezolizumab [22].

Recent studies have investigated the safety of atezolizumab-containing regimens, highlighting the incidence of FN as an adverse effect. For example, a study conducted by Hata et al. compared the safety of ABCP with that of the BCP combination for treating advanced non-squamous (NSQ)-NSCLC in Japanese patients, observing higher incidences of FN in the ABCP group [23]. Similarly, Shiraishi et al. compared the safety of atezolizumab and platinum plus pemetrexed, with or without bevacizumab, for metastatic NSQ-NSCLC in Japanese patients, noting higher incidences of FN in the atezolizumab-containing regimens, with FN rates of 10% vs 6% (P = 0.15) among patients treated with atezolizumab, carboplatin plus pemetrexed, and bevacizumab vs atezolizumab plus carboplatin with pemetrexed [24]. In addition, Amari et al. evaluated the safety and patient-reported outcomes of atezolizumab plus chemotherapy, with or without bevacizumab, in stage IIIB/IV NSQ-NSCLC patients, finding higher incidences of adverse events, particularly neutropenia (19.1% in platinum-pemetrexed-atezolizumab-bevacizumab; 23.6% in platinum-pemetrexed-atezolizumab.) [25]. Endo et al. analysed the adverse event profile of the ABCP combination therapy based on Japanese Adverse Drug Event Report data, emphasising the importance of effectively managing FN and skin-related adverse events [26]. Although the indications and regimen are not the same as in this study, the incidence of FN tends to be high, increasing the robustness of the results. All these studies highlight the increased risk of adverse events, including FN, with atezolizumab-containing regimens.

The growing use of ICIs in diverse cancer treatments has heightened the detection of rare immune-related adverse events (irAE). Among these, immune-related neutropenia, although infrequent, poses a significant risk of mortality due to its potential to lead to sepsis. Recognising and promptly addressing haematological irAE are vital for ensuring favourable outcomes in patients undergoing ICI therapy [27]. The mechanism of action for the increase in FN when atezolizumab is combined with other factors is not fully understood. FN can occur as a side effect for a few cancer treatments, including immunotherapies such as atezolizumab. It is believed to result from the suppression of the immune system, particularly the neutrophil count, which can leave the patient vulnerable to infections. Atezolizumab use, when combined with other patient-related risk factors, such as age, comorbidities, and prior chemotherapy, may further affect the immune system, potentially increasing the risk of FN [6, 7]. However, the specific mechanisms and interactions leading to this increased risk may vary depending on the context and individual patient characteristics, and further research is needed to fully elucidate this process. In addition, the genetic factors contributing to FN can vary among individuals, and there may be a few differences in FN susceptibility between populations in Japan and elsewhere. While there may not be a single conclusive genetic factor, several genetic variations can influence the risk of FN in patients undergoing chemotherapy, such as variations in genes responsible for drug metabolism, immune response, and the production of blood cells [28]. Comparing the frequency of FN between patients in Japan and other regions could potentially reveal differences in the prevalence of these genetic factors. To understand these genetic factors and their interactions, studies involving large cohorts from both regions would be necessary. Researchers would need to investigate specific genetic markers and variations that may predispose certain individuals to FN when undergoing chemotherapy. This research could help identify genetic factors that contribute to the varying frequencies of FN between different populations and inform personalised treatment approaches to reduce the FN risk.

The limitations of this study stem from the medical information database utilised. This database contains data from visits to specific medical institutions. Consequently, patients receiving anticancer drugs for NSCLC at the same institution were excluded based on predefined criteria. However, if a patient had prior anticancer treatment elsewhere, the data might only reflect cases occurring after second-line treatment. In addition, relevant events or outcomes from other medical facilities cannot be accurately identified, including the patient’s medical history, surgical history, radiotherapy, and prior treatment. Outcome validation was not performed, limiting the interpretation of point estimates such as incidence rates. Propensity scores could not be adjusted due to limited control cases and uneven group distribution. In addition, the database, collected for insurance claims, lacks information on non-essential items. This analysis only considers beta-lactams based on treatment guidelines [29]. Other antibiotics, like new quinolones, may be used in practice. Furthermore, in this study, the total dose of Carboplatin and Paclitaxel was recorded for each patient group, but the total dose per dose was not calculated. Therefore, it cannot be denied that the high dose per dose may have contributed to the high incidence of FN. However, Table 1 shows the mean and standard deviation (SD) of the number of cycles and drug administration. While there was no general difference in the distribution of the number of cycles between the exposure group and the control group, the total dose in the exposure group was small and the incidence of FN was higher than in the control group, so it is likely that the dose per dose did not contribute. The current study design cannot fully distinguish the independent contribution of neutropenia to the observed outcomes from the potential confounding effect of fever. The high co-occurrence of fever and FN makes it challenging to isolate the specific effects of neutropenia on the investigated parameters. Future studies employing stratification or matching based on the presence of fever may be necessary to definitively attribute the observed effects to neutropenia alone. Lastly, relying on the receipt database for FN identification may underestimate prevalence. The absence of neutrophil count data and the potential for missed cases due to terminology limitations are key constraints. In addition, since the database used in this survey does not allow for the acquisition of clinical test values, it was not possible to define outcomes using clinical test values.

As mentioned before, an elevated FN occurrence was observed with atezolizumab plus bevacizumab plus chemotherapy in patients with metastatic NSCLC in the IMpower150 clinical trial [12]. Although this international phase III clinical trial was conducted among a limited number of Japanese patients in a global population and direct comparisons were not possible due to differences in patient backgrounds with this DB study, there were a few similarities between these 2 studies. The IMpower150 trial showed a higher incidence (19.4%) of FN with atezolizumab involving chemotherapy in the ACP or ABCP group than in the control BCP group (4.2%). This clearly demonstrates that FN is a concern in both studies, with higher incidence rates noted in exposed groups.

To the best of our knowledge, the present DB study represents a pioneering effort in exploring the association between atezolizumab and the incidence of FN in a real-world context, using a comprehensive Japanese medical records database. While acknowledging the existence of the IMpower150 trial, which shares certain similarities with our DB study, it is important to note that, apart from this trial, no other investigations of comparable scope and nature have been conducted. Thus, this study contributes valuable insights into the real-world implications of atezolizumab use in relation to FN occurrence, filling a notable gap in the existing research landscape.