Comparison of the outcome between immunotherapy alone or in combination with chemotherapy in EGFR-mutant non-small cell lung cancer

Whether ICIs combined with chemotherapy can improve outcomes in EGFR-mutant non-small cell lung cancer (NSCLC) remains uncertain. Patients with EGFR-mutant NSCLC and who progressed on first-line EGFR-TKIs treatment were retrospectively collected. We reviewed the outcome of these patients treated with ICIs or ICIs combined chemotherapy (ICI + C). Total 30 patients were included. The ORR were 9.1% and 25.0% for the ICI and ICI + C groups. The ICI + C group showed the trend of longer progression-free survival and overall survival periods. Patients without the T790M mutation had a significantly longer PFS than did those without this mutation (4.23 [95% CI: 2.75–5.72] vs. 1.70 [95% CI: 0.00–3.51] months, HR:4.45, p = 0.019). ICIs combined with chemotherapy tended to be more effective than ICIs alone in pretreated EGFR-mutant NSCLC. The T790M mutation may be a potential biomarker.


Statistical analysis.
We compared patients' characteristics by using Pearson's chi-square test or Fisher's exact test. We performed a survival analysis using Kaplan-Meier survival curves for PFS and OS. The Cox proportional hazard regression model and logistic regression were applied to analyze clinical features and outcomes. A p value of < 0.05 was considered to be statistically significant, and all p values were two-sided. We used SPSS software (version 24.0, IBM Corp., Armonk, NY, USA) for all statistical analyses.
Ethical disclosure. This study was approved by the Institutional Review Board of Taipei Veterans General Hospital (2020-07-046CC).

Results
Patient characteristics. This study included a total of 30 patients. The median age was 66.5 years (45-85 years). Approximately 86.7% of the patients were never smokers, and 43.3% of the patients were men. Most of the patients (83.3%) had a favorable performance status (ECOG: 0-1). All patients had stage IV NSCLC with adenocarcinoma confirmed by histology. Moreover, all patients possessed EGFR mutations, with 56.7% having exon 19 deletion, 30.0% having an L858R mutation, and 13.3% having another uncommon mutation. The PD-L1 assay had been conducted for only 20% of the patients. The median number of treatment lines before ICIs therapy was 4 [3][4][5][6][7][8][9][10][11] . All patients had been treated with first-or second-generation EGFR-TKIs and subsequent chemotherapy. The objective response rate (ORR) was 76.7%. The median PFS for EGFR-TKIs was 12.20 months (95% confidence interval [CI]: 8.99-15.41). Before subsequent ICIs treatment, 53.3% of the patients had brain metastasis, and 10% of the patients had liver metastasis. Considering immunotherapy, 22 patients received immunotherapy alone (ICI group) and 8 patients received immunotherapy in combination with chemotherapy (ICI + C group). We observed no significant difference between the two treatment groups at baseline. Table 1 presents a summary of the characteristics. Different ICIs were used, including nivolumab (50.0%), pembrolizumab (6.7%), atezolizumab (10.0%), and durvalumab (6.7%). Moreover, 23.3% of the patients received nivolumab combined with chemotherapy, and 3.3% received pembrolizumab with chemotherapy (Supplementary Table 1).

Treatment outcomes of immunotherapy and combination chemotherapy. The ORRs in the ICI
and ICI + C groups were 9.1% and 25.0%, respectively. Furthermore, the disease control rates in the ICI and ICI + C groups were 54.6% and 87.5%, respectively ( Fig. 1; Supplementary Table 2). Among the 30 patients, only 8 had documented immunotherapy-related toxicity, which ranged from grade 1 to 2 (skin rash, liver enzyme elevation, fatigue). The median follow-up period of this cohort was 16 T790M mutation and treatment response. We then analyzed the patients for their EGFR mutation status and their treatment response to immunotherapy. We re-evaluated a total of 21 patients for their EGFR mutation profile (re-biopsy tissue or liquid biopsy) before ICIs treatment. These patients were divided into a T790M-positive group, namely T790M(+), and a T790M-negative group, namely T790M(−). Eighteen patients had no T790M mutation, and only 3 patients were found to be T790M positive. The baseline characteristics are presented in Supplementary    Table 4). These associations among clinical factors and prognosis were not noted in the OS analysis (Supplementary Table 5).

Discussion
We present the results of patients with EGFR-mutant NSCLC who were treated with ICIs or ICIs combined with chemotherapy in a real-world setting. We found that ICIs in combination with chemotherapy showed a trend of superior efficacy along with acceptable treatment response compared with monotherapy involving ICIs. These results are consistent with those reported by previous studies [18][19][20] . Several possible mechanisms underlie the unsatisfactory efficacy of ICI therapy in patients with EGFR mutations. Patients with EGFR mutations have lower tumor mutation burden (TMB) levels and reduced tumor-infiltrating lymphocytes compared with other patients; they thus have lower immunogenicity and antitumor immunity 22,23 . The application of TKIs may affect PD-L1 expression and the tumor microenvironment [25][26][27][28][29] , both of which may have an adverse effect on the efficacy of immunotherapy. For example, a study proposed that chemotherapy may contribute to an increase in neoantigens and boost the reaction to immunotherapy 30  For clinicians, this study demonstrated that ICI alone in EGFR-mutant NSCLC has poor outcome. Concomitant chemotherapy may improve the benefits of immunotherapy, however, the optimal strategy for such combinations requires further investigation. We evaluated the possible clinical characteristics affecting outcomes. After adjusting for age, sex, smoking history, performance status, and uncommon mutations, we observed that patients with liver metastasis may have poorer PFS than did those without such metastasis. A previous study also indicated that liver metastasis is an issue affecting immunotherapy because ICIs alone provide minimal therapeutic benefits 31 . This might be associated with the specific microenvironments and immunoregulation of the liver 20,32 . We observed that patients with EGFR-mutant NSCLC and liver metastasis showed inferior PFS. By analyzing our data, we identified that in late-line settings, liver metastasis was predictive of poor response. However, not all clinical factors were associated  www.nature.com/scientificreports/ with OS; only poor ECOG showed any predictive value for patient survival. This might be because the patients were all heavily pretreated and the small number of study groups caused no significant OS results. Nevertheless, one can reasonably assume that the performance status may be useful for predicting OS in these patients, given the complicated condition of patients in this late-line setting.
To identify patients that may have benefited from therapy, several biomarkers have been studied for ICIs in EGFR-positive cohorts. Smoking status have been reported as a clinical predictor of response to ICIs in EGFRmutant NSCLC 33 . A previous cohort study reported that T790M-negative patients may benefit from ICI treatment after TKI failure 25 . Yamada et al. also demonstrated that uncommon mutations and the absence of T790M mutations are predictive of positive ICIs outcomes 24 . A study on the IMMUNOTARGET cohort also supported the finding that different subtypes of EGFR mutations have different PFS periods after ICI treatment 34 . Lau et al. also reported the significant difference of treatment response to ICIs between common and uncommon mutations in retrospective study 35 . Patients with uncommon mutations may have a higher TMB 16 . In addition, patients showing acquired resistance not engendered by T790M mutations are likely to exhibit high PD-L1 levels 24,25 . High PD-L1 expression may be result from the activation of other alternative oncogenic pathway and these pateints may benefit from ICIs administration 36 . Our study confirmed that T790M remains a poor prognostic marker not only for ICIs alone but also for ICIs combined with chemotherapy. In addition, a longer duration of treatment with first-line EGFR-TKIs was associated with a longer PFS, but this finding is not consistent with previous reports. Ichihara et al. reported that the PFS for prior treatment with EGFR-TKIs was negatively associated with that for prior treatment with ICIs 37 . Liu et al. also reported a better response to subsequent immunotherapy in EGFR-mutant NSCLC with shorter PFS during EGFR-TKIs treatment 38 . They also conducted single cell RNAsequencing to prove the different tumor microenvironment between longer and shorter first-line TKI treatment groups. Comparing to their study, our cohort focusing on those with late-line treatment group. Our patients may receive more lines of treatment after EGFR-TKIs failure. In this cohort, tumor mutation loads were more strongly affected by T790M mutation than by previous EGFR-TKI treatment response. Therefore, for PFS, prior treatment with EGFR-TKIs was less informative than T790M as a biomarker under clinical circumstances.  www.nature.com/scientificreports/ Our study has several limitations. First, the retrospective observational cohort study design has inherent restrictions on the data available for analysis. For example, some molecular profiles, such as PD-L1 expression, was evaluated only in a subset of patients. This was probably due to PD-L1 expression was not extensively applied in the late-line settings. However, PD-L1 expression has been reported to be a significant predictor of ICIs efficacy for EGFR-mutated NSCLC patients in the subgroup analysis of the ATLANTIC study 39 . Further investigation focusing on the role of PD-L1 expression in ICIs combined chemotherapy may offer more information. Similarly, there was no data of TMB reported. Second, the total sample size was small. Thirdly, the chemotherapy regimens in our study were heterogeneous, including single agent navelbine, docetaxel, and doublet combining navelbine plus gemcitabine (Supplement Table 1). One patient received paclitaxel, carboplatin and bevacizumab. The heterogenicity was due to the lack of standard-of-care in late-line treatment. Meanwhile, it also reflected the unmet need in clinical practice. Finally, the previous treatment pathways and chemotherapy combinations may have differed among patients. Nevertheless, this is the first study examining the efficacy and outcomes of ICIs administered alone and in combination with chemotherapy in patients with EGFR-mutant NSCLC in an area with a high prevalence of EGFR mutations.

Conclusion
ICIs combined with chemotherapy may be more effective and beneficial than ICIs alone in pretreated patients with EGFR-mutant NSCLC in the real world. Furthermore, the T790M mutation can be used as a predictive biomarker for poor response to treatments comprising both ICIs alone and ICIs combined chemotherapy.

Data availability
The datasets generated during and/or analysed during the current study are not publicly available due to patients' privacy but are available from the corresponding author on reasonable request.