New Actions on Actionable Mutations in Lung Cancers

Actionable mutations refer to DNA alterations that, if detected, would be expected to affect patients' response to treatments [...].

NSCLC patients with targetable mutations generally benefit less from immune checkpoint blockade (ICB), with possibly the exception of BRAF mutations and KRAS mutations [10,11]. In this Special Issue, Hong et al. specifically evaluated the efficacy of adding anti-PD1/L1 therapy to platinum-based chemotherapy in TKI-resistant EGFR-mutant NSCLC using a relatively large real-world cohort (n = 178). The study found that immunotherapy adds limited benefit to platinum doublets regardless of PD-L1 levels [12]. In addition, it is known that mutations in cancer genes such as STK-11 or KEAP1 are associated with ICB benefit independent of PD-L1 and tumor mutation burden [13,14], highlighting the critical potential of these genomic features as predictive biomarkers for ICB treatment. Three studies in this Special Issue attempted to identify novel genomic features as predictive markers to guide ICB-based therapy. Zhang et al. reported that mutations in HSPG2 were associated with benefits from ICB in melanoma and NSCLC patients [15]. Similarly, Wang et al. reported that mutations in fatty acid synthase were related to superior benefits from ICB in a large cohort of melanoma and NSCLC patients [16]. Furthermore, a study led by Yu et al. on lung squamous cell carcinoma observed that TP53 wildtype, especially when co-occurring with LRP1B wildtype, is associated with improved survival after anti-PD1 therapy [17]. It is anticipated that with the accumulation of genomic profiling data from patients who received ICB-based treatment, additional genomic features will emerge as potential predictive biomarkers in NSCLC patients with or without actionable mutations.
Biomarker-based therapeutic decision-making is the foundation of modern precision oncology. However, tissue-based tests often face limitations due to inadequate specimens [5,6,[18][19][20] and intra-tumor heterogeneity [5][6][7]21]. Moreover, longitudinal tissuebased profiling is often not feasible in clinical practice. Radiological images contain rich information that reflects the anatomical and functional characteristics of the tumor and its microenvironment. However, these images' complex anatomical and morphological features surpass the analytic capacity of human eyes. Therefore, artificial intelligence (AI), such as machine learning, has become a promising modality for extracting informative data from these intricate images [22]. Two radiogenomics studies applied machine learning approaches to predict oncodriver mutation status and PD-L1 level in this issue, showing promise. He et al. used a machine-learning approach to predict EGFR mutation status [23]. At the same time, Shao et al. developed a multi-label multi-task deep learning system for the same purpose to predict the mutation status of multiple oncodrivers and PD-L1 levels [24]. As the performance of machine learning depends heavily on sample size and data quantity, future studies with larger sample sizes and high-quality image/molecular data are expected to improve radiogenomic predictions further.
In addition to predicting therapeutic response, actionable mutations have also been used for prognostication. For example, in this Special Issue, Tian et al. reported the value of testing for oncodriver alterations in detecting occult metastatic disease in morphologically negative lymph nodes [25]. In contrast, Zhao et al. reported that oncogenic EFNA4 amplification may promote lymph node metastasis and be associated with poor prognosis in lung adenocarcinomas [26].
With ongoing efforts in molecular profiling of NSCLC, more actionable mutations are being discovered, and more actionable mutations are becoming targetable, such as the recent example of Kras G12C [27]. Furthermore, patients with NSCLC are subtyped into different molecular subgroups with varying response profiles based on co-mutations [28]. Therefore, it is reasonable to anticipate that most, if not all, NSCLC tumors will eventually be found to carry mutations that we can act on.

Conflicts of Interest:
The authors declare no conflict of interest.