PIK3CA in the transformation of EGFR‑mutant lung adenocarcinoma to small cell lung cancer following acquired targeted therapy resistance

To the Editor: Histologic transformation of epidermal growth factor receptor (EGFR)-mutant lung adenocarcinoma (LUAD) into small cell lung cancer (SCLC) represents a recognized mechanism of resistance to EGFR tyrosine kinase inhibitors (TKIs).^[1] This transformation is frequently associated with an aggressive clinical course and rapid deterioration. Although inactivation of tumor protein p53 (TP53) and retinoblastoma 1 (RB1) has been implicated in this process,^[2] the comprehensive clinicopathological features and molecular mechanisms remain to be elucidated. This retrospective cohort study aimed to describe the clinical characteristics and prognosis of patients with transformed SCLC (T-SCLC), and to investigate the genetic alterations associated with this phenomenon.

We identified patients with EGFR-mutant LUAD at The 900 Hospital of the Joint Logistics Support Force of the People’s Liberation Army of China, and conducted a retrospective analysis of their clinical and pathological data following treatment with EGFR-TKIs from January 2012 to January 2023. Histological classification followed the 2015 World Health Organization guidelines. Follow-up concluded in June 2024, with a minimum follow-up of 18 months for all patients. All cases were confirmed stage IV EGFR-mutant LUAD according to the tumor-node-metastasis (TNM) classification (Version 8) and had received first-, second-, or third-generation EGFR-TKIs. At the end of follow-up, all patients had developed TKI resistance. The study protocol was approved by the Ethics Committee of the 900th Hospital of the Joint Logistic Support Force of the Chinese People’s Liberation Army (No. 2024-045) and was conducted in accordance with the Declaration of Helsinki (2013 revision). Owing to the retrospective nature of the study, the requirement for written informed consent was waived by the Ethics Committee.

In total, 570 patients with advanced EGFR-mutant LUADs were evaluated, and nine patients showed evidence of T-SCLC, with an incidence rate of 1.58% [Supplementary Figure 1, http://links.lww.com/CM9/D14]. The median age was 54 years (range, 27–67 years), with four males and five females. Seven patients (77.8%) were nonsmokers [Supplementary Table 1, http://links.lww.com/CM9/D14]. Histological diagnoses were made from puncture biopsy tissue samples. Biopsy specimens of primary lung lesions provided 77.8% (7/9) and 66.7% (6/9) of evaluable tissue samples before and after transformation, respectively. Pure SCLC was identified in six patients, whereas three patients had mixed LUAD and SCLC [Supplementary Table 2, http://links.lww.com/CM9/D14]. Ki-67, a marker of cellular proliferation, showed a significant increase following transformation. The median Ki-67 expression values were 30% (range: 10–80%) in LUADs and 85% (80–95%) in T-SCLCs, with statistically significant differences (P <0.01), indicating significantly elevated proliferative activity [Supplementary Figure 2, http://links.lww.com/CM9/D14].

All patients in this study cohort received first-line treatment with EGFR-TKIs, and 66.7% (6/9) had baseline brain metastases [Supplementary Table 3, http://links.lww.com/CM9/D14]. The median progression-free survival (mPFS) with first-line TKIs was 12.0 months (95% CI: 10.5–13.5 months). The median time to transformation (mTTT), defined as the interval from the initial diagnosis of LUAD to histological confirmation of SCLC transformation, was 26.0 months (95% CI: 16.4–57.1 months) across all cases. In stratified analysis of different generations of TKIs, the mPFS was 11.0 months (95% CI: 7.8–14.2 months) for first-generation and 15.0 months (95% CI: 10.2–19.8 months) for third-generation TKIs. The corresponding mTTT was 14.0 months (95% CI: 8.0–29.6 months) and 25.0 months (95% CI: 16.4–57.1 months), respectively. Whether frontline third-generation TKIs delay SCLC transformation requires further real-world validation and mechanistic studies. After transformation, immune checkpoint markers were commonly negative, and platinum–etoposide with or without EGFR-TKI was the most frequently used regimen [Supplementary Table 4, http://links.lww.com/CM9/D14]. Posttransformation, the first-line mPFS (T-mPFS) was 6.0 months (95% CI: 5.2–6.8 months), and the median overall survival (T-mOS) was 17.0 months (95% CI: 9.0–25.0 months). Median OS from initial diagnosis of LUAD was 43.0 months (95% CI: 35.1–50.9 months) [Supplementary Figure 3, http://links.lww.com/CM9/D14]. In subgroup analysis, T-mPFS was 6.0 months (95% CI: 4.3–7.7 months) with chemotherapy alone and 5.0 months (95% CI: 1.1–9.0 months) with chemotherapy plus TKI, while T-mOS was 11.0 months (95% CI: 4.6–17.4 months) and 15.0 months (95% CI: 3.2–26.8 months), respectively.

In our study cohort, 88.9% (8/9) of cases harbored TP53 mutations and/or RB1 inactivation during SCLC transformation, and 33.3% (3/9) acquired new phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha (PIK3CA) mutations during tumor progression. After SCLC transformation, the initial EGFR common mutations (exon 19 del and L858R) were mostly retained, whereas T790M and MET proto-oncogene receptor tyrosine kinase (MET) amplification were not retained in the transformed tumors [Supplementary Table 5, http://links.lww.com/CM9/D14]. A representative case in our cohort demonstrated that the initial transformation event was characterized by concurrent TP53, RB1, and PIK3CA mutation. With further tumor progression, the PIK3CA mutation was persistently retained, suggesting a potential role in maintaining the transformed phenotype [Supplementary Figure 4, http://links.lww.com/CM9/D14]. In PIK3CA-mutant versus wild-type subgroups, T-mOS was 8.0 months (95% CI: 3.2–12.8 months) vs. 15.0 months (95% CI: 7.8–22.2 months), suggesting a possible poorer prognosis associated with PIK3CA mutation. No formal statistical comparisons were performed due to the limited sample size, and this observation requires validation in a larger cohort.

To determine whether genomic characteristics in our transformation cohort differed from other research cohorts, mutation data from The Cancer Genome Atlas lung adenocarcinoma study (TCGA-LUAD, n = 516) and a separate Chinese SCLC cohort (GSE60052, n = 99) available in the GEO database were downloaded for comparative analysis. The frequency of PIK3CA mutations was significantly higher in our cohort (33.3% [3/9] vs. 5.2% [27/516] in TCGA-LUAD, P <0.05). Among EGFR-mutant cases in TCGA (n = 66), the co-mutation rate of PIK3CA was 7.6% (5/66), still lower than in our cohort. Compared with classic SCLC in the Chinese population, PIK3CA mutations also remained more frequent in our cohort (33.3% [3/9] vs. 1.0% [1/99], P <0.05). Previous studies have suggested that T-SCLC arises primarily through clonal evolution in adenocarcinoma cells following mutation of TP53 and RB1. Using data from the TCGA-LUAD cohort, we evaluated the correlation between PIK3CA, EGFR, TP53, and RB1 at the transcriptional level. A positive correlation was observed between PIK3CA expression and both EGFR and RB1 expression [Supplementary Figure 5, http://links.lww.com/CM9/D14]. To explore PIK3CA’s role in SCLC transformation, we re-examined the longitudinal mutation profiles of our cohort. PIK3CA mutations often occur after TP53/RB1 inactivation (e.g., Patients 1 and 9), and their upregulation may play a role in maintaining the SCLC transition state. However, validation in a larger, independent cohort and laboratory models is required to definitively establish the frequency and clinical impact of PIK3CA mutations in SCLC transformation.

Histological transformation involves a shift in tumor morphology under selective pressure from targeted therapy. Although SCLC and non-small cell lung cancer (NSCLC) were historically regarded as distinct entities, shared oncogenic drivers such as PIK3CA mutations and common cellular origins provide plausible mechanistic overlap. TP53 and RB1 inactivation remain central to SCLC pathogenesis, and type II (surfactant protein C-expressing) cells, typically precursors of adenocarcinoma, can exhibit SCLC phenotypes following TP53/RB1 loss.^[3] T-SCLC is characterized by aggressive behavior, poor prognosis, and absence of standardized treatment. Our previous meta-analysis demonstrated response rates of 53% with chemotherapy and 50% with chemo-targeted combinations, but mPFS was limited to 3.5–4 months.^[4] In this cohort, programmed death receptor 1/programmed death ligand 1 (PD-1/PD-L1) expression was frequently negative after transformation. An artificial intelligence-based classification of TCGA NSCLC samples identified EGFR and PIK3CA mutations as prevalent in the immune-excluded subtype (41.9% and 51.6%, respectively). In TCGA-LUAD, PIK3CA expression correlated positively with tumor mutational burden, supporting the possibility that PIK3CA mutations contribute to genomic heterogeneity and immune evasion. It is crucial to emphasize that the identification of PIK3CA mutations in patient samples is a correlative finding and does not, by itself, prove causality. The definitive establishment of PIK3CA mutations as a driver of transformation requires further functional validation in larger model systems.

The phosphoinositide 3-kinase (PI3K) pathway is essential for cell proliferation and survival, and is frequently altered in tumors, including LUAD. In this study, PIK3CA mutations were detected in three of the nine patients with T-SCLC. In an exploratory analysis of the TCGA-LUAD cohort, the presence of PIK3CA mutations was positively correlated with EGFR and RB1 mRNA expression levels. However, given the limited sample size of our transformation cohort, the relevance of these correlations to SCLC transformation warrants further investigation. Marcoux et al^[5] reported TP53 (79%), RB1 (58%), and PIK3CA (27%) mutations in 58 T-SCLC cases, consistent with our findings. Correlation analysis of gene expression in our study showed significant positive associations between PIK3CA and EGFR, as well as RB1. These findings are consistent with previous reports that TP53 and RB1 alterations are commonly observed during SCLC transformation, while PIK3CA mutations may represent a candidate cooperating alteration that warrants further validation. PIK3CA alterations have also been implicated in EGFR‑TKI resistance. In the AURA3 trial, they were more frequent in tumors with subclonal vs. clonal T790M. These baseline samples were collected after progression on prior therapy and before osimertinib, so whether they pre‑existed or emerged during prior therapy could not be determined. PIK3CA H1047R reduced osimertinib sensitivity in vitro, which was partially reversed by PI3K inhibition.^[6] In our T‑SCLC cohort, PIK3CA mutations occurred in 3/9 patients (33.3%), higher than in EGFR‑mutant TCGA‑LUAD (7.6%) and classic SCLC (1.0%). Additionally, Quintanal-Villalonga A et al^[7] found that PI3K/AKT inhibition delayed neuroendocrine transformation in an EGFR‑mutant LUAD PDX model. Collectively, these observations suggest that PIK3CA alterations may act as secondary cooperating events in SCLC transformation, not merely resistance mediators. The primary limitation of this study is the small cohort size (n = 9), which is inherent due to the rarity of histologically confirmed T-SCLC; however, this severely limits the statistical power and robustness of our findings. Subgroup analyses are underpowered and should be interpreted with caution as hypothesis-generating only. Therefore, our findings, particularly regarding the frequency of PIK3CA mutations and the observed survival trends, require confirmation in larger, independent cohorts before any definitive conclusions can be drawn. Future prospective, multicenter studies with larger sample sizes are urgently needed to validate the frequency and clinical impact of PIK3CA mutations in T-SCLC and to establish optimal treatment strategies.

In conclusion, we propose an expanded molecular model for T-SCLC: TP53, RB1, and PIK3CA alterations may be involved in SCLC transformation in EGFR-mutant LUAD. PIK3CA genotyping at the time of T-SCLC diagnosis may enable patient stratification and support clinical trials of PI3K pathway inhibitors as potential novel therapeutic approaches. Due to the small sample size, these results warrant further validation in larger cohorts, complemented by functional assays and in vivo models to elucidate the mechanistic role of PIK3CA mutations in T-SCLC.

Funding

This work was supported by the 900th Hospital of the Joint Logistic Support Force of China:Youth Incubation Special Program (No. 2023QN04) and Fujian Medical University Sailing Fund Project (No. 2022QH1333).

Conflicts of interest

None.

Supplementary Material

cm9-139-2994-s001.docx ^{(979.1KB, docx)}