Patients
We retrospectively studied patients who underwent complete resection of NSCLC at the National Hospital Organization Kinki-Chuo Chest Medical Center (KCMC) between February 2002 and January 2021, and selected 67 patients with a pathological diagnosis of PPC. We defined complete resection as a grossly or microscopically removed tumor, which corresponds to R0 in the residual tumor (R) classification. The histopathological diagnosis according to the current 2015 World Health Organization classification was performed by pathologists belonging to KCMC [16]. Medical information, including age, sex, smoking status, surgical procedure, pathological tumor-node-metastasis (TNM) classification (eighth edition), PD-L1 expression status, and the outcome after surgery were collected from medical records. The present study was approved by the Institutional Review Board of KCMC (Approval number: 2020–067). The cut-off date for the analysis was set at June 30, 2022. The Institutional Review Board of KCMC waived the requirement for informed consent from all research participants due to the retrospective and anonymous nature of this study. Information about opting out of the study was provided on the homepage of KCMC. All methods were in accordance with relevant guidelines and regulations.
PD-L1 immunohistochemistry
We evaluated all viable cancer cells in the entire pathological tissue section of each tumor sample. The PD-L1 clone 22C3 pharmDx kit and Dako Automated Link 48 platform (Agilent Technologies, Dako, Carpinteria, CA, USA) were used to examine the PD-L1 expression. We calculated the PD-L1 tumor proportion score (TPS) as the percentage of complete or partial membrane staining in a sample. The cut-off values for the expression of PD-L1 were set at 50 and 1% based on a previous clinical study. We categorized the tumor samples of each patient into 3 groups (< 1% [negative], 1–49% [low expression], and ≥ 50% [high expression]) based on the presence of positively stained cells in specimen.
Genomic DNA extraction and quality check
To investigate alterations in cancer-related genes, we obtained 67 lung tissues specimens from patients with PPC by surgical resection or trans-bronchial lung biopsy. All samples were fixed by formalin and embedded in paraffin (FFPE). Genomic DNA (gDNA) was extracted from FFPE samples using QIAamp DNA FFPE Tissue Kit (QIAGEN, Japan) according to manufacturer’s protocol. After extraction, the DNA concentration was quantified by Qubit® 3.0 Fluorometer (Thermo Fisher Scientific, USA). The gDNA quality was determined by the A260/A280 and A260/A230 ratios using NanoDrop 1000 (Thermo Fisher Scientific). We excluded 8 of the 67 samples due to failure to extract a sufficient amount of DNA.
Library preparation and sequencing with the Cancer HotSpot panel
DNA libraries were generated by PCR using AmpliSeq for Illumina Cancer HotSpot Panel v2 (Illimina, USA) in 59 samples. AmpliSeq for Illumina Cancer HotSpot Panel v2 targets 2800 mutations in 50 cancer-related genes, including the genes of interest—TP53, ATM, PIK3CA and EGFR—in our study. gDNA (20 ng) was used to amplify the panel range of genes. The libraries were amplified using AmpliSeq Library PLUS for Illumina (Illumina, USA) according to manufacturer’s instructions. Each library was quantified using an Agilent DNA 100 kit with an Agilent 2100 Bioanalyzer (Agilent Technologies, USA). Sequencing was performed using a MiSeq NGS system (Illumina, USA). The MiSeq sequence data was processed and analyzed using BaseSpace Sequence Hub. Of the 59 samples, we excluded one sample because it was impossible to analyze due to an error. In comparison to the UCSC hg19 reference genome, sequences with amino acid changes were identified as somatic mutations.
Overall survival and relapse-free survival
The primary outcome of this study was overall survival (OS) after surgery, defined as the time from the date of curative resection to the date of death by any cause or the date of the last follow-up examination. The secondary outcome was relapse-free survival (RFS), which was defined as the length of time for which the patient survived after curative resection (as the primary treatment) without any signs or symptoms of cancer or death from any cause.
Statistical analyses
Finally, 55 patients were included in the statistical analysis (three patients of 58 patients in whom a DNA analysis of PPC was performed were excluded because we were unable to obtain medical information). The probability of OS was assessed using the Kaplan–Meier method and log-rank tests. A multivariable Cox proportional hazards analysis was performed to estimate the hazard ratios (HRs), with adjustment by risk factors for mortality or relapse. In a multivariable Cox proportional hazards analysis, the number of covariates that be can analyzed is the number of cases with a particular outcome ÷ 10. In other words, a multivariable Cox proportional hazards model requires a minimum of 10 outcome events per predictor variable (EPV) [17, 18]. However, it also suggests that the rule of using ≥10 EPV in Cox models is not clearly defined and can be relaxed. That is, a simulation study suggested that, in comparison to a Cox model with 10–16 EPV, a Cox model with 5–9 EPV provided acceptable results [19]. Therefore, we accepted 5–9 EPV the in the Cox model. In this study, the number of covariates that could be included in the Cox proportional hazards analysis of OS was 15 (i.e., the number of patients who died) divided by 5–9 (result: 1–3) and the number of RFS was 28 (i.e., the number of patients who had a confirmed relapse) divided by 5–9 (result: 3–5). It has been reported that older age, male sex, and advanced pathological stage tend to be associated with a poor prognosis in patients with treated PPC [20, 21]. Therefore, in addition to the gene status variable of interest, we selected age, sex, and pathological stage as confounding factors. Since “elderly” is generally defined as age ≥ 65 years, we set 65 years as the cut-off value for age. We selected the following three factors in the multivariable Cox hazards analysis of OS: gene status (pathogenic mutation [reference: variants of unknown significance (VUSs) or wild type (WT)]) as the explanatory variable; and pathological stage (stage III–IV [reference: stage I–II]) and age (≥65 [reference: < 65]) as confounding factors. Sensitivity analyses were performed by the multivariable Cox hazards analyses of OS in which pathological stages (stage III–IV [reference: stage I–II]) and sex (female [reference: male]), and age (≥65 [reference: < 65]) and sex (female [reference: male]), respectively, were included as confounding factors were performed to assess the robustness of the results of the analysis. On the other hand, we selected the following four factors according to the multivariable Cox hazards analysis of RFS: gene status (pathogenic mutation [reference: VUSs or WT] as the explanatory variable; and pathological stage (stage III–IV [reference: stage I–II]), age (≥65 [reference: < 65] and sex (female [reference: male]) as confounding factors. All statistical analyses were conducted using Easy R (EZR) (Saitama Medical Center, Jichi Medical University, Saitama, Japan), which is a graphical user interface for R (The R Foundation for Statistical Computing, Vienna, Austria). EZR is a modified version of R commander with added biostatistical functions [22]. P values of < 0.05 were considered to indicate statistical significance.