We conducted a multi-center retrospective study screening all consecutive NSCLC patients diagnosed with stage I-II NSCLC and molecular assessment performed between 2016–2018 in Västra Götaland, Sweden (n = 354). Further inclusion criteria included the availability of tumor size from CT scanning or a pathology report as well as follow-up data. Patients were excluded if diagnosed before 2016, had no digitally accessible patient charts, no tumor measurements noted in the patient charts, or had recurrent disease (study cohort n = 310).

Patient demographics (age, sex, Eastern Cooperative Oncology Group [ECOG] performance status [PS], and smoking history), cancer stage, pathological details (histology, mutational status including KRAS mutational status and subtype), first-line treatment and outcome data were retrospectively collected from patient charts and the Swedish Lung Cancer Registry. Clinical staging was based on TNM staging guidelines 7^th edition (4). TNM staging 8^th edition released in 2017 was introduced in Swedish guidelines in 2018, and full implementation was reached in 2019. Ethical approval was obtained from the Swedish Ethical Review Authority prior to study commencement (Dnr 2019–04771 and 2021–04987). No informed consent was required due to all data presented in a de-identified form according to the Swedish Ethical Review Authority.

Patients were assessed with next-generation sequencing (NGS) for mutational status on DNA from formalin-fixed paraffin-embedded (FFPE) blocks or cytological smears using the Ion AmpliSeq™ Colon and Lung Cancer Panel v2 from Thermo Fisher Scientific as part of the diagnostic workup process at the Department of Clinical Pathology at Sahlgrenska University Hospital, assessing hotspot mutations in EGFR, BRAF, KRAS, and NRAS. Until June 2017, ALK-fusions were assessed with immunohistochemistry (IHC), and with fluorescence in situ hybridization (FISH) if positive or inconclusive IHC. ROS1 was analyzed upon request with FISH. Thereafter, ALK, ROS1, and RET fusions were assessed on RNA using the Oncomine Solid Tumor Fusion Panel from Thermo Fisher Scientific.

To obtain the most recent and accurate untreated primary tumor size, measurements were collected from the radiology report of computed tomography (CT) performed before a final diagnosis of NSCLC was established; this is referred as clinical staging. In patients who underwent surgical resection, the actual primary tumor size was also collected from the pathology report, also referred as pathological staging (PAD). The largest tumor diameter was collected and reported in millimeters.

The primary outcome of this study was OS, defined as the interval between the date of first treatment and the date of death from any cause. Patients alive or lost to follow-up at the cut-off date were censored at last contact. Median follow-up time was estimated using the reverse Kaplan-Meier method. We compared OS and risk of death stratified by KRAS mutational status, i.e., with no mutation in KRAS (wildtype, KRAS ^WT), all KRAS mutations (KRAS ^MUT), KRAS G12C mutations (KRAS ^{MUT G12C}) and all KRAS mutations other than G12C (KRAS ^{MUT not G12C}).

Clinical characteristics were summarized using descriptive statistics and analysis of associations between KRAS mutational status and clinicopathological parameters was performed using Pearson´s X² test or T-test. Survival was estimated using the Kaplan-Meier method, visualized at 5-year follow up. The log-rank test was used to assess significant differences in OS between KRAS ^WT and KRAS ^MUT groups. To evaluate if there was a significant difference in primary tumor size between KRAS ^MUT and KRAS ^WT, the Mann-Whitney U test was used. We defined an interaction term between tumor size (largest diameter in mm) and KRAS mutational status to assess the combined impact on the risk of death (HR). First, the mean size of all primary tumors was calculated. Thereafter a dummy variable was calculated by subtracting the mean size from all individual measurements following multiplication with 1 if KRAS was mutated and 0 if KRAS was WT. The interaction term was included in Multivariable Cox regression analysis, also correcting for sex, age, tumor size in mm and KRAS mutational status. Statistical significance was set at p < 0.05 and no adjustments were made for multiple comparisons. Data analysis was conducted using IBM SPSS Statistics version 27 and GraphPad Prism version 9.

A total of 310 patients, who were diagnosed with stage I-II NSCLC during 2016–2018 in Västra Götaland, Sweden and for whom genetic data was available, were included in this retrospective cohort study ( Figure 1 ). In the total population, majority of patients were female (187, 60.3%), with a median age of 70 years, and most were current or former smokers (267, 86%) ( Table 1 ). Most patients had good PS with ECOG 0–1 at diagnosis (285, 92%) and the proportion of N1 was low (18, 5.8%). NSCLC was predominantly adenocarcinoma of the lung (281, 90.6%), while squamous cell carcinoma incidence was relatively low (11, 3.5%), which was expected due to the selection of histological type for NGS assessment. Of included patients, over a third (115, 37%) had a KRAS mutation ( Table 1 ). This percentage matches what has been previously reported (9), showing good representativeness of the patient group studied here. When comparing the baseline characteristics of KRAS ^WT with KRAS ^MUT patients, a greater proportion of those with KRAS ^MUT were female and current or former smokers. There were no cases of squamous cell carcinoma in the KRAS ^MUT group. The most common KRAS mutation was G12C (47%). In the total population, majority of patients underwent surgical resection (273, 88%; Table 2 ). Three patients did not receive any treatment and were excluded from further survival analyses. Median follow-up time was 63 months (95% CI, 59.7–68.3) and the data cut-off date was 31 October 2022.

When comparing OS for all (stage I-II) patients stratified by KRAS mutational status, no significant difference was detected with a mean OS (median not reached) of 74 months for KRAS ^WT vs 63 months for KRAS ^MUT (p = 0.847; Figure 2A ). Further stratification of the KRAS mutated group by the G12C mutation also did not significantly change survival: 74 months for KRAS ^WT, 61 months for KRAS ^{MUT not G12C} and 63 months for KRAS ^{MUT G12C} (p = 0.834; Figure 2B ).

There were also no significant differences according to KRAS mutational status in stage I ( Figures 2C, D ) or Stage II ( Figures 2E, F ). Similarly in resected patients, no significant difference was observed with a mean OS (median not reached) of 78 months for KRAS ^WT vs 65 months for KRAS ^MUT (p = 0.856; Supplementary Figure 1A ), or between the subgroups of KRAS ^MUT (p = 0.471; Supplementary Figure 1B ).

Next, we stratified by stage and found mean OS (median not reached) of 79 months for stage I vs 50 months for stage II ( Supplementary Figure 2A ). We then conducted the analysis separately according to KRAS mutational status. For KRAS ^WT, the mean OS (median not reached) was 78 months for stage I vs 46 months for stage II ( Supplementary Figure 2B ), and for KRAS ^MUT, 65 months for stage I vs 53 months for stage II ( Supplementary Figure 2C ).

Next, we stratified patients using T-staging and studied OS according to KRAS mutational status. Among those with T1 disease, KRAS ^WT had 83 months while KRAS ^MUT had 66 months OS (p = 0.751; Figure 3A ). Further, KRAS ^{MUT not G12C} patients had survival of 70 months and KRAS ^{MUT G12C} had 61 months (p = 0.344; Figure 3B ). In the T2 group, KRAS ^WT had 53 months while KRAS ^MUT had 59 months OS (p = 0.495; Figure 3C ). KRAS ^{MUT not G12C} patients had survival of 51 months and KRAS ^{MUT G12C} had 66 months (p = 0.389; Figure 3D ). Similarly, in the T3 group, KRAS ^WT had 47 months while KRAS ^MUT had 50 months OS (p = 0.966; Figure 3E ). KRAS ^{MUT not G12C} patients had survival of 50 months and KRAS ^{MUT G12C} had 53 months (p = 0.984; Figure 3F ).

We further analyzed the impact of T stage on survival and found that it correlated as expected with mean OS of 82 months for T1, 55 months for T2, and 46 months for T3 (p < 0.001; Supplementary Figure 3A ). The same trend was observed when separately analyzing KRAS ^WT with a mean OS (median not reached) of 83 months for T1, 53 months for T2, and 45 months for T2 (p < 0.001; Supplementary Figure 3B ), and KRAS ^MUT with a mean OS of 65 months for T1, 58 months for T2, and 48 months for T3 (p < 0.023; Supplementary Figure 3C ).

To evaluate differences between primary tumor size from CT scans at diagnosis stratified by KRAS mutational status, we used the Mann-Whitney U test. The test revealed that KRAS ^MUT primary tumors were significantly smaller at diagnosis, with a median size of 20 mm (n = 115) vs KRAS ^WT primary tumors with a median size of 25 mm (n = 190) (p = 0.043; Figure 4A ). However, when looking at tumor size as assessed in resected specimens, there were no differences; KRAS ^WT median size 22 mm (n = 171) vs KRAS ^MUT median size 21 mm (n = 102) (p = 0.16; Figure 4B ).

We found that increase in primary tumor size determined from CT scans did not have a significant effect on risk of death (HR, 1.006; 95% CI, 0.922–1.021; p > 0.5) ( Figure 4C ). However, when testing the correlation between primary tumor size as assessed in resection specimens, we found a significantly increased risk of death (HR, 1.029; 95% CI, 1.012–1.046; p < 0.001) ( Figure 4D ). The risk of death increases with 2.9% for every mm increase of size.

When analyzing stage I and stage II patients separately we found that the primary tumor size, as assessed in resection specimens, correlated to a significantly increased risk of death (HR, 1.051; 95% CI, 1.026–1.077; p < 0.001) ( Supplementary Table 1 ) for stage I patients, with 5.1% for every mm increase in tumor size. However, for stage II patients, no correlation was found between tumor size and risk of death. Furthermore, the primary tumor size determined by CT did not impact the risk of death in either stage. Along these lines, we analyzed the risk of death separately for T1, T2 and T3 groups regarding tumor size from CT and resected specimens and no correlation was found ( Supplementary Table 1 ).

To test if the combination of tumor size and KRAS mutational status impacts the risk of death, we defined an interaction term including both variables. For primary tumor size from CT scans and KRAS mutational status, no significant difference in the risk of death was detected (HR, 1.008; 95% CI, 0.988–1.030; p > 0.5) ( Figure 4C ). Similarly, there were no significant differences for primary tumor size and KRAS mutational status when measured in resection specimens (HR, 1.002; 95% CI, 0.978–1.027; p = 0.807) ( Figure 4D ). Along these lines, we analyzed the risk of death separately for stage I and II, as well as for T1, T2 and T3 groups regarding the interaction term combining tumor size and KRAS mutational status and no correlation was found ( Supplementary Table 1 ).