KRAS G12C mutated advanced non-small cell lung cancer (NSCLC) Characteristics, treatment patterns and overall survival from a Danish nationwide observational register study

Objectives: We aimed to characterize the advanced NSCLC population in terms of KRAS G12C prevalence, patient characteristics, and survival outcomes after the introduction of immunotherapies


Introduction
Lung cancer is the leading cause of cancer mortality and the most common cancer worldwide with non-small cell lung cancer (NSCLC) being the most frequent subtype [1,2].At the time of diagnosis, most patients have progressed to advanced disease and are left with few available treatment options.The prognosis for NSCLC remains poor with an estimated five-year overall survival (OS) of approximately 26% in the US [3] despite earlier detection of NSCLC, improved radiation techniques, and advances in pharmacologic treatments including targeted therapies for anaplastic lymphoma kinase [ALK] and epidermal growth factor receptor [EGFR]) and inhibitors for other genetic alterations such as ROS1, BRAF, NTRK, RET, MET Ex 14 as well as immunotherapies (anti-PD-1/anti-PD-L1 treatment).
Mutations in Rat Sarcoma (RAS) are the most frequent oncogenic drivers in human cancer dominated by mutations in Kirsten RAS (KRAS) occurring in 20-40% of all lung cancers [4][5][6].Oncogenic driver mutations are changes in the DNA sequence of genes that cause cells to become malignant.In NSCLC, the KRAS G12C mutation is one of the most common, accounting for approximately 13% of all mutations in Western populations [7][8][9][10].
The advances in the understanding of tumor biology during the past years have led to the development of specific covalent KRAS G12C inhibitors such as sotorasib and adagrasib.Based on the results of the single-arm, multicenter, phase I/II CodeBreak 100 trial (NCT03600883), sotorasib received accelerated approval by the U.S. Food and Drug Administration (FDA) in May 2021 [11].In addition, a positive opinion by the European Medicines Agency (EMA) was given in January 2022 for the treatment of adults with advanced NSCLC harboring the KRAS G12C mutation who have progressed after at least one prior line of systemic therapy [12,13].The efficacy of sotorasib was tested in the CodeBreaK 200, open-label, phase 3 trial (NCT04303780) where patients with locally advanced or metastatic KRAS G12C mutated NSCLC who had previously been treated with a platinum-doublet chemotherapy and checkpoint inhibitor were randomized to receive sotorasib or docetaxel.CodeBreaK 200 found an improvement in the primary endpoint of progression-free survival (PFS), but overall survival (OS), a secondary endpoint, was not significantly different between treatment arms [14].
Prior to the availability of targeted therapy for KRAS G12C NSCLC patients, treatment options were limited to chemotherapy until the introduction of anti PD-1/L1 treatment (according to PD-L1 expression levels) in 2015 (Supplementary Fig. S1).Patient and tumor characteristics, treatment patterns, and clinical outcomes of KRAS G12C NSCLC patients in Denmark have not been well-characterized following the introduction of anti PD-1/L1 treatment.However, there is a need to understand the treatment-eligible real-world advanced NSCLC population before the introduction of KRAS G12C targeted therapy.Furthermore, it is uncertain whether we can extrapolate the outcomes found in the clinical trials to routine clinical care, and we therefore need to bridge this evidentiary gap by comparing the outcomes from the clinical trials to the current standard of care in real world populations.Using the Danish national registries, we characterized the advanced NSCLC population after the introduction of immunotherapies in terms of KRAS G12C prevalence, patient characteristics and survival outcomes.

Study design and participants
A Danish nationwide register-based cohort study spanning from January 1, 2018 to December 10, 2021 was conducted.Patients eligible for study inclusion were those aged ≥18 years with an incident locally advanced or metastatic (stage IIIb/c or IV) NSCLC diagnosed between January 1, 2018 to June 30, 2021.Patients were excluded if they immigrated to Denmark less than one year prior to diagnosis or were diagnosed with early stage (I-IIIa) NSCLC regardless of progression to stage IIIb/c or IV during the study period.We defined four groups based on mutational status in order to compare the 1) KRAS G12C mutated patients to 2) all NSCLC patients, 3) patients with any KRAS mutation, including KRAS G12C (KRAS mutation unspecified subtype; KRAS gene deletion, unspecified subtype; KRAS gene amplification, unspecified subtype; KRAS mutation in codon 12, 13, 61, 117 and 146, unspecified subtype; KRAS G12A; KRAS G12C; KRAS G12D; KRAS G12R; KRAS G12S and KRAS G12V; and KRAS G13D), or 4) patients who were KRAS/ EGFR/ALK WT (Triple WT) (Fig. 1).KRAS, EGFR and ALK status were extracted from the Danish Pathology Register.The mutational testing was a mixture of Next-Generation-Sequencing (NGS) testing and PCR testing (Supplementary Table S1).PD-L1 expression status was also extracted from the Danish Pathology Register.The PD-L1 expression as assessed by immunohistochemistry was reported as PD-L1 tumor proportion scores (TPS) for all patients.During the study period, the analyses were based on the antibody 22C3, either as Dako-Agilent's PD-L1 IHC 22C3 pharmDx or as a validated 22C3-test on Dako-Agilent Omni's machine [15].

Data sources
Patients with NSCLC diagnoses were identified in the Danish Lung Cancer Register (DLCR), which contains data on all Danish lung cancer patients since 2003 [16].Sociodemographic information (gender, age, region of treatment, smoking number of pack-years, level of income and education), genetic mutation profile (ALK, BRAF, EGFR, ERBB2, KRAS, MET, NTRK1/2/3, RET, ROS, STK11 and TP53), clinical characteristics (comorbidities [defined by diagnosis codes assigned at hospital contacts within the previous 5 years of NSCLC diagnosis], ECOG performance status, PD-L1 expression status at NSCLC diagnosis), tumor characteristics (NSCLC diagnosis, date, histology, and stage of disease), treatment history (line of treatment (LOT) and type of systemic therapy) and death were gathered from the following registers: the Danish National Patient Register [17], the Danish Pathology Register (DPR) [18], the Danish Civil Registration System [19], the Danish Education Register [20], and the Income Statistics Register [21] (Supplementary Table S2).The registers were linked using the unique CPR-number (Civil Personal Registration number) which is an identification number assigned to all Danish residents at birth or immigration.

KRAS mutation status
All advanced NSCLC patients included in the analyses were described overall (All NSCLC); and patients with known KRAS mutation status before start of first line systemic treatment (LOT1) were classified into the following three groups (Fig. 1): KRAS G12C, Any KRAS mutation (G12C and non-G12C), and Triple WT [ALK, EGFR and KRAS WT]).After the implementation of NGS-testing in the Capital Region and the Central Jutland Region, KRAS mutational testing was carried out systematically from January 1, 2019 and April 1, 2018, respectively, but the WT results were not systematically reported into the DPR.If a patient had a biopsy before LOT1 with a test result for ALK/EGFR after the abovementioned dates, a missing KRAS status for patients in the two regions, was categorized as KRAS WT.No further categorization was performed in the remaining three regions (Region Zealand, Region of Southern Denmark, and Region of Northern Jutland) as KRAS testing was not performed systematically in these regions during the study period.In Region Zealand, NGS-testing was implemented (non-systematic testing) from January 14, 2019 with a panel from QIAGEN (AIT Basic FFPE v6.0: QIAact Actionable Insights Tumor Panel on FFPE).In addition to pathogenic alterations in the EGFR gene exon 18-21, this panel inaccurately reported changes in codon 521 as being pathogenic.This was not corrected until August 24, 2020.Consequently, all EGFR mutations reported from Region Zealand between January 14, 2019 and August 24, 2020 was categorized as 'unknown'.
LOT1 was defined as the first administration of an eligible therapy drug plus other eligible drugs given within the first 28 days following the advanced NSCLC diagnosis, where start date of LOT1 was defined as the first administration of the last type of therapy that defined the treatment grouping.
The treatment groups were defined by treatments administered within the first 28 days of a line of therapy (LOT) (i.e., those that began on day 1, with additional treatments on day 14, day 20, etc. were considered one treatment line if the overlap between individual treatments was less than 28 days).
As end dates of therapy were not available in the registers, a LOT was considered discontinued at the initiation of a subsequent LOT.If a gap in drug administration of more than 120 days occurred, the LOT was advanced regardless of the type of therapy (i.e., even if the same type of treatment was repeated).Time-to-next-therapy (TTNT) was defined as the interval between the start date of a LOT and start date of subsequent LOT or death, whichever occurred first, as a surrogate measure for PFS.

Analyses
Descriptive statistics were used to characterize KRAS G12C prevalence, patient characteristics and treatment patterns.For continuous variables, medians, means, and standard deviations (SDs) are presented, and for categorical variables, the number and percentage are reported.
The primary outcome was OS from date of mutational test result (the date of the mutational test defining allocation to KRAS G12C, Any KRAS mutation, or triple WT).All NSCLC was not included in the comparison as not all patients in this group had a mutational test done.Secondary outcomes were OS from start of LOT1 and LOT2, TTNT from LOT1 and LOT2.OS and TTNT were calculated as Kaplan Meier (KM) estimates; and the median with 95% confidence intervals (CIs) and numbers at risk are presented in KM curves.Start date of LOT1 and LOT2 were used as index dates for OS and TTNT analyses.Censoring criteria were death, emigration, or end of study period, whichever occurred first.
For outcome measures or groups in a stratum with a number of patients less than five, the numbers were rounded to the nearest five due to the data discretion rules of Statistics Denmark.
Analyses were validated by parallel coding and carried out using SAS Enterprise Guide (7.1) and R version (4.1).

Ethics
According to Danish law, ethical approval from the ethics committee or informed written consent is not needed for anonymized observational registry studies.The study was approved by the Danish Data Protection Agency (ref.no.P-2020-39).

Patient characteristics
Age, median follow-up time, level of income, education and distribution of comorbidities were similar between the groups (Table 1).Regardless of KRAS mutation status, the majority of patients were diagnosed with Stage IV at NSCLC diagnosis and the most frequent histologic type was adenocarcinoma.The KRAS G12C group differed from the other groups in some respects; they were more likely women (67%) and smokers (86%) compared to all NSCLC, any KRAS mutation and Triple WT groups.About one-quarter of KRAS G12C (26%), any KRAS mutation (28%) and 27% of Triple WT had ECOG performance status >1 at diagnosis, but this was slightly higher among the all NSCLC group (31%).KRAS G12C had higher levels of PD-L1 expression ≥50% (54%) than any KRAS mutation (46%), Triple WT (36%) and the all NSCLC (38%) groups.
Compared to patients who were systemically treated, the patients that did not receive systemic anticancer treatment (40% of all NSCLC) were older (median 74 years vs. 69 years), had a lower level of income and education, a higher Charlson Comorbidity Score, and a higher ECOG performance status.

Co-mutation profile
Of all NSCLC, 75% had a record of any mutational testing prior to LOT1 (Table 1).The median time from diagnosis to the test result defining the group (i.e.KRAS G12C, Any KRAS mutation, or Triple WT) was 13 days (IQR: 7-21 days).The test result defining the mutational group was received no later than one and three months after diagnosis in 89% and 98% of patients, respectively.KRAS G12C patients were widely tested for ALK (>98%) and EGFR (90%).Less than five KRAS G12C patients had an EGFR or ALK co-mutation.We investigated co-mutation status for a range of other co-mutations (ERBB2, BRAF, MET, NTRK123, RET, ROS, STK11, and TP53) but these were poorly reported or rarely tested for during the study period (data not shown).The third and fourth most reported mutation tests during the study period were ROS (45% of KRAS G12C) and BRAF (32% of KRAS G12C).In all NSCLC patients, 16% (n = 1,215) were tested for BRAF prior to LOT1, of whom 4.0% (n = 47) had a documented BRAF V600E mutation.For those with a known test result, parallel to EGFR and ALK mutations, ROS and BRAF were also almost mutually exclusive with a KRAS G12C mutation.

Treatment characteristics
In the all NSCLC, any KRAS mutation, and Triple WT groups, the most frequent LOT1 treatment was platinum-based chemotherapy (without anti-PD-1), whereas the most common LOT1 treatment for KRAS G12C patients was anti-PD-1/PD-L1 monotherapy (Fig. 2).

Survival outcomes and TTNT
Median OS was analyzed from the date of mutational test result instead of date of NSCLC diagnosis to account for the immortal time bias that otherwise occurs in the time span from the date of NSCLC diagnosis to mutational testing.The OS from mutational test result was comparable for KRAS G12C, all NSCLC, any KRAS mutation and Triple WT (Fig. 3).Regardless of mutational group, a longer OS was observed for patients with PD-L1 expression ≥50% compared to those with a PD-L1 expression of <1% and 1%-49%.
Patients that did not receive systemic treatment had a shorter median OS of 2.7 months (95% CI 2.5-2.8months) compared to patients who did receive systemic treatment (median OS of 14.0 months; 95% CI 13.4-14.6months).
From LOT1 and LOT2, the KRAS G12C mutated group had a numerically longer median OS than any other mutational group (Fig. 4 and Supplementary Fig. S2).However, when stratifying the mutational groups by PD-L1 expression level, the OS from LOT1 and LOT2 were similar between the groups.Regardless of mutational group, patients with a PD-L1 expression of ≥50% had a longer OS compared to those with a PD-L1 expression of <1% and 1%-49%.Also from LOT2, the KRAS G12C mutated group had longer OS than the other groups.
When evaluating OS in LOT1 according to type of treatment received, the longest OS was observed for patients receiving TKI.For patients receiving anti PD-1/PD-L1 treatment (with or without platinum-based chemotherapy), OS was longer for all groups compared to patients receiving platinum-based chemotherapy without PD-1/PD-L1i (Supplementary Fig. S3).

Discussion
This nationwide study provides real-world evidence on the characteristics and outcomes of advanced NSCLC patients after the introduction of targeted therapies and immunotherapies in routine clinical practice in Denmark, with a specific focus on patients with the KRAS G12C mutation.Since there were no approved targeted therapies for NSCLC patients harboring KRAS G12C until recently, these patients were treated as patients with unknown driver mutations.The Triple WT group was included to compare outcomes with patients for whom there are no or few approved targeted therapies.
We found that 40% of patients were KRAS tested prior to LOT1 during the study period.Of all patients with a known KRAS test result before LOT1, 11% harbored the KRAS G12C mutation.Demographic and clinical characteristics differed in the KRAS G12C group compared to the other groups.KRAS G12C mutated patients were more likely women, smokers, and had a lower level of education and income.
The higher proportion of females in the KRAS G12C group is in concordance with some recent studies [22][23][24][25][26], but not all identified studies [26][27][28][29][30].It is well established that KRAS G12C is associated with smoking and that non-squamous cell carcinoma is the most frequent histological subtype [30].The low proportion of squamous cell carcinoma in KRAS G12C is further explained by NGS-testing occurring almost exclusively in patients with adenocarcinoma.Consequently, it is plausible that the prevalence of KRAS in squamous cell carcinoma is under-estimated in this study.
Of the patients who were tested prior to LOT1, we found KRAS G12C to be mutually exclusive with other driver mutations.This is in

Table 1
Patient and tumor characteristics at NSCLC diagnosis.Due to data discretion rules of Statistics Denmark, numbers <5 patients are not reported or rounded to nearest 5. Abbreviations: ALK: anaplastic lymphoma kinase, BRAF: serine/threonine-protein kinase B-Raf, ECOG: Eastern Cooperative Oncology Group, EGFR: epidermal growth factor receptor, KRAS: Kirsten rat sarcoma virus, LOT1: first line of therapy, n: numbers, NOS: Not otherwise specified, NSCLC: Non-small Cell Lung Cancer, PD-1/PD-L1: programmed cell death-1/programmed death-ligand-1 and SD: standard deviation, TPS: Tumor proportion score.accordance with the current understanding that KRAS G12C is mutually exclusive with other actionable driver mutations, and only occurs as a resistance mechanism after anti-EGFR therapy [22,27,31,32].
As a consequence of the high proportion of patients with PD-L1 expression level ≥50% in the KRAS G12C group, this group received anti PD-1/PD-L1 treatment more frequently compared to the other groups.The most common LOT1 therapies for KRAS G12C mutated patients were anti-PD-1/PD-L1 monotherapy (46%), platinum-based chemotherapy without anti-PD-1/PD-L1 treatment (39%), and platinum-based chemotherapy with anti-PD-1/PD-L1 treatment (13%).Thus, use of any anti-PD-1/L1 treatment in LOT1 was 59%.Combination treatment with platinum-based chemotherapy with anti-PD-1/PD-L1 treatment was not recommended by The Danish Medicines Council before April 2019 and October 2020 in Denmark for non-squamous NSCLC patients with PD-L1 expression level 1-50% and <1%, respectively (Supplementary Fig. S1).Consequently, we found an increase in the proportion of patients receiving anti-PD-1/PD-L1 treatment during the study period (data not shown).
Previous studies have shown diverging results on the impact of the KRAS G12C mutation on outcomes and survival in NSCLC  [ 22,26,[34][35][36][37][38][39].In this study, the median OS was similar for KRAS G12C, any KRAS mutation and Triple WT from the time of mutational test result, but from the start of LOT1 and LOT2, the median OS was longer for KRAS G12C mutated patients.When analyzing OS by PD-L1 expression levels, the longer OS from LOT1 and LOT2 for KRAS G12C disappeared indicating that the high level of PD-L1 expression in KRAS G12C mutated patients was likely accountable for the difference in OS between the mutational groups.In KRAS G12C patients, those with a PD-L1 expression of ≥50% had a longer OS from LOT1 compared to KRAS G12C patients with a PD-L1 expression level of <1% and 1%-49%.Another observational study also found a positive impact of high PD-L1 expression (≥50%) on survival [26].In the phase 1/2 CodeBreak100 trial evaluating KRAS G12C mutated advanced NSCLC patients, a lower percentage of patients with a high level of PD-L1 expression (>50%) experienced objective response to sotorasib than patients with low (<1%) or moderate (1-49%) PD-L1 expression level [40].Since PD-L1 positive KRAS G12C patients seem to benefit from anti-PD-1/PD-L1 treatment, it is plausible that anti PD-1/PD-L1 treatment with anti KRAS G12C treatment could be a promising combination therapy for this subgroup of patients.However, the first report of safety/efficacy of sotorasib in combination with anti PD-1/PD-L1 treatment in advanced KRAS G12C mutated NSCLC presented at World Conference on Lung Cancer 2022, showed that this combination may be challenged by hepatotoxicity [41].
It is a major strength that this study included both systemically treated and untreated patients as the untreated patient group constitutes a large percentage (approximately 40% in each group) of the advanced NSCLC population.Furthermore, this study is one of the largest population-based investigations of PD-L1 expression in patients with NSCLC treated with current standard of care.This nationwide study design reduces selection bias and limitations of missing data as the Danish administrative and health registries allow complete nationwide capture of an unselected cohort of all individuals diagnosed with NSCLC with independent individual-level ascertainment of all previous hospital Fig. 3. Overall survival (OS) from the date of mutational test result.To account for immortal time bias, rather than including patients from NSCLC diagnosis, patients were included in the analysis from the date of the mutational test result defining the mutational group.Only the mutational groups requiring a KRAS mutation test (and for KRAS WT also EGFR and ALK tests) were included in this analysis, hence the all NSCLC group is not reported.The Kaplan-Meier curves show A) Overall OS for each group from LOT1, B) For patients in each group with PD-L1 expression <1%, C) PD-L1 expression 1-49% and D) PD-L1 expression ≥50%.
contacts [17].Having access to smoking status, and socio-economic deprivation, that are likely to influence the outcomes, is a strength as well.Moreover, cohort studies are often challenged by immortal time bias.However, in our study, we accounted for immortal time bias by analyzing OS from mutational test result defining the mutational group rather than analyzing OS from the date of NSCLC diagnosis.The analyses from the start of LOT1 and LOT2 (both OS and TTNT analyses) only included patients tested prior to the respective dates, and therefore immortal time bias was not an issue for these analyses.Thus, we do not consider immortal time bias to have an impact on our results.
This study has some limitations.Firstly, the prevalence of KRAS G12C found in our study may not reflect the complete prevalence in Denmark as not all advanced NSCLC patients were tested in the study period, and the subtype of KRAS mutation was not specified for all KRAS mutated patients.Hence, in the any KRAS mutation group, there is likely an underestimation of KRAS G12C patients.However, the prevalence of KRAS G12C (11%) found in our study is in concordance with the prevalence found in other Western populations (10.5%-15.4%)[22,23,26].
In the any KRAS mutation group, we chose to include the few (<5) patients that were registered with KRAS gene amplification as they had an altered KRAS gene expression.However, it is unknown whether these patients were KRAS mutated.Due to the small number of registered patients with a KRAS gene amplification, we do not consider this decision to have any significant impact on the results.
Furthermore, there were regional differences in testing uptake with the Region of Southern Denmark and the Region of Northern Jutland testing less frequently than the other regions.In addition to these two regions, Region Zealand did not perform KRAS mutational testing systematically.Hence, it is likely that some patient selection occurred due to the non-systematic KRAS testing in these regions.
A small number of patients may have received systemic anti-cancer treatment without it being included in the treatment analyses.The list of different types of anti-cancer systemic therapies included in this study is very inclusive, but there are a few exceptions where the systemic anticancer treatments are not included: 1) Anti-cancer systemic treatments that are not registered and unavailable from the registries, such as clinical trial drugs (including anti KRAS G12C treatment which is not yet recommended by the Danish Medicines Council), and other existing drugs that do not have an SKS procedure (treatment) code in the Danish National Patient Registry (DNPR).These occur at very low frequencies but are indeed a possibility also in the "No systemic treatment" group; 2) Other cytotoxic treatments not included in Table S3 administered presumably for other cancer types.Co-cancers may occur, and these treatments are not considered in the present analysis; and 3) Systemic treatments of unspecified type in the registries.On a negligible number of occasions, treatments are registered using unspecific codes, and when these occur alone it is not possible to determine the specific nature of the administered treatment.However, these are very rare and would not impact substantially on any of the analyses.
Moreover, the study is limited to advanced NSCLC patients, and only those who were KRAS tested prior to the start of LOT1 were eligible for the subgroup analyses.Thus, the results may not be generalizable to all NSCLC patients.TTNT (a surrogate measure of the clinical outcome measure 'Progression Free Survival' [PFS]) included in our analyses may not be comparable to PFS determined in clinical trials and should be interpreted with caution.Our data did not allow for a direct analysis of progression (e.g.CT-scan confirmation), and we used change of therapy or break of therapy >120 days (whichever occurred first) as an indication of progression.Also, there may be reasons for change of therapy (e. g. toxicity) other than progression.

Conclusion
In patients diagnosed with advanced NSCLC after the implementation of anti PD-1/L1 therapies as part of routine standard of care, the survival from LOT1 and LOT2 in KRAS G12C mutated patients is comparable to patients with any KRAS mutation, Triple WT, and all NSCLC patients.

Fig. 1 .
Fig. 1.Patient selection flow of all patients diagnosed with NSCLC between January 1, 2018 and June 30, 2021 identified in the Danish Lung Cancer Register.Numbers may be rounded to nearest 5 if belonging to a cluster in which a subgroup is n < 5, in order to comply with the discretion rules of Statistics Denmark.Abbreviations: ALK: anaplastic lymphoma kinase, EGFR: epidermal growth factor receptor, KRAS G12C: Kirsten rat sarcoma virus p.G12C KRAS: Kirsten rat sarcoma virus, LOT: line of therapy (e.g.LOT1 = first line of therapy), NSCLC: Non-Small Cell Lung Cancer, Triple WT: ALK/EGFR/KRAS wildtype, WT: wildtype.Percentages for the subgroups by mutational status (Any KRAS mutation, KRAS G12C and Triple WT) are percentages out of all patients with a KRAS test result before LOT1 (n = 2,969).Percentages for patients in LOT1-LOT4 are percentages out of the specific subgroup.'No treatment' only refers to systemic treatment, i.e. patients may have received radiation therapy or surgery.

Fig. 2 .
Fig. 2. Sunburst Chart displaying the distribution and sequence of therapeutic regimens for each of the groups.Each color represents a unique regimen or death, each ring a line of therapy (starting with LOT1 in the inner ring).Due to data discretion rules, in cases where <5 patients were in a treatment sequence e.g.treatment X in LOT1 followed by treatment Y in LOT2, treatment Y are not shown in LOT2.Abbreviations: KRAS: Kirsten rat sarcoma virus, Mono: monotherapy, NSCLC: Non-small Cell Lung Cancer, PD-1/PD-L1: programmed cell death-1/programmed death-ligand-1, TKI: tyrosine kinase inhibitor, w/o: without, WT: wildtype.