2.2. Results of NGS Analysis

The results of NGS analysis are summarized in Table 2.

In this study cohort, NGS was performed in 55% and 45% at diagnosis and later stage, respectively. Patients in cohort A underwent NGS analyses more frequently at a later stage (59%) than those in cohort B (NGS at diagnosis, 62%). The Oncomine Comprehensive Assay Version 3 (OCAv3) was the most frequently used NSG panel (87% of the study cohort) and covered 161 of the most relevant cancer driver genes. A total of 109 different mutations were identified overall (71 and 94 in cohorts A and B, respectively). The five most commonly mutated genes were tp53, KRAS, SETD2, SLX4, and ATR (Figure 2 and Figure 3).

Open in a separate window

Figure 2

Frequency and name of all mutations.

Open in a separate window

Figure 3

The mutations that were detected were stratified according to the diagnosis.

Overall, 38%, 9%, and 1% of the patients had 1, 2, and 3 pathogenic mutations, respectively (Figure 4). In cohort A, 12 (35%) and 4 (12%) patients had 1 and 2 pathogenic mutations, respectively. In cohort B, 26 (39%), 5 (8%), and 1 (1.5%) patient(s) had 1, 2, and 3 pathogenic mutations, respectively.

Open in a separate window

Figure 4

Heatmap of all pathogenic and not-pathogenic mutations. 0 = no pathogenic mutation; 1 = pathogenic mutation.

The most frequent pathogenic mutation (Figure 5) overall was KRAS (27%), followed by BRAF other than V600E and PIK3CA (8% each), and ERBB2, ESR1, and NRAS (7% each). The most frequent pathogenic mutations in cohort A were PIK3CA (25%), KRAS (20%), BRAF V600E (10%), whereas in cohort B, these included KRAS (31%), BRAF other than V600E, ESR1 and NRAS (each 10%), and ERBB2 (8%).

Open in a separate window

Figure 5

Pathogenic mutations. (a) Frequency of pathogenic mutations in our collective according to diagnosis. (b) Frequencies of pathogenic mutations in cohorts A (left) and B (right).

Overall, 27% of our patients had targetable mutations (22%, 4%, and 1% of the patients had 1, 2, and 3 targetable mutations, respectively). In cohort A, 8 (23.5%) and 3 (9%) patients had 1 and 2 targetable mutations, respectively. In cohort B, 14 (21%), 1 (1.5%), and 1 (1.5%) patient(s) had 1, 2, and 3 targetable mutations, respectively (Figure 6).

Open in a separate window

Figure 6

Targetable mutations.

The most frequent targetable mutations were PIK3CA (15%), ESR1 and NRAS (each 12%) overall; PIK3CA (36%) and BRAF V600E (14%) in cohort A; and ESR1 and NRAS (each 21%) and MET (11%) in cohort B (Figure 7).

Open in a separate window

Figure 7

Targetable mutations stratified by the diagnosis. (a) Frequency of targetable mutations in the study cohort. (b) Frequency of targetable mutations in cohorts A (left) and B (right).

2.3. Discussion of Results at Molecular Tumor Board/Tumor Board

The details of most of the patients were discussed with the institutional tumor board. In 2020, only a minority of patients were discussed with the molecular tumor board of the university hospital in Basel.

3.1. Objectives

1. To compare the clinicopathologic and NGS data in the two cohorts (A and B)

2. To compare the rate of presentation to the molecular tumor board and discussion with experts

3. To compare the outcomes of patients in cohorts A and B based on the successful allocation of NGS-data-based precision therapy (time of treatment)

3.2. Endpoints

1. Clinical characteristics in descriptive statistics in cohorts A and B

2. Rate of molecular tumor board presentation (optimal 1 NGS = 1 tumor board presentation in cohorts A and B)

3. Time on treatment with the targeted agent after the NGS-based treatment allocation

3.3. Patients Disposition and Cohort Description

Patient selection was based on a report prepared by the Institute of Pathology of our hospital, in which all patients with tumors who underwent an NGS examination in 2020 were listed in random order. The selection criteria for the patients are presented in Appendix A.

The first 100 consecutive patients were included in this study. All clinicodemographic patient information was collected from the KSBL’s electronic medical records/charts (see Appendix B for detailed information) and transferred to an MS Excel^® file (https://www.office.com/).

In the next step, all information on the NGS analyses was compiled from electronic pathology reports (see Appendix C). All patients who were under 70 years old at the time of NGS analysis were randomized in cohort A; all patients who were 70 years or older at the time of NGS analysis were randomized in cohort B. All mutations and amplifications identified by NGS were recorded, and the predictive/pathogenic mutations according to the pathology report were noted. The latest (data: 24 January 2024) OnkoKB^® [23] database was used to determine whether a target or targeted therapy was available for these predictive/pathogenic mutations. Patient files were then searched to determine whether the NGS data had been discussed at a molecular tumor conference and if the patients for whom the targeted therapy was available had received it. Further information about targeted therapy, such as the name of therapy, time on therapy, best response, progression-free survival (PFS), benefit of therapy (PFS > 6 months), and time of overall survival (OS), was noted. Finally, the date of last follow-up or death was recorded. Patients who received NGS-based treatment were classified according to the current ESMO (European Society for Medical Oncology) ESCAT (ESMO Scale for Clinical Actionability of Molecular Targets) criteria [24].

3.4. Molecular Diagnostics

All NGS analyses were performed at the Institute of Pathology of our hospital using nationally certified tests. Ninety-seven percent of the analyses were performed on FFPE (formalin-fixed paraffin-embedded) tissues. Three analyses were performed on the liquid biopsies. The assays were performed using local standard operating procedures according to the manufacturer’s guidelines.

The following NGS panels were used (see Appendix D for detailed information about the NGS panels and a list of the whole genes that were examined).

• (A)

  Oncomine Comprehensive Assay^®, Version 3 (OCAv3) (ThermoFisher Scientific, Waltham, MA, USA): a multi-biomarker NGS assay that covers 161 of the most relevant cancer genes

• (B)

  Oncomine Tumor Mutation Load Assay^® (TML) (ThermoFisher Scientific, Waltham, MA, USA): a large targeted NGS assay that covers 1.65 Mb across 409 oncogenes relevant across major cancer types

• (C)

  Oncomine Solid Tumor DNA Panel^® (ThermoFisher Scientific, Waltham, MA, USA): next-generation sequencing (NGS) kits for both DNA and RNA workflows that enable the assessment of actionable mutations involved in solid tumors, particularly those originating from the lung and colon, and cover 22 hotspot genes

• (D)

  KSBL Hotspot Panel^® (HOP), (Institute of Pathology of Liestal, Switzerland): The KSBL Hotspot Panel is a custom panel designed by the Institute of Pathology of Liestal, based on the Ion Ampliseq Custom DNA panels (Thermo Fisher Scientific, Waltham, MA, USA). It partially covers 124 genes for hotspot alterations and 34 genes for copy number variations.

• (E)

  Ion Torrent™ Oncomine™ cfDNA Assays (ThermoFischer Scientific, Waltham, MA, USA): tumor type-specific multi-biomarker next-generation sequencing (NGS) assays that enable the detection of somatic mutations in genes found in plasma samples

  1. Oncomine Lung cfDNA Assay^® covers 11 hotspot genes

  2. Oncomine Colon cfDNA Assay^® (cfColon) covers 14 hotspot genes

Data readout and analyses were performed by an accredited molecular pathologist at the Institute of Pathology.

3.5. Statistical Analysis

All data were analyzed by a statistician at our hospital. A comprehensive set of statistical and analytical methods was employed to elucidate the key aspects of the dataset. Using the R programming language (The R Foundation for Statistical Computing, c/o Institute for Statistics and Mathematics, University of Economics and Business Vienna, Austria), various functions and techniques were applied to derive meaningful insights.

The following statistical methods were utilized:

1. Data Exploration and Grouping:

   • ○ The count and group functions in R (The R Foundation for Statistical Computing c/o Institute for Statistics and Mathematics, University of Economics and Business Vienna, Austria) were used to systematically examine and categorize cases based on specific criteria, enabling a detailed exploration of the dataset.

2. Descriptive Statistics:

   • ○ To calculate the median, minimum, and maximum values for selected variables to provide a summary of the central tendency and variability of the data.

3. Statistical Testing:

   • ○ Applied the two-sided Mann–Whitney U test to assess the significance of differences between the median values for two distinct groups. This method provides a robust nonparametric approach for identifying potential contrasts in central tendencies.

4. Survival Analysis: (survival package)

   • ○ Implemented the surv() function in R to create survival objects that define survival times and event indicators for each participant.
   • ○ The survfit () function was used to compute the Kaplan–Meier survival curves, which visually represent the probability of survival over time.

5. Cox Proportional Hazards Model: (survival package)

   • ○ The coxph () function for the Cox proportional hazards model was used to calculate the hazard ratios and their associated confidence intervals.
   • ○ Generated a forest plot using the forestplot package to visualize the results.

6. Plotting:

   • ○ Plots were generated using the ggplot2 package in R (https://ggplot2.tidyverse.org/), which provides clear and visually informative representations of various analyses, including descriptive statistics, survival curves, and forest plots.

4.1. Clinical Differences of the Cohorts and Differences in NGS

Despite random patient selection, our cohort was not well-balanced, with one-third patients in cohort A and two-thirds in cohort B. This contradicts our hypothesis that NGS is used more frequently in younger patients with cancer. However, this reflects that cancer is a disease of older patients, and NGS is currently a standard diagnostic tool that is used in many different tumor entities, for example, lung cancer, colorectal cancer, prostate cancer, and breast cancer, depending on the approval status of the targeted agents [25].

The number of comorbidities increased with age, and the difference between the two cohorts was statistically significant. This finding aligns with the current literature and should be considered when performing NGS testing [26]. Benderra et al. demonstrated that, in their ELCAPA cohort of more than 1500 older patients with cancer, the comorbidity type, number, and severity were independently associated with a shorter OS. However, the number of comorbidities should not be a barrier to testing older patients with cancer. This is because NGS-driven personalized treatment approaches might increase the tolerability and quality of life of older patients with cancer [27].

Considering the results of the ECOF PS, one can assume that our patient population was quite fit at the time of NGS analysis, with 80% having a documented ECOG PS between 0 and 1 and 20% with an ECOG PS of 2–3. No patient had an ECOG PS of 4, indicating that patients who were completely disabled (ECOG PS 4) were not the target audience for NGS analysis. Other studies have shown that NGS is not valuable when performed in cases with a poor performance status [28]. In general, it is important to question whether the ECOG PS is the correct screening tool for older patients to determine a therapeutic strategy. The ECOC PS describes only the patient’s level of functioning in terms of their ability to care for themselves, daily activity, and physical ability and does not account for the heterogeneity among older adults with respect to their biological age [29,30]. There are special geriatric assessment tools that refer to the evaluation of geriatric impairments that are not routinely captured in oncology assessments by evaluating the functional, cognitive, psychological, and nutritional status, physical performance, history of falls, comorbid medical conditions, and social support, which helps to estimate the biological age [31]. As we aimed to assess the data of the G8 questionnaire (a screening tool that was developed to identify older patients with cancer who would benefit from a former comprehensive geriatric assessment) and correlate with our findings [32]. However, systematic data collection has only been implemented at our cancer center since 2024.

Thus, the clinical management of older patients with cancer should follow the rules of the current guidelines and include a full geriatric assessment if the results of screening procedures, such as the G8 Score, are positive [33,34]. Treatment decisions based on NGS testing might be better facilitated by a more personalized geriatric oncology approach. A personalized geriatric oncology approach for older cancer patients can enhance treatment decisions based on NGS testing by tailoring interventions to their unique health profiles. This includes using comprehensive geriatric assessments (CGA) to adjust treatment intensity, prioritizing targeted therapies with fewer side effects, setting individualized treatment goals focused on quality of life, and applying pharmacogenomic data to reduce toxicity and manage polypharmacy. These strategies help clinicians balance effective cancer treatment with the complex needs of older adults, improving both outcomes and quality of life.

Our patient cohort comprised patients with a heterogeneous group of cancers. We observed a relatively high proportion of patients with newly diagnosed advanced NSCLC, colorectal cancer, and melanoma in cohort B, in which NGS was the standard of care at primary diagnosis [35,36,37]. This suggests that lung cancer is the most common newly diagnosed cancer in patients older than 70 years, followed by colorectal, prostate, and breast cancer [38]. In our study, we observed a relatively low rate of prostate cancer.

It is possible that the more frequent occurrence of one tumor entity may have influenced the results for other tumor entities. However, the most frequently diagnosed tumor entities are also frequently represented in our study population, which also corresponds to the real-world setting.

Patients in cohort A were more frequently diagnosed with localized cancer (65%) than those in cohort B, of whom 50% had metastatic cancer. This indicates that older patients with cancer are often diagnosed at later stages. However, we estimated that our cohort was too small to draw any conclusions on this topic.

4.2. Number of Treatment Lines in Metastatic Cancers

Patients in cohort B received a significantly lower total number of treatment lines than those from cohort A (median number of 6 treatment lines in cohort A vs. 2 in cohort B). Based on current literature, there is a risk of undertreatment in older patients. Older patients are more likely to be frail; therefore, treatment decisions may be more conservative [39,40,41]. A large proportion of older patients also have rapidly progressive disease, refuse therapy, and die rapidly. Only a minority of patients received more than three lines prior to NGS testing (three patients each from cohorts A and B; total 6%).

4.3. Results of NGS Analysis

NGS was performed in 55% at diagnosis and in 45% at a later timepoint. In cohort B, NGS analysis was more common at diagnosis than in cohort A (62% vs. 41%, p = 0.0747). Based on our selected NGS panels, common cancer mutations were detected in both cohorts, including tp53 (40%), KRAS (24%), SETD2 (12%), SLX4 (12%), and ATR (11%) (Figure 2 and Figure 3). TP53 was the most commonly mutated gene, and the number of KRAS mutations in our study was higher than that reported in other studies [42].

The proportion of patients with pathogenic mutations was similar in the two cohorts (47% in cohort A and 48.5% in cohort B). The proportion of patients with targetable mutations was numerically higher in cohort A than in cohort B (32.5% vs. 24%); this difference was not statistically significant (p = 0.476). This demonstrates that older patients with cancer might have a different biology of their disease; this has been shown in several publications [43]. This result limits the treatment options with targeted therapy in older patients. However, the cohort is too small and heterogeneous to draw general conclusions from it.

4.4. Outcome After NGS

In total (both cohorts), 27% of patients had targetable mutations, and 8% of our patients received targeted therapy. In cohort A, 5/11 (45%) patients with a targetable mutation (15% of the whole cohort A) received targeted therapy, whereas in cohort B, 3/16 (19%) patients with a targetable mutation (4.5% of the whole cohort B) received targeted therapy. Although the difference between 45% treatment initiation in cohort A versus 19% in cohort B was not statistically significant, these figures show that targeted therapy was started less frequently in older cancer patients. It is possible that the low number of cases in both cohorts influenced the statistical significance. However, we think that the clinical importance of this result is relevant for older patients because they were less likely to be treated with targeted therapy tailored to their tumor mutation.

But compared to other cohorts, this was in general a comparable outcome for our patients. A study from 2020, which analyzed the impact of NGS on clinical practice in oncology in France, analyzed NGS data from 2013 to 2016 and found actionable alterations in 53% of which 8% received targeted therapy finally [44]. They reported also from studies with sample sizes ranging from 97 to 17 664 patients, with the percentage of patients with actionable alterations ranging from 40% to 94% and of patients receiving matched treatment ranging from 4% to 44% [45]. In the analysis from Gibbs et al. (2023), who performed a comprehensive literature review on the clinical impact of NGS for the management of advanced cancer, they found a mean of 29% and a median of 25% (range 2–66%) of patients that were matched to targeted treatment after NGS testing [46].

Some reasons why less than 50% of the patients with targetable mutations in cohort A (5/11 patients, 45%) and only 19% from cohort B (3/16) started targeted therapy include rapidly progressive disease and stable disease under prior therapy for both cohorts. In cohort B, also age-specific factors such as reduced general condition, no therapy request, and comorbidities, as well as status after targeted therapy, were reasons to not start a targeted therapy. Similar reasons are mentioned in other studies [45,46,47].

Of the 8% who started targeted therapy, 3 patients (38%) or 3% of the whole cohort experienced a clinical benefit from the targeted therapy. One patient from cohort A with NSLCL treated with lorlatinib was treated for at least 168 weeks, with ongoing therapy and stable disease. Two patients from cohort B had partial remission; one patient with GIST was treated with imatinib for 104 weeks, and the other patient with breast cancer who was treated with fulvestrant for at least 136 weeks had ongoing therapy. For those three patients, targeted therapy changed the way of treatment for the last years, and according to the medical history, they are all in a good clinical condition due to the precision medicine.

The reasons for the other five patients without clinical benefits were progressive disease in four patients and toxicity-related discontinuation of therapy in one patient.

We did not find a difference in the clinical outcomes after targeted therapy between cohorts A and B because of the very small number of patients. However, we highlight that a small but relevant portion of older patients benefited from NGS-determined therapies. In addition, despite the lower uptake of targeted therapies among older patients, the positive outcome for those who did receive them suggests a need to reconsider current clinical approaches to reduce the risk of undertreatment of older cancer patients.

Our result of the benefit of targeted therapy after NGS testing of 3% for the whole cohort is consistent with published data. Marquart et al. reported in an American observational study that 8.3% of patients received targeted therapy and 4.9% benefited from a genome-targeted treatment [14]. In the MOSCATO Trial, they reported a rate of 19% of patients matched to targeted therapy and a clinical benefit of 33% of these patients, or 7% of the whole cohort [15]. Another study, part of the TAPUR (Targeted Agent and Profiling Utilization Registry) and NCI-MATCH (Molecular Analysis for Therapy Choice) trials, analyzed the impact of NGS on therapy decisions in patients with metastatic cancer. The study found that NGS testing led to treatment changes in approximately 16% of cases, with a notable proportion of these patients experiencing significant clinical benefits from targeted therapies identified through NGS. The study highlighted that while NGS can reveal actionable mutations, the benefit varies, with some patients showing exceptional responses to therapies tailored to their genetic profile [16,17]. We could not find any other trails, that examined the effectiveness of NGS results in relation to patient age.

Limiting treatment success includes late testing, poor prognosis, therapy-refractory disease, and patient death.

Due to the small number of patients and, in some cases, only a few patients with the same tumor diseases, comparative cancer subgroup analyses, i.e., to analyze cancer-specific trends in the utility or success of NGS-based therapies, were not possible. We found targetable mutations in patients with NSCLC, breast cancer, colorectal cancer, melanoma, sarcoma, GIST, and CUP using NGS examination, and targeted therapies were started in patients with NSCLC (2 patients from cohort A), breast cancer (1 patient from cohort A and 2 patients from cohort B), colorectal cancer (1 patient from cohort A), sarcoma (1 patient from cohort A), and GIST (1 patient from cohort B). From cohort A, one patient with NSCLC, and from cohort B, one patient with breast cancer, and one patient with GIST have benefited from the NGS-guided treatment. However, it is important to note that this result is the result of our cohort and cannot be considered as generally valid because many individual factors led to no therapy being started for other cancer subtypes in both cohort A and cohort B. In cohort B, for example, 4 patients with melanoma had targetable mutations, and all patients were refused the start of targeted therapy (due to bad general condition, no therapy request, loss of FU, and therapy not started yet). It could be that in another cohort all patients are started on therapy, which would lead to different results than in our collective. Therefore, larger data collections are necessary to confirm our results.

4.5. Overall Survival

After NGS, the prognosis was generally very poor, with a median OS of less than 12 months in both cohorts. There was no statistically significant difference in OS between cohorts A and B, either after the date of diagnosis or after the date of NGS, with a trend toward better overall survival in younger patients after the date of diagnosis (p = 0.08).

One reason why younger patients did not have better overall survival than older patients could be that cancer death is related to treatment, treatment duration, and number of lines. We observed that our patients from cohort A were treated with a significantly higher number of therapy lines and had a higher cancer burden and more refractory disease than older patients. On the other hand, a non-significant numerical difference was observed in older patients. Older patients with cancer have a higher competing risk to die because of comorbidities and side effects of cancer therapy. Nevertheless, questions remain, and one might hypothesize about the factors influencing overall survival (OS), including decision-making based on NGS results in older cancer patients. Further research is needed to clarify these factors, and more studies are needed to confirm our observations and achieve significant results.

Factors that had a statistically significant negative effect on overall survival were a higher number of comorbidities at diagnosis (>3) and metastatic stage at diagnosis.

The administration of targeted therapy had no statistically significant effect on OS in our cohort. We attribute this to the small number of patients who received targeted therapy. Another large retrospective cohort study conducted at the Institut Curie in France assessed the survival impact of targeted therapies in lung cancer patients over two decades. The study found that the introduction of targeted therapies, guided by NGS, significantly improved overall survival (OS) in patients with non-small cell lung cancer (NSCLC). The median OS increased notably following the adoption of these therapies, highlighting the clinical benefit of NGS-driven treatment strategies in this specific population [18].

4.6. Discussion of Results at Molecular Tumor Board/Tumor Board

Our second endpoint was to evaluate the number of NGS results presented and discussed on the molecular tumor board. At our institution, we joined the molecular tumor board together with the University Hospital of Basel. In 2020, the results of NGS analyses were discussed at the molecular tumor board for only a minority of patients. Most of our patients were discussed with our institutional tumor board, where the decision to perform the NGS analysis was made. In only 14% of our patients (18% of patients in cohort A and 12% of patients in cohort B), the results of the NGS analysis were discussed again by the tumor board. On the one hand, this is due to the fact that we had many patients in whom no predictive mutations were found (53%). However, data from pathology reports are used in routine clinical practice, and common mutations such as EGFR in lung cancer and ESR1 or PIK3CA mutations in breast or colon cancer are not present on the molecular board. Furthermore, in 2020, the majority of patients with molecular alterations in early disease did not receive any molecular-driven treatment and were therefore not regularly presented on the molecular tumor board.

4.7. Limitations

The limitations of our data include the small sample size, single center data, hypothesis generating data, retrospective nature, and heterogeneous tumor entities.

4.8. Potential Biases

One potential bias could be the different tumor stages at the time of NGS. Fifty-nine percent of patients of cohort A had NGS not at the timepoint of diagnosis but later in the course of disease, while 62% of the patients of cohort B had the NGS examination at the timepoint of diagnosis. This maybe has influenced the decision for treatment (although we could not prove this with our data because of the lack of information about the reasons for not starting targeted therapy for many patients from cohort A). Another weakness of our analysis might be the bias in terms of OS/PFS analysis due to the high number of comorbidities in both cohorts, which could have influenced our OS and PFS results. However, due to the random selection of patients, we have reduced biases and present a Swiss real-world cohort.