Abstract
Background/Aim: Targeted therapy is an important and fast developing aspect of modern tumor therapy including therapy of head and neck cancer (HNC). Surgically treated patients often experience significant limitations to their ability to swallow, speak, or mimic expressions. In cases of recurrent tumors or palliative situations, targeted therapies such as immune checkpoint inhibitors (ICI) are frequently employed. This study compared different targeted therapies focusing on survival probability. Patients and Methods: Data from patients with head and neck cancer treated with different therapy regimens from the TriNetX network were analyzed. Two groups were formed: Cohort I received one targeted therapy, whereas patients in cohort II received a different targeted therapy. Cohorts I and II were matched 1:1 with respect to certain confounders. After defining the primary outcome as “death”, a Kaplan–Meier analysis was performed, and the risk ratio (RR), odds ratio (OR), and hazard ratio (HR) were calculated. Results: A total of 18,331 patients with HNC treated with targeted therapy were analyzed. Patients treated with VEGF inhibitors had a significantly longer overall survival than patients treated with c-MET or EGFR inhibitors. Patients treated with PI3K inhibitors showed a significantly reduced survival probability compared to those treated with c-MET, mTOR, and RET inhibitors. Conclusion: EGFR inhibitors are one of the most frequently used targeted therapies in HNC. However, in the present analysis, a survival advantage of patients treated with c-MET inhibitors or VEGF inhibitors was observed compared to those treated with EGFR inhibitors.
Head and neck cancer (HNC) is one of the most common malignancies in the world. The most frequent histological tumor entity of HNCs is the Head and Neck Squamous Cell Carcinoma (HNSCC) (1). Treatment typically involves extensive surgical procedures, including the use of free or pedicled flaps, and adjuvant combined radiochemotherapy if required, in accordance with national guidelines. However, the 5-year overall survival rate for patients with this type of cancer is currently only 50% (2, 3). Surgically treated patients often experience significant limitations to their ability to swallow, speak, or mimic expressions (4). In cases of recurrent tumors or palliative situations, targeted therapies such as immune checkpoint inhibitors (ICIs) are frequently employed. International guidelines recommend the use of targeted therapy as a first- or second-line palliative treatment (5-8). Therefore, it may be useful to compare different targeted therapies in terms of overall survival (OS) in order to offer the palliative patient the therapy with the longest OS when choosing between two targeted therapies.
The first-line targeted therapy for recurrent or metastatic HNSCC is the use of programmed cell death ligand-1 (PD-L1) inhibitors, also known as ICIs such as pembrolizumab or nivolumab (9). PD-1 is expressed on both immune and cancer cells, suppressing T cell activity and preventing the immune system from killing cancer cells (10). The expression of PD-L1 is evaluated through PD-L1 combined positive score (CPS) and tumor proportion score (TPS). The biomarkers indicate the number or percentage of cells that express PD-L1 (9). Higher scores are linked to a better response to ICI therapy (11). Therefore, the current American Society of Clinical Oncology (ASCO) guideline recommends ICI therapy (pembrolizumab) as the first-line treatment for recurrent or metastatic HNSCC with positive scores. It is also recommended for patients with platinum-refractory recurrent or metastatic HNSCC, regardless of CPS status (9). Compared to standard care treatment for recurrent, metastatic, and platinum-refractory HNSCC, nivolumab has demonstrated an improved response rate, longer OS, and lower toxicity in phase III clinical trials (12). Additionally, research has shown that smokers and HPV-negative patients respond better to ICI therapy (13).
An example of an established targeted therapy is the use of monoclonal antibodies against epidermal growth factor receptor (EGFR). EGFR is over-expressed in approximately 90% of HNSCC (14). Mutation of EGFR is associated with poor prognosis due to the resulting rapid tumor proliferation, increased invasiveness, metastasis rate and angiogenesis. The growth of EGFR-expressing tumors can be inhibited either by blocking EGFR binding sites on the extracellular domain of the receptor or by inhibiting intracellular tyrosine kinase activity (15). Cetuximab, the most commonly used EGFR inhibitor, was approved by the FDA for the treatment of HNSCC in the United States of America in 2006 (16). Research has demonstrated that the addition of EGFR inhibitors to chemotherapy for advanced HNSCC patients can increase patient survival rates with relatively good tolerability (17). However, the use of EGFR inhibitors in modern tumor therapy has presented significant challenges due to serious side effects and EGFR mutations that can lead to drug resistance (17). Combination therapies with other targeted therapies are currently being studied in several clinical trials as a means of reducing side effects and preventing the development of resistance (1).
Another aspect of current research is the use of targeted therapy and immunotherapy in the neoadjuvant and adjuvant settings. A small randomized Phase II study has already demonstrated positive effects of neoadjuvant pembrolizumab in terms of tolerance and response (18). Currently, a randomized Phase III trial (KEYNOTE-689) is evaluating the efficacy and safety of pembrolizumab as neoadjuvant and adjuvant therapy (19). Targeted therapy aims to reduce tumor volume, slow tumor progression, and ultimately prolong OS. In some tumor types, neoadjuvant use of targeted therapy has been shown to be more effective than adjuvant use (20-25). However, the neoadjuvant use of targeted therapy in HNSCC is often limited by the occurrence of comorbidities, field cancerization (26), and expected worse outcomes with delayed surgical treatment (27).
To the best of our knowledge, there are currently no valid data on the influence of different targeted therapies on 5-year OS in patients with head and neck cancer using real-word data. The primary aim of this study was to investigate the influence of different targeted therapies on the 5-year OS of patients with head and neck cancer by comparing the current standard of care targeted therapies with the other less established targeted therapies in a real-world setting. Secondary aim of this analysis was to compare the current standard of care targeted therapies with the other less well-established targeted therapies with regard to a possible survival benefit. We used data from the TriNetX Global Health Research Network (TriNetX, Cambridge, MA, USA). Detailed data analyses were performed to improve understanding of how the use of different targeted therapies might change clinical prognoses.
Patients and Methods
Ethics Statement. An exception was granted by the institutional ethics committee because the study was retrospective, and the data used was de-identified. This is because all Healthcare Organizations (HCOs) that contributed data to TriNetX obtained written informed consent from either the patients themselves or their legal guardians. TriNetX complies fully with the Health Insurance Portability and Accountability Act (HIPAA) and holds certification to the ISO 27001:2013 standard. It maintains a robust Information Security Management System (ISMS) to safeguard the healthcare data it accesses. Additionally, any aggregated data within TriNetX comprises solely de-identified information, following the de-identification standard outlined in Section §164.514(a) of the HIPAA Privacy Rule. The de-identification process undergoes formal verification by a qualified expert, as required by Section §164.514(b)(1) of the HIPAA Privacy Rule. This validation eliminates the need for TriNetX’s previous waiver from the Western Institutional Review Board (IRB). The TriNetX network comprises data contributed by participating Healthcare Organizations. Each healthcare organization (HCO) affirms that it has all the necessary rights, consents, approvals, and authority to supply data to TriNetX under a Business Associate Agreement (BAA). Importantly, these HCOs ensure that their identity remains anonymous as a data source, and the data they provide is exclusively used for research purposes. To enhance patient privacy, data shared via the TriNetX Platform is attenuated to prevent identification of the specific HCO that provided patient information.
Real-world data. As per the USA’s FDA definition, real-world data (RWD) in the medical and healthcare field includes information on patients’ health status and healthcare delivery, collected from various sources.
RWD is obtained independently of randomized clinical trials (RCTs), which remain the standard for generating clinical knowledge. Randomization of patients into distinct treatment groups and stringent selection criteria in RCTs can make it difficult to apply their results to real-world clinical practice.
Real-world evidence (RWE), derived from the utilization of Real-World Data, offers a more accurate reflection of the actual clinical environment in which therapeutic interventions are administered. It includes considerations such as patient demographics, comorbidities, adherence to treatment regimens, and concurrent treatments. RWD represents a valuable and comprehensive source of data that extends beyond the confines of traditional epidemiological studies, clinical trials, and laboratory-based experiments. Furthermore, it offers a cost-effective alternative for data collection compared to the latter methods.
When used and analyzed appropriately, RWD has the potential to generate valid and unbiased RWE. This not only saves time and money compared to controlled trials but also improves the efficiency of research and decision-making in medicine and healthcare.
Data acquisition, allocation, and matching. The TriNetX Global Health Research Network provides access to medical records from more than 120 healthcare organizations (HCOs) in over 30 countries in North and South America, Europe, the Middle East, and Africa (EMEA) and Asia-Pacific including over 250,000,000 patients with more than 5 years of clinical history enabling the collection and exchange of longitudinal clinical data (28).
Patients enrolled in this study met the following inclusion criteria: 1) diagnosis of head and neck cancer (International Classification of Diseases (ICD)-10 codes C00-C14, C30-32, C76) and (2) treatment with targeted therapy including EGFR, VEGF, PI3K, mTOR, c-MET, and RET inhibitors sourced from the TriNetX network (Figure 1).
Consort flow diagram. ICD-10: International Classification of Diseases 10, C00-C14, C30-32, C76.
Exclusion criteria for this study included: 1) the absence of targeted therapy, 2) follow-up data of less than five years, and 3) medical records older than 20 years. No other specific selection criteria were used to ensure the best possible use of real-world data. To be included, patients’ medical records had to cover at least five years (1,825 days) of follow-up after visiting the healthcare organization for an inpatient encounter. To improve the efficiency of processing and delivering outcome analytics results, the TriNetX Global Health Research Network excludes patients who met the criteria more than two decades ago. This category only applies to a minor fraction of patients in most cohorts, as TriNetX mainly contains data from patient encounters within the past two decades.
The analyses were conducted on two groups. Cohort I received one targeted therapy, whereas patients in cohort II received a different targeted therapy. This allowed for comparison between each targeted therapy
To exclude confounders, cohorts I and II were matched 1:1 based on age, sex, lymph node metastases, nicotine dependence, alcohol dependence, and body mass index (BMI), as in previous published studies. Additionally, matching via lymph node metastases allows for comparison of patients with and without advanced and metastatic states. Other confounders were not considered in this study. This approach allowed for conditions that were as close to randomized as possible (29, 30).
Data analysis. Descriptive statistics were utilized to present baseline patient characteristics. Categorical variables are expressed as absolute values (n) and relative incidence (%), while patient age is presented as the mean value and standard deviation. A multivariable analysis was conducted to identify any associations between targeted therapy and 5-year OS. The outcome was defined as ‘death’ within a period of 5 years after the commencement of therapy. The study focused on OS as progression is not typically recorded in structured electronic health records (EHR), which is the primary source of data provided by healthcare organizations to TriNetX. Kaplan-Meier analysis, Cox proportional hazards regression, and calculation of risk ratio (RR), odds ratio (OR), and hazard ratio (HR) were conducted. Data analysis was restricted to a 5-year period following the initial diagnosis of HNC. Patients were considered healed if there was no recurrence of HNC or metastases within this period. Statistical analysis was conducted using the log-rank test, with a p-value of ≤0.05 considered statistically significant. For 1:1 matching, the propensity score-matching algorithm was used, which employs logistic regression to calculate a propensity score for each patient in the group. This can range from 0 to 1 and indicated the predicted probability that a patient was in cohort I or II given the patients covariates. The greedy nearest-neighbor matching algorithm with a caliper of 0.1 pooled standard deviation was used. The caliper setting within TriNetX has a fixed value of 0.1. In addition, the analysis of treatment pathways for the use of targeted therapy drugs was performed.
Results
Assessment, allocation, and matching. The database was queried on 09/20/2023, therefore 10 patient data had to be excluded due to an index event of more than 20 years old. A total of 372,423 patients with HNC, of which 18,331 patients with targeted therapy from 84 HCOs were included in the study. 72% of the patients treated with targeted therapy were male, 27% showed a nicotine dependence, and 50% had lymph node metastases (Table I). A total of 9,829 patients received immune checkpoint inhibitors, 7,402 patients received EGFR inhibitors, 2,276 patients received VEGF inhibitors, 1,019 patients received mTOR inhibitors, 160 patients received c-MET inhibitors, 46 patients received RET inhibitors and 34 patients received PI3K inhibitors (Table II).
Comparison of patients with and without targeted therapy in the study population of 372,423 patients.
Targeted therapy groups included in this study, with their group-specific active ingredients and the number of patients with each received targeted therapy.
A total of 3,933 (40%) of the patients receiving immune checkpoint inhibitors died after 5 years. Furthermore, 3,3341 (45.19%) of the patients with EGRF inhibitors, 862 (37.9%) of the patients with VEGF inhibitors, 448 (44.1%) of the patients with mTOR inhibitors, 67 (41.9%) of the patients with c-MET inhibitors, 14 (30.4%) of the patients with RET inhibitors and 14 (41.2%) of the patients with PI3K inhibitors died within a 5-year survival follow up (Table III).
Targeted therapies considered in this study, with their number and percentage of deaths.
With targeted therapy (I) vs. patients without targeted therapy (II). Before propensity score matching (PSM), the “targeted therapy” cohort included 18,068 patients and the “non-targeted therapy” cohort included 361,261 patients. Both cohorts differed significantly (p<0.05) for age, sex, lymph node metastases, nicotine dependence, alcohol dependence and body mass index. After PSM, both cohorts consisted of 17,748 patients and only differed significantly for alcohol dependence and body mass index (Figure 2, Table IV).
Propensity score density function. Before propensity score matching (PSM) (below) and after PSM (top). (purple: targeted therapy, green: without targeted therapy).
Patient’s characteristics in the “targeted therapy” cohort and “non-targeted therapy” cohort before and after 1:1 matching for age, sex, lymph node metastases, nicotine dependence, alcohol dependence, and body mass index (BMI).
Immune checkpoint inhibitors (I) vs. VEGF inhibitors (II). Before PSM, the ICI cohort included 9,731 patients and the VEGF inhibitor cohort included 2,260 patients. Both cohorts differed significantly (p<0.05) for age, sex, lymph node metastases, nicotine dependence, alcohol dependence and body mass index. After PSM, both groups consisted of 2,260 patients and differed significantly for body mass index (Figure 3, Table V).
Propensity score density function. Before propensity score matching (PSM) (below) and after PSM (top). (purple: ICI, green: VEGF inhibitors).
Patient’s characteristics in the imune checkpoint inhibitor (ICI) cohort and VEGF inhibitor cohort before and after 1:1 matching for age, sex, lymph node metastases, nicotine dependence, alcohol dependence, and body mass index (BMI).
EGFR inhibitors (I) vs. VEGF inhibitors (II). Before PSM, the EGFR inhibitor cohort included 7,338 patients and the VEGF inhibitor cohort included 2,260 patients. Both cohorts differed significantly (p<0.05) for sex, lymph node metastases, nicotine dependence, alcohol dependence and body mass index. After PSM, both groups consisted of 2,260 patients and differed significantly for body mass index (Figure 4, Table VI).
Propensity score density function. Before propensity score matching (PSM) (below) and after PSM (top). (purple: EGFR inhibitors, green: VEGF inhibitors).
Patient’s characteristics in the EGFR inhibitor cohort and VEGF inhibitor cohort before and after 1:1 matching for age, sex, lymph node metastases, nicotine dependence, alcohol dependence, and body mass index (BMI).
VEGF inhibitors (I) vs. c-MET inhibitors (II). Before PSM, the VEGF inhibitor cohort included 2,260 patients and the c-MET inhibitor cohort included 159 patients. Both cohorts differed significantly (p<0.05) for age, sex, and lymph node metastases. After PSM, both groups consisted of 159 patients and did not differ significantly for any confounder (Figure 5, Table VII).
Propensity score density function. Before PSM (below) and after PSM (top). (purple: VEGF inhibitors, green: c-MET inhibitors).
Patient’s characteristics in the VEGF inhibitor cohort and c-MET inhibitor cohort before and after 1:1 matching for age, sex, lymph node metastases, nicotine dependence, alcohol dependence, and body mass index (BMI).
PI3K inhibitors (I) vs. mTOR inhibitors (II). Before PSM, the PI3K inhibitor cohort included 33 patients and the mTOR inhibitor cohort included 1,016 patients. Both cohorts differed significantly (p<0.05) for sex, lymph node metastases and nicotine dependence. After PSM, both groups consisted of 31 patients and did not differ significantly for any confounder (Figure 6, Table VIII).
Propensity score density function. Before PSM (below) and after PSM (top). (purple: PI3K inhibitors, green: mTOR inhibitors).
Patient’s characteristics in the PI3K inhibitor cohort and mTOR inhibitor cohort before and after 1:1 matching for age, sex, lymph node metastases, nicotine dependence, alcohol dependence, and body mass index (BMI).
PI3K inhibitors (I) vs. c-MET inhibitors (II). Before PSM, the PI3K cohort included 33 patients and the c-MET inhibitor cohort included 159 patients. Both cohorts differed significantly (p<0.05) for sex. After PSM, both groups consisted of 29 patients and did not differ significantly for any confounder (Figure 7, Table IX).
Propensity score density function. Before PSM (below) and after PSM (top). (purple: PI3K inhibitors, green: c-MET inhibitors).
Patient’s characteristics in the PI3K cohort and c-MET inhibitor before and after 1:1 matching for age, sex, lymph node metastases, nicotine dependence, alcohol dependence, and body mass index (BMI).
PI3K inhibitors (I) vs. RET inhibitors (II). Before PSM, the PI3K inhibitor cohort included 33 patients and the RET inhibitor cohort included 46 patients. Both cohorts differed significantly (p<0.05) for sex. After PSM, both groups consisted of 25 patients and did not differ significantly for any confounder (Figure 8, Table X).
Propensity score density function. Before PSM (below) and after PSM (top). (purple: PI3K inhibitors, green: RET inhibitors).
Patient’s characteristics in the PI3K inhibitor and RET inhibitor cohort before and after 1:1 matching for age, sex, lymph node metastases, nicotine dependence, alcohol dependence, and body mass index (BMI).
Immune checkpoint inhibitors (I) vs. EGFR inhibitors (II). Before PSM, the immune checkpoint inhibitor cohort included 9,731 patients and the EGFR inhibitor cohort included 7,338 patients. Both cohorts differed significantly (p<0.05) for age, sex, lymph node metastases and nicotine dependence. After PSM, both groups consisted of 7,186 patients and did not differ significantly for any confounder (Figure 9, Table XI).
Propensity score density function. Before PSM (below) and after PSM (top). (purple: Immune checkpoint inhibitors, green: EGFR inhibitors).
Patient’s characteristics in the immune checkpoint inhibitors and EGFR inhibitors cohorts before and after 1:1 matching for age, sex, lymph node metastases, nicotine dependence, alcohol dependence, and body mass index (BMI).
Survival analysis. With targeted therapy (I) vs. without targeted therapy (II). Within 5 years, 7,129 patients (40.2%) from the “targeted therapy” cohort and 3,847 patients (21.7%) from the “non-targeted therapy” cohort II died. A significant risk difference of 18.5% (p=0.0001) was observed. The risk ratio (RR) was 1.853 [95% confidence interval (CI)=1.793-1.916], odds ratio (OR) was 2.426 (95%CI=2.315-2.543) and hazard ratio (HR) was 2.271 (95%CI=2.183-2.362) (Table XII, Figure 10).
Kaplan–Meier survival analysis for a 5-year interval in patients with and without targeted therapy.
Kaplan–Meier survival analysis for a 5-year interval in patients with and without targeted therapy. (purple: With targeted therapy, green: Without targeted therapy).
Table XII presents the number of patients in each cohort with the primary outcome of ‘death’, as well as the median survival and survival probability at the end of 5 years. Additionally, a log-rank test, HR, and z-test for proportionality were conducted. Median survival is defined as the number of days when survival dropped below 50%; ‘−’ indicates that survival did not drop below 50% during the 5-year period.
Immune checkpoint inhibitors (I) vs. VEGF inhibitors (II). Within 5 years, 839 patients (37.1%) from the immune checkpoint inhibitor cohort and 857 patients (37.9%) from the VEGF inhibitor cohort died. A non-significant risk difference of 0.8% (p=0.58) was observed. The RR was 0.979 (95%CI=0.908-1.056), OR was 0.857 (95%CI=0.857-1.09) and HR was 1.243 (95%CI=1.130-1.368) (Table XIII, Figure 11).
Kaplan–Meier survival analysis for a 5-year interval in patients with immune checkpoint inhibitors and VEGF inhibitors.
Kaplan–Meier survival analysis for a 5-year interval in patients with immune checkpoint inhibitors and VEGF inhibitors. (purple: Immune checkpoint inhibitors, green: VEGF inhibitors).
Table XIII presents the number of patients in each cohort with the primary outcome of ‘death’, as well as the median survival and survival probability at the end of 5 years. Additionally, a log-rank test, HR, and z-test for proportionality were conducted.
EGFR inhibitors (I) vs. VEGF inhibitors (II). Within 5 years, 1,012 patients (44.8%) from the EGFR inhibitor cohort and 857 patients (37.9%) from the VEGF inhibitor cohort died. A significant risk difference of 6.9% (p=0.0001) was observed. The RR was 1.181 (95%CI=0.04-0.097, OR was 1.328 (95%CI=1.179-1.495) and HR was 1.234 (95%CI=1.127-1.351) (Table XIV, Figure 12). Table XIV presents the number of patients in each cohort with the primary outcome of ‘death’, as well as the median survival and survival probability at the end of 5 years. Additionally, a log-rank test, HR, and z-test for proportionality were conducted.
Kaplan–Meier survival analysis for a 5-year interval in patients with EGFR inhibitors and VEGF inhibitors.
Kaplan–Meier survival analysis for a 5-year interval in patients with EGFR inhibitors and VEGF inhibitors.. (purple: EGFR inhibitors, green: VEGF inhibitors).
VEGF inhibitors (I) vs. c-MET inhibitors (II). Within 5 years, 66 patients (41.5%) from the VEGF inhibitor cohort and 66 patients (41.5%) from the c-MET inhibitor cohort died. A non-significant risk difference of 0% (p=1) was observed. The RR was 1 (95%CI=0.77-1.298), OR was 1 (95%CI=0.64-1.562) and HR was 0.696 (95%CI=0.493-0.982) (Table XV, Figure 13).
Kaplan–Meier survival analysis for a 5-year interval in patients with VEGF inhibitors and c-MET inhibitors.
Kaplan–Meier survival analysis for a 5-year interval in patients with VEGF inhibitors and c-MET inhibitors. (purple: VEGF inhibitors, green: c-MET inhibitors).
Table XV presents the number of patients in each cohort with the primary outcome of ‘death’, as well as the median survival and survival probability at the end of 5 years. Additionally, a log-rank test, HR, and z-test for proportionality were conducted. PI3K inhibitors (I) vs. mTOR inhibitors (II). Within 5 years, 14 patients (45.2%) from the PI3K inhibitor cohort and 15 patients (21.7%) from the mTOR inhibitor cohort died. A non-significant risk difference of 3.2% (p=0.799) was observed. The RR was 0.933 (95%CI=0.548-1.588), OR was 0.878 (95%CI=0.324-2.384) and HR was 2.181 (95%CI=0.985-4.83) (Table XVI, Figure 14).
Kaplan–Meier survival analysis for a 5-year interval in patients with PI3K inhibitors and mTOR inhibitors.
Kaplan–Meier survival analysis for a 5-year interval in patients with PI3K inhibitors and mTOR inhibitors. (purple: PI3K inhibitors, green: mTOR inhibitors).
Table XVI presents the number of patients in each cohort with the primary outcome of ‘death’, as well as the median survival and survival probability at the end of 5 years. Additionally, a log-rank test, HR, and z-test for proportionality were conducted.
PI3K inhibitors (I) vs. c-MET inhibitors (II). Within 5 years, 13 patients (44.8%) from the PI3K cohort and 10 patients (34.5%) from the c-MET inhibitor cohort died. A non-significant risk difference of 10.3% (p=0.421) was observed. The RR was 1.3 (95%CI=0.683-2.475), OR was 1.544 (95%CI=0.535-4.452) and HR was 2.514 (95%CI=1.024-6.173) (Table XVII, Figure 15).
Kaplan–Meier survival analysis for a 5-year interval in patients with PI3K inhibitors and c-MET inhibitors.
Kaplan–Meier survival analysis for a 5-year interval in patients with PI3K inhibitors and c-MET inhibitors. (purple: PI3K inhibitors, green: c-MET inhibitors).
Table XVII presents the number of patients in each cohort with the primary outcome of ‘death’, as well as the median survival and survival probability at the end of 5 years. Additionally, a log-rank test, HR, and z-test for proportionality were conducted.
PI3K inhibitors (I) vs. RET inhibitors (II). Within 5 years, 13 patients (52%) from the PI3K inhibitor cohort and 10 patients (40%) from the RET inhibitor cohort died. A non-significant risk difference of 12% (p=0.395) was observed. The RR was 1.3 (95%CI=0.706-2.393), OR was 1.625 (95%CI=0.53-4.984) and HR was 3.574 (95%CI=1.321-9.67) (Table XVIII, Figure 16).
Kaplan–Meier survival analysis for a 5-year interval in patients with PI3K inhibitors and RET inhibitors.
Kaplan–Meier survival analysis for a 5-year interval in patients with PI3K inhibitors and RET inhibitors. (purple: PI3K inhibitors, green: RET inhibitors).
Table XVIII presents the number of patients in each cohort with the primary outcome of ‘death’, as well as the median survival and survival probability at the end of 5 years. Additionally, a log-rank test, HR, and z-test for proportionality were conducted.
Immune checkpoint inhibitors (I) vs. EGFR inhibitors (II). Within 5 years, 2,821 patients (39.3%) from the immune checkpoint inhibitor cohort and 3,267 patients (45.5%) from the EGFR inhibitor cohort died. A significant risk difference of 6.2% (p=0.0001) was observed. The RR was 0.863 (95%CI=0.831-0.897), OR was 0.775 (95%CI=0.726-0.828) and HR was 1.045 (95%CI=0.994-1.099) (Table XIX, Figure 17).
Kaplan–Meier survival analysis for a 5-year interval in patients with immune checkpoint inhibitors and EGFR inhibitors.
Kaplan–Meier survival analysis for a 5-year interval in patients with Immune checkpoint inhibitors and EGFR inhibitors. (purple: Immune checkpoint inhibitors, green: EGFR inhibitors).
Table XIX presents the number of patients in each cohort with the primary outcome of ‘death’, as well as the median survival and survival probability at the end of 5 years. Additionally, a log-rank test, HR, and z-test for proportionality were conducted.
Pathway analysis. For our study cohort, a treatment pathway analysis was run. In the first line of treatment (LOT1), the most administered regimen was cetuximab, followed by pembrolizumab and nivolumab. Pembrolizumab is more present in the second line of treatment (Figure 18).
Treatment pathways. The left side shows the ten most administered treatments, and the right side shows six lines of treatment (LOT). : Ipilimumab and nivolumab.
Discussion
The aim of this study was to examine the impact of different targeted therapies on the 5-year OS of patients with HNC. Additionally, we aimed to compare the effectiveness of current standard of care targeted therapies with other less established therapies in terms of potential survival benefits in a real-world setting. One of the most used targeted therapies in HNC were EGFR inhibitors. However, the present analysis observed a survival advantage for patients treated with c-MET inhibitors or VEGF inhibitors compared to those treated with EGFR inhibitors.
In modern oncology, targeted therapies are used to treat HNC with advanced tumor stages, as well as recurrent or metastatic states, which often result in a palliative situation (9). The selection of a suitable therapeutic agent is influenced by various factors, including tumor entity, the type of mutation present, expected toxicity, the patient’s wishes and interdisciplinary expert consensus. In a palliative setting, the primary goal is to maintain the patient’s quality of life while also maximizing their OS. This analysis compares various targeted therapies in terms of OS to determine the therapy that provides the longest OS in cases of drug choice uncertainty.
The American Society of Clinical Oncology (ASCO) guideline recommends pembrolizumab as the first-line treatment for patients with recurrent or metastatic HNSCC (9). This recommendation is based on the KEYNOTE-048 phase III trial, which compared the OS of patients receiving pembrolizumab (ICI) and those receiving cetuximab (EGFR inhibitor). In this study, patients who received pembrolizumab alone or in combination with chemotherapy had longer OS compared to those who received cetuximab alone or in combination with chemotherapy, regardless of their PD-L1 status (5, 28). However, in the present 5-year survival analysis of real-world data, no significant difference was found between treatment with pembrolizumab and treatment with cetuximab in terms of OS. Researchers came to a similar conclusion in the randomized phase II study (GORTEC 2015-01 PembroRad). They compared the combination of pembrolizumab and radiotherapy (RT) to cetuximab with RT and found no survival advantage of pembrolizumab over cetuximab. However, the toxicity of pembrolizumab plus RT was lower (29).
The study also compared the effectiveness of VEGF inhibitors and immune checkpoint inhibitors (ICI) and found that therapy with VEGF inhibitors resulted in significantly longer OS compared to therapy with ICI. This contradicts previous data from other studies, which showed limited efficacy and a low response to therapy with VEGF inhibitor monotherapy (30, 31). While there are currently no FDA-approved anti-angiogenesis therapies for HNSCC, a randomized phase III trial showed that bevacizumab in combination with chemotherapy resulted in prolonged OS compared to chemotherapy alone (32). Additionally, Laenens et al. (2022) discovered that therapy with ICI was linked to an increased risk of cardiovascular events (33), and resistance to ICI is common (34). This may explain the statistical survival advantage of patients treated with VEGF inhibitors over those treated with ICI in this collective. A systematic review (35) showed prolonged OS with tolerable toxicity in patients with advanced hepatocellular carcinoma treated with a combination of VEGF inhibitors and ICI.
A comparison of VEGF inhibitors and EGFR inhibitors, as well as c-MET inhibitors, showed that patients treated with VEGF inhibitors had a significantly longer OS than those treated with EGFR inhibitors or c-MET inhibitors. A study by Su et al. showed an advantage of cetuximab (EGFR inhibitor) over bevacizumab (VEGF inhibitor) in the treatment of irresectable colorectal cancer (36), whereas a meta-analysis by Cui et al. found a survival benefit with bevacizumab (37). The analysis revealed a survival benefit in patients treated with VEGF inhibitors compared to those treated with EGFR inhibitors. This may be due to the frequent development of resistance to EGFR inhibitors (38) but also to c-MET inhibitors (39) and therefore the efficacy of both may be reduced compared to that of VEGF inhibitors. Both EGFR and VEGF play important roles in tumor growth. Wang et al. demonstrated the benefits of combining EGFR inhibitors and VEGF inhibitors in advanced non-small cell lung cancer (NSCLC) (40). Additionally, this combination therapy showed a positive trend in OS in HNSCC (41).
The extended OS in patients treated with VEGF inhibitors compared to those treated with standard of care ICIs or EGFR inhibitors should be scrutinized. However, the current data on monotherapy or combination therapy with VEGF inhibitors in HNSCC is mostly outdated and typically involves only a small number of patients (31, 42-44). Due to the current design of the analysis, statements on ECOG status, biomarkers, comorbidities, and tumor stage are limited. Therefore, it is not possible to make a final evaluation of the benefit of VEGF inhibitors compared to the standard of care regime. These results should instead serve as an incentive to reconsider the current use of VEGF inhibitors and to investigate and secure a possible survival benefit in combination with other targeted therapies in subsequent studies.
When comparing the OS rates of patients with HNSCC who were treated with PI3K inhibitors to those treated with mTOR inhibitors, c-MET inhibitors, or RET inhibitors, it appears that patients treated with PI3K inhibitors had a lower likelihood of survival than those treated with a therapeutic agent from one of the other three targeted therapy groups. However, therapy with PI3K inhibitors has been found to result in increased resistance, inadequate inhibition of the target, increased adverse effects (45), and higher toxicity (46). As a result, the FDA has only approved a few different PI3K inhibitors to date. Combination therapy of PI3K inhibitors with other targeted therapies is believed to be more likely to achieve a relevant therapeutic response (45).
The real-world data analysis has several strengths, including a substantial patient cohort, a multi-institutional approach, and the use of PSM to mitigate the influence of confounding factors. Notably, multi-institutional outcomes have been lacking in the literature to date. The study conducted matching based on age and common risk factors such as smoking and alcohol consumption. This retrospective study relied on data sourced from the TriNetX database. This study did not investigate HPV status or other histopathological aspects, such as UICC stage, lymph node metastasis, or extracapsular spread (ECS), due to limitations in the database. TNM data is only available for patients whose data comes from the cancer registry. Outside of cancer registries, TNM data is recorded in free-text, requiring Natural-Language-Processing (NLP) extraction. NLP remains a challenging field due to the need for financial resources and the necessity of implementing data quality control measures to ensure trustworthy data. The use of ICD-10 codes also prevented the retrieval of information regarding differences in tumor stage and pathohistological features. Future studies should include a more detailed examination of histopathological details, such as tumor stage, grading, HPV status, and surgical resection status.
It is important to acknowledge the study’s constraints, such as the unavailability of data on the cause of death to determine whether it was cancer related. This information could not be obtained from the TriNetX database. Additionally, the real-world data used in this study was collected globally from various healthcare organizations (HCOs) in Europe, the Middle East, Africa, Asia, North America, and South America. Unfortunately, this multi-center analysis did not consider national and international differences in the treatment of oral cancer patients and epidemiological variations.
The study has a retrospective and descriptive nature, and the data integrity is outside of our control. This means that some patients may have been lost to follow-up for various reasons, including reestablishing care with a healthcare organization outside of the TriNetX network. Additionally, we were unable to perform additional or separate analyses due to the lack of access to the raw data. The limitations of our study also restrict the applicability of the results to guide clinical practice.
No data is available on the duration of intake, dosage, and form of administration (intravenous or oral). Therefore, it cannot be excluded that these key points have an impact on the survival of the patients. In particular, the difference in bioavailability between oral and intravenous administration and the associated differences in excretion cannot be taken into account here.
The findings need to be interpreted cautiously regarding the limitations discussed above. At a minimum, the harvested data support the hypotheses of longer OS with VEGF inhibitor therapy in comparison to other targeted therapies. In previous studies ICI therapy and EGFR inhibitor therapy as standard of care were mostly superior in comparison to other targeted therapies. A possible explanation is that in our analysis exclusive real-world data were used. The data collected are clinical data from real patients and were not obtained from studies. This could prompt additional research and subsequently offer researchers information on acceptable combinations. In the long term, if expert therapy decisions are to be made between two drugs, they may choose the therapeutic agent with the greater survival advantage especially in the palliative setting.
Conclusion
Targeted therapies are a crucial aspect of modern tumor therapy and are particularly significant in the treatment of HNSCC, especially in cases of tumor recurrence and advanced stages. The most used targeted therapies in HNSCC are ICIs and EGFR inhibitors. However, the present analysis shows an OS advantage for patients treated with VEGF inhibitors compared to those treated with EGFR inhibitors or ICIs. Before commencing therapy in multimorbid patients, it is important to consider the development of resistance, adverse effects, and increased toxicity. However, it should be noted that patients treated with targeted therapy still experience reduced OS. Future studies should investigate the neoadjuvant and combinatory use of targeted therapy.
Footnotes
Authors’ Contributions
All Authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by Josefine Baudrexl and Marcel Ebeling. The first draft of the manuscript was written by Josefine Baudrexl and Marcel Ebeling and all Authors commented on previous versions of the manuscript. All Authors read and approved the final manuscript. The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.
Conflicts of Interest
The Authors have no relevant financial or non-financial interests to disclose.
Funding
The Authors declare that no funds, grants, or other support were received during the preparation of this manuscript.
- Received January 17, 2024.
- Revision received February 8, 2024.
- Accepted February 9, 2024.
- Copyright © 2024 The Author(s). Published by the International Institute of Anticancer Research.
This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY-NC-ND) 4.0 international license (https://creativecommons.org/licenses/by-nc-nd/4.0).
References
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.