Abstract
Background/Aim: Pembrolizumab alone or combined with chemotherapy is now approved in PD-L1-positive patients with recurrent or metastatic squamous cell carcinoma of the head and neck (SCCHN). Since real-world data are pending, our goal was to evaluate the efficacy and safety of immune checkpoint inhibitor (CPI) therapy in an unselected cohort of patients with SCCHN. Patients and Methods: We analyzed 78 patients with recurrent or metastatic SCCHN from three Austrian cancer centers that received CPI therapy alone or with chemotherapy as palliative first-line systemic treatment for this retrospective study. Patient characteristics, details on treatment, and survival were analyzed by a chart-based review. Results: Of the 78 patients analyzed, 55 patients were treated with CPI alone (45 with Pembrolizumab, 10 with Nivolumab) and 23 patients received chemotherapy with a platinum and 5-FU in addition to CPI. With a median follow-up of twelve months, the median PFS of all patients was 4 months [95% confidence interval (CI)=2.2-5.8] and the median OS was 11 months (95% CI=7.1-14.9). The overall response and disease control rates were 20.5% and 46.1%, respectively. There was no statistically significant difference in clinical outcome between patient groups with a different combined positive score (CPS). The rate of reported immune related adverse events was comparable to existing data. Conclusion: Our findings confirm the results of the KEYNOTE-048 trial that CPI therapy alone or together with chemotherapy is an effective treatment for patients with recurrent or metastatic CPS-positive SCCHN.
An estimated 650,000 new cases of squamous cell carcinoma of the head and neck (SCCHN) are diagnosed each year worldwide, making it the sixth most common cancer type, accounting for around 1-2% of all cancer deaths (1). Around one third of patients with locally advanced disease relapse after primary treatment with curative intention, besides those 10 percent primarily diagnosed in the metastatic setting, creating a high medical need for effective palliative therapy concepts (2-4). Despite modern treatment combinations, survival of patients with recurrent or metastatic (R/M) SCCHN remains poor. The standard of therapy for these patients has long been intensive combination chemotherapy based on platinum agents together with fluorouracil (5-FU) and taxanes, alone or in combination with the anti-EGFR antibody cetuximab (5, 6). Since patients with SCCHN frequently suffer from multiple comorbidities, many of them have been unable to take advantage of this effective but toxic treatment option.
As the evasion of the immune system plays a critical role in carcinogenesis, reactivation of the anti-tumor immune response employing immune-checkpoint inhibitors (CPI) has become a cornerstone in the therapy of multiple malignancies during the last ten years (7). Only a few years ago, the first clinical trials evaluating the safety and efficacy of CPI therapy in SCCHN were conducted, showing promising results. In 2016, Ferris et al. showed the PD-1 inhibitor Nivolumab to be more effective in treating platinum-refractory R/M SCCHN than the standard systemic therapy at the time, consisting of cetuximab, docetaxel or methothrexate (8). In unselected patients, the overall response to CPI monotherapy unfortunately remains relatively low, with around 10-20% (9).
Nevertheless, CPI therapy alone or in combination with chemotherapy has since managed to replace the prior standard therapies in palliative first-line treatment of SCCHN and now several trials analyzing its use in definitive multimodality treatment concepts are underway (10). Most notably, published in late 2019, the phase-3 KEYNOTE-048 trial compared the PD-1 inhibitor pembrolizumab as monotherapy and in combination with a platinum and 5-FU backbone to the EXTREME-regimen (platinum, 5-FU and cetuximab) in a palliative platinum-sensitive first-line setting (11). A significant overall survival (OS) benefit of pembrolizumab, both in the monotherapy and in the combination therapy groups, compared to the EXTREME regimen and a favorable safety profile was demonstrated. In subgroup analyses, this benefit was predominantly observed in patients with PD-L1 combined positive score (CPS) of at least one in the immunohistochemical examination, suggesting a higher activity and therefore response rate in tumors expressing PD-L1. The results of the KEYNOTE-048 trial led to the approval of pembrolizumab alone or in combination with chemotherapy in this setting.
While the efficacy of CPI therapy was shown in randomized clinical trials, real-world data in unselected patients are scarce, with the only published data so far describing a purely Japanese population (1, 12, 13). Therefore, we set out to analyze the outcome of unselected R/M SCCHN patients treated with these different CPI therapies for routine purposes and compare it to the published literature.
Patients and Methods
Patients. We analyzed 78 patients with recurrent or metastatic R/M SCCHN from three Austrian cancer centers (General Hospital Medical University of Vienna/General Hospital, University Hospitals Salzburg and Innsbruck) who received CPI therapy with either pembrolizumab or nivolumab between 2018 and 2020 for this retrospective study. In analogy to the KEYNOTE-048 study, patients receiving CPI as palliative first-line systemic treatment were included, unless progression within 6 months after platinum-based therapy with curative intention was observed.
All patients were retrospectively included in the analysis. We collected the clinical characteristics and follow-up data of all included patients. Clinical data including patient characteristics, the stage of disease, laboratory values, OS, and progression free survival (PFS) were analyzed by chart-based review. Charlson Comorbidity Index (CCI) was used to evaluate prognostic significance of comorbidities (14). PFS was calculated from the date of first CPI therapy until disease progression or death from any cause and OS was defined as the time between first CPI therapy and death from any cause. Response assessment using the RECIST 1.0 criteria was performed by bi- or three-monthly computed tomography (CT) imaging (15).
Ethics approval and consent to participate. This study was approved by the ethical committees at the participating institutions and the Central Ethical Committee of the Province of Salzburg (415-EP/73/662/2016). The study was conducted according to the guidelines of the Declaration of Helsinki. Informed consent was obtained from all subjects involved in this study.
Immunohistochemical analysis. CPS analysis of the patients’ routine tumor samples was done as part of standard histopathological workup by a trained pathologist, using either the Dako Agilent PD-L1 IHC 22C3 pharmDx kit (Innsbruck and Salzburg) or the Nordic BioSite rabbit monoclonal PD-L1 antibody (clone BSR-90) (Vienna). HPV association in oropharyngeal cancers was assessed mainly via p16 immunohistochemistry, while an HPV PCR was not routinely done in all tumor samples.
In the presence of CPS analyses of more than one tumor sample (e.g., taken at time of primary diagnosis and from distant metastasis), we used the most recent one before initiation of palliative therapy in our analysis. Three CPS subgroups were defined in analogy to the KEYNOTE-048 study – CPS below 1, CPS 1 to 19, and CPS 20 and above.
Statistical analysis. Statistical analyses were performed using IBM® SPSS® statistics software, version 24. Where appropriate, Kruskal-Wallis test and Pearson’s chi-squared test were used to compare the clinical characteristics of patient subgroups. For survival analyses, Kaplan-Meier curve analyses and the log-rank test were applied. A two-sided p-value of 0.05 was considered statistically significant.
Results
Patient characteristics. Table I and Table II depict the characteristics, as well as the outcome and response rates of all included patients, separated by treatment group and CPS. Of the 78 patients analyzed, 55 (70.5%) were treated with a CPI alone (81.8% of those with pembrolizumab and 18.2% with nivolumab) and 23 (29.5%) were treated with a combination therapy of pembrolizumab and chemotherapy (5-FU and a platinum-based agent).
Patient characteristics. The clinical characteristics and response data of all patients, as well as the comparison between the different patient subgroups, divided according to therapeutic regimen, are shown.
Patient characteristics. The clinical characteristics and response data of all patients, as well as the comparison between the different patient subgroups, divided according to CPS-subgroups, are shown.
Patients were predominantly male (82.1%), and the majority had a performance score of 0 or 1 (55.2% of all patients or 74.1% of patients of which the performance score was documented). The median age of patients at time of first diagnosis was 61.0 years, while the median age when first requiring palliative therapy was 62.5 years.
Indication for CPI therapy was distant failure in 17 patients (21.8%) and local recurrence or extensive disease in 40 patients (51.3%), while both distant metastases and local recurrence was present in 29 patients (26.9%). A numerically lower number of patients with isolated locoregional recurrence were treated with a combination therapy of CPI and chemotherapy.
Most primary tumors were localized in the oral cavity (30.8%) or oropharynx (30.8%). Of the 25 oropharyngeal tumors, 10 (40.0%) were HPV-associated. A history of tobacco smoking was documented in 46 patients (59.0%).
Approximately one third of patients (30 patients or 38.5%) had none or few relevant comorbidities, with a CCI score of 0 to 2 (estimated 10-year-survival of above 90%), while most patients (37 or 47.4%) had some comorbidities carrying p>prognostic significance, with a CCI score of 3 or 4 (estimated 10-year-survival of 77% and 53%, respectively). A minority of patients (10 patients or 12.8%) had severe comorbidities limiting their life-expectancy, with a CCI score of 5 or 6 (21% or 10% estimated 10-year-survival, respectively), and one patient had a CCI score of 7 and therefore an estimated 10-year-survival chance of 0%.
Response and survival data. The overall response rate (ORR) in all 78 patients was 20.5% (16 patients). Four patients (5.1%) achieved a complete response (CR), while 12 (15.4%) achieved a partial response (PR), and a stable disease (SD) was documented in 20 patients (25.6%). No response with disease progression at time of first restaging was seen in 25 patients (32.1%). At the time of data collection, primary response assessment was not yet available in 17 patients (21.8%) due to the short follow-up. After a median follow-up of 12 months, the median PFS of all patients was 4.0 (95% CI=2.2-5.8) months and the median OS was 11.0 (95% CI=7.1-14.9) months. There was no statistically significant survival difference between the different CPS groups. Immune related adverse events were observed in 16.7% of patients (most commonly hyperthyroidism and skin reactions), with 2.6% being grade 3 or higher (one colitis and one hepatitis).
When comparing the 55 patients that were treated with CPI alone to the 23 receiving combination therapy with a chemotherapy containing 5-FU and a platinum agent (Table I), there was no statistically significant difference in PFS [median PFS with CPI monotherapy was 4.0 (95% CI=1.5-6.5) months vs. 3.0 months with combination therapy (95% CI=0.7-5.3); p=0.674; Figure 1]. Patients receiving combination therapy showed a non-statistically significant trend towards longer OS [median OS with CPI monotherapy was 10.0 (95% CI=6.3-13.7) months vs. 15.0 months in combination therapy (95% CI=8.7-21.3); p=0.293; Figure 1]. Notably, the percentage of patients with a CPS of 20 or more was significantly higher in the CPI monotherapy group (50.9% vs. 21.7%, p=0.009). There were no significant differences in relevant patient characteristics (e.g. sex, age, performance score) within the different CPS groups. Response to therapy was not statistically different either; however, by tendency more patients treated with CPI monotherapy achieved a SD or higher (52.3% vs. 30.4%; p=0.174).
Progression-free survival (left) and overall survival (right) of patients treated with either checkpoint-inhibitor therapy alone (n=23) or in combination with chemotherapy (n=55). p-Values were generated using the log-rank test.
CPS subgroups and dynamics. Naturally, most patients treated with CPI therapy had tumors with a positive CPS ≥1 (91.0%). A CPS of 1-19 was recorded in 38 patients (48.7%) and a high CPS (≥20) was observed in 33 patients (42.3%). CPS analysis was not available in 5 patients (6.4%).
More than one tumor sample was available in 13 patients, often from different time points (e.g., at primary diagnosis and at relapse or metastasis). In that case, only the CPS of the most recent biopsy was used in our analysis. Notably, a change of the CPS-subgroup was seen in 8 patients; a relevant CPS increase compared to the biopsy taken at primary diagnosis leading to change of the CPS subgroup was seen in 3 patients (one of which was CPS negative at first diagnosis), and a relevant CPS decrease leading to change of the subgroup was seen in 5 patients (two of which completely lost PD-L1 expression). In the remaining five patients, no change in CPS-subgroup was seen (Figure 2). The two patients (2.6%) with a negative CPS in the most recent available biopsy with an initially positive CPS at first diagnosis were treated with CPI therapy; both of them received CPI monotherapy as first-line palliative treatment (one of them achieving SD for 5 months) and were treated with chemotherapy in later therapy lines after progression to CPI therapy.
Combined positive score of tumor samples from the same patients taken at time of first diagnosis and local recurrence/distant metastasis.
There were no significant differences in PFS of patients with a high CPS≥ 20 compared to patients with a CPS of 1-19 [median PFS of 4.0 (95% CI=0.2-7.8) months vs. 4.0 (95% CI=1.7-6.3) months; p=0.563; Figure 3]. OS was also not statistically significant different either [CPS≥20 vs. CPS 1-19, median OS of 11.0 (95% CI=6.9-15.1) months vs. 12.0 (95% CI=7.1-16.9) months; p=0.939; Figure 3]. The clinical characteristics according to CPS subgroups are also shown in Table I. As noted above, significantly more patients in the group with CPS≥20 were treated with CPI monotherapy (84.8% vs. 52.6%, p=0.009). When we analyzed patients receiving CPI monotherapy or combination therapy separately, there was no significant PFS or OS benefit for those with a CPS ≥20 compared to the other CPS-positive patients.
Progression-free survival (left) and overall survival (right) of patients, divided according to CPS-subgroups (CPS<1, n=2; CPS 1-19, n=38; CPS≥20, n=33). p-Values were generated using the log-rank test.
Discussion
To our knowledge, we are the first western study group reporting real-world data about the efficacy and safety of palliative first-line CPI therapy in platinum-sensitive patients with R/M SCCHN that were treated outside of a clinical trial, thereby closely reflecting clinical practice. We documented nearly 50% disease control and a response rate of around 20% in a disease with a historically unfavorable prognosis using treatments that were relatively well tolerable in patients with multiple comorbidities.
Our cohort closely resembles the general population of SCCHN patients in clinical practice without much referral or selection bias, as there are no independent oncologists outside of hospitals in Austria and, in contrast to clinical trials, we did not apply any inclusion criteria besides the treating physician’s decision. To ease indirect comparison to the KEYNOTE-048 trial, we only included platinum-sensitive patients with an interval of at least 6 months after completing curative systemic therapy before starting palliative first-line CPI therapy in our analysis.
Comparing our data to the KEYNOTE-048 trial, patient and tumor characteristics (e.g., sex, age, HPV status and CPS distribution) are very similar. The rate of patients with a history of tobacco smoking was around 60%. In patients with an oropharyngeal primary tumor, 40% were HPV-associated. This is in line with other published data for patients in western Europe (16). However, we also included patients with tumor localizations that were excluded in the KEYNOTE-048 trials (e.g., nasopharyngeal and paranasal sinuses) in our study.
Unfit patients with a higher ECOG performance score of 2 and more were included in our analysis, although they were not included in larger prospective studies as the KEYNOTE-048 trial. Most patients in our study had relevant comorbidities that by themselves reduced their 10-year-survival expectancy, documented via the CCI (14). The documentation of relevant comorbidities and calculation of the CCI for every included patient makes a more objective comparison with other patient cohorts significantly easier. The fact that only 38.5% of our patients had no comorbidities that affected their estimated survival (CCI 0, 1 or 2) supports the clinical impression that this is a highly vulnerable and complex to treat patient population.
We also included ten patients receiving CPI monotherapy with nivolumab in our analysis. No statistical analyses comparing the efficacy of nivolumab to pembrolizumab could be done, as the number of patients receiving either drug as monotherapy was relatively small. However, retrospective analyses comparing these two PD-L1 inhibitors in other tumor entities show no significant differences in efficacy (17-19).
The rate of patients with a CPS≥20 was virtually identical in our analysis (42%) and the KEYNOTE-048 study (43%). Only two patients with a negative CPS were included in our analysis, so no reliable statement can be made about this group of patients. Interestingly, both CPS-negative patients were initially CPS positive at the time of first diagnosis and lost CPS-positivity in later biopsies of distant metastases. They both received CPS monotherapy and one of them could achieve SD for 5 months.
More than one tumor sample at different points in time and/or localizations were available in 13 patients, 9 of which showed changes in the CPS subgroup; one newly gaining and two completely losing CPS-positivity. Such discrepancies in PD-L1 status of different biopsies have already been reported in SCCHN and other tumor entities, with several factors possibly attributing to such findings (20-22). Clonal evolution could lead to the emergence of new dominant tumor clones over time and thereby to temporal changes in PD-L1 expression of the tumor. Such evolution can occur naturally on its’ own, especially in polyclonal tumors, or under the influence of external selective pressure (e.g., chemotherapy given as part of curative first-line treatment in our patients) (23, 24). Differences in PD-L1 expression between different biopsies taken at the same time could also be explained by inter- and intratumoral heterogeneity, as observed in several tumor entities (20-22, 25, 26). Further studies analyzing the clinical implications of such discrepancy in PD-L1 status of different biopsies, and the importance of multiple biopsies are needed.
The median follow-up of 12 months in our study is similar to the initial report of the KEYNOTE-048 study (10 to 13 months, depending on the treatment arm). However, due to several patients having a very short follow-up, primary response assessment via computed tomography was still outstanding at the time of data collection in around one fifth of patients.
The ORR of 20.5% in all our patients is very similar to the results reported in the KEYNOTE-048 study; there, the ORR of patients with a CPS≥1 to pembrolizumab alone was 19%, while the ORR to pembrolizumab in combination with chemotherapy was 36%. As no response assessment was yet available for one fifth of our patients, the ORR could be higher after longer follow-up. We did see a numerically higher ORR (27.3%) and a higher number of complete remissions (12.1%) in patients with a high CPS ≥20, compared to patients with a lower positive CPS of 1-19 (ORR 15.8%, no CR). However, the low numbers of patients in our analysis may preclude the achievement of a statistically significant difference (p=0.200). This could also explain the higher response rates in patients receiving CPI monotherapy, as significantly more patients with a high CPS were included in that group. As responses to CPI monotherapy were less likely in patients with a lower CPS, these patients were more often treated with additional chemotherapy in clinical practice, provided there were no contraindications such as a poor performance score.
In clinical practice, several factors influence the choice of therapy. The combination of CPI and chemotherapy is usually preferred in patients that need a fast response due to high tumor burden or have a good performance score, while we reserve CPI monotherapy for frail patients or those with a low tumor burden. Interestingly, while there were no statistically significant differences, numerically fewer patients with isolated locoregional recurrence were treated with combination therapy in our analysis, leading to the assumption that the overall tumor burden was an important factor contributing to the choice of treatment. We also observed numerically fewer patients with a higher CCI and therefore more relevant comorbidities were receiving combination therapy. This could also explain the trend towards a longer OS in the combination therapy group.
Survival data in our analysis resemble the findings of the KEYNOTE-048 study. We report a median PFS of 4.0 months in patients receiving CPI monotherapy and 3.0 months in patients receiving combination therapy, versus around 3 and 5 months in the KEYNOTE-048 study, respectively. Our OS data are similar as well, with a median OS of 10.0 months in the CPI monotherapy patients (vs. 11.6 months in the KEYNOTE-048 study) and 15.0 months after combination therapy (vs. 13.0 months in the KEYNOTE-048 study). Our findings are also in line with recently published real-world data in a Japanese population, also describing similar results to the KEYNOTE-048 trial in non-Caucasian patients. We therefore conclude that the response and efficacy of CPI therapy alone and in combination with chemotherapy in unselected real-world patients is similar to the available data from clinical trials.
As always, in retrospective studies, assessment of adverse events is hampered by incomplete and unstandardized documentation. Often, relatively minor adverse events not requiring intervention of the treating physician are not as thoroughly documented in clinical routine. Immune related adverse events (irAEs) are of special clinical interest due to the need of early recognition and specific treatment and the possible influence on further therapeutic procedures. We made the choice to therefore document them in more detail compared to other adverse events in clinical practice. Therefore, we focused on irAEs in our study. The rate of irAEs of special interest in patients receiving CPI therapy in the KEYNOTE-048 study was around 26-31%, with 5-7% being grade 3 or higher. In contrast, we documented 16.7% irAES, with 2.6% being grade 3 or higher, perhaps due to the retrospective collection of data. Types of irAES were similar, with hypothyroidism being one of the most common irAEs in both our analysis and the KEYNOTE-048 study (11). No treatment-related deaths were documented in our patients. Our findings therefore confirm the safety of CPI therapy, reiterating clinical experience and prior published data; however, treating physicians must be aware of severe irAEs occurring in around 5% of patients treated with CPI therapy.
Limitations of our study include the retrospective nature of the study design, the unstandardized and presumably incomplete documentation of adverse events in some patients and the relatively small number of patients. Also, HPV-PCR was not done in all patients with oropharyngeal tumors due to differences in local standards. Furthermore, different CPS antibody clones were used for CPS assessment during histopathological workup at the different centers, highlighting the lack of general standard outside of clinical trials. While extensive data is available about the widely used 22C3 antibody, there exists no published work comparing it to the BSR-90 clone used in Vienna. 22C3 has binding profiles in the extracellular domain of PD-L1, while no information about the binding epitope of BSR-90 is available. This is clearly a limitation of our study, since results of PD-L1 testing could differ between the different sites. However, the BSR-90 antibody is used routinely for clinical purposes in Vienna and was validated and approved at the Department of Pathology – Medical University of Vienna according to the quality management system. In addition, due to the small number of patients available for analysis, our results present only statistical trends and should be viewed in context with already published data to draw conclusions about clinical practice.
Conclusion
In conclusion, our findings could reproduce and consecutively validate the results of the KEYNOTE-048 trial that CPI therapy alone or together with chemotherapy is an effective and well tolerable treatment for patients with R/M CPS-positive SCCHN. Particularly patients with a CPS≥ 20 show promising response rates, with complete remission being possible. CPI monotherapy is well tolerable in patients with significant comorbidities, but special caution must be taken as serious immune-related adverse events can occur in a small percentage of patients. As response rates and overall prognosis of patients with SCCHN remain low even in the era of CPI therapy, further studies to evaluate new treatment modalities are needed.
Footnotes
Authors’ Contributions
Conceptualization, Sandro M. Wagner, Thomas Melchardt, Lukas Weiss, Gabriele Gamerith, Georg Pall, Richard Greil and Thorsten Fuereder; Data curation, Sandro M. Wagner, Teresa Magnes, Dominik Kiem, Daniel Neureiter, Christina Wagner, Marie-Bernadette Aretin, Stefan F. Nemec, Gabriele Gamerith, Georg Pall and Thorsten Fuereder; Formal analysis, Sandro M. Wagner and Thomas Melchardt; Supervision, Thomas Melchardt, Georg Pall, Richard Greil and Thorsten Fuereder; Writing – original draft, Sandro M. Wagner; Writing – review & editing, Teresa Magnes, Thomas Melchardt, Dominik Kiem, Lukas Weiss, Daniel Neureiter, Christina Wagner, Marie-Bernadette Aretin, Stefan F. Nemec, Gabriele Gamerith, Georg Pall, Richard Greil and Thorsten Fuereder. All Authors have read and agreed to the final version of this manuscript.
Conflicts of Interest
Conflicts of interest in relation to our research work: TMa: speaking fees from MSD; travel support from BMS. TMe: speaking fees and travel support from Merck, MSD and BMS. LWe: speaking fees from BMS, MSD and Merck Serono; travel support from Merck Serono; advisory role for Merck Serono, MSD and BMS. GPa: speaking fees from MSD and BMS; advisory role for MSD and BMS. RGr: speaking fees Merck and BMS; advisory role for Merck and BMS; research funding from Merck and BMS; travel support from MSD. TFu: research funding from MSD and Merck KGaA; speaking fees from MSD, Merck KGaA and BMS. All other Authors declare no potential conflicts of interest.
- Received January 5, 2023.
- Revision received January 19, 2023.
- Accepted January 20, 2023.
- Copyright © 2023 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.
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.