Salvage Chemotherapy After Pembrolizumab for Recurrent or Metastatic Head and Neck Squamous Cell Carcinoma

Abstract

Background/Aim

Pembrolizumab-based therapy is widely used for recurrent or metastatic head and neck squamous cell carcinoma (R/M HNSCC). However, clinical outcomes after subsequent salvage chemotherapy (SCT) remain insufficiently defined. This study evaluated the efficacy of SCT after pembrolizumab in patients with R/M HNSCC.

Patients and Methods

This multicenter retrospective study included 36 evaluable patients with R/M HNSCC who received SCT after pembrolizumab between December 2019 and February 2022. The primary endpoint was the objective response rate. Secondary endpoints included the disease control rate, progression-free survival, overall survival, and exploratory clinical factors associated with treatment outcomes.

Results

The best overall response to SCT was a complete response in 2 patients, partial response in 8, stable disease in 12, and progressive disease in 14. The overall response rate was 27.8%, and the disease control rate was 61.1%. Median progression-free and overall survival were 4.6 months and 11.5 months, respectively. The overall response rate was significantly higher (>4-fold) in patients with prior surgery and in platinum-naive patients. A combined positive score of 20 or higher for tumor programmed cell death-ligand 1, initial M1 status, and prior surgery were associated with longer progression-free survival in exploratory analyses.

Conclusion

SCT after pembrolizumab showed clinically meaningful activity in selected patients with R/M HNSCC. These findings support SCT as a reasonable post-pembrolizumab option, although prospective validation is warranted.

Introduction

Recurrent or metastatic head and neck squamous cell carcinoma (R/M HNSCC) remains difficult to treat. Current clinical guidelines include pembrolizumab-based therapy as an important systemic treatment option for R/M HNSCC, and the KEYNOTE-048 trial established the immune checkpoint inhibitor (ICI) pembrolizumab, alone or with chemotherapy, as a first-line standard in appropriate patients (1-3). As pembrolizumab is used more frequently, disease progression after pembrolizumab is more common and patients then require subsequent systemic therapy. Several real-world studies from our multicenter group have described pembrolizumab outcomes, clinical factors, hematological markers, immune-related adverse events (irAEs), combined positive score (CPS)-related outcomes, and quality of life in patients with R/M HNSCC (4-8). Salvage chemotherapy (SCT) is often considered when patients retain adequate performance status and organ function. However, the clinical outcomes of SCT after pembrolizumab remain insufficiently defined.

Several retrospective studies have reported meaningful activity of chemotherapy after ICI therapy in R/M HNSCC. Saleh et al. reported an objective response rate (ORR) of approximately 30% after SCT following ICI progression (9). Other studies also suggested that post-ICI systemic therapy may provide clinically relevant tumor control (10, 11). In a previous study, Fushimi et al. reported favorable outcomes with SCT after nivolumab compared with best supportive care, supporting the clinical relevance of active systemic treatment after anti-programmed cell death protein 1 (PD-1) therapy (12). Cetuximab- and taxane-containing regimens have also shown activity in this setting (13). More recently, Abe et al. reported the clinical role of SCT after ICI therapy in patients with R/M HNSCC (14). These studies are important, but many included mixed ICI exposure, nivolumab-focused cohorts, or heterogeneous treatment sequences. Pembrolizumab-specific real-world data remain limited.

The present multicenter retrospective study therefore evaluated SCT after pembrolizumab in patients with R/M HNSCC. The primary endpoint was the objective response rate after SCT. Secondary endpoints included the disease control rate (DCR), progression-free survival (PFS), overall survival (OS), and exploratory clinical factors associated with treatment outcomes. All subgroup analyses were interpreted as exploratory because this study only included patients who were able to receive SCT after pembrolizumab.

Patients and Methods

Patients. This multicenter retrospective observational study is reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) framework where applicable (15). We reviewed patients with R/M HNSCC who received SCT after pembrolizumab at four participating institutions between December 2019 and February 2022. The participating institutions were Tokyo Medical University Hospital, Tokyo Medical University Hachioji Medical Center, International University of Health and Welfare Mita Hospital, and Kitasato University Hospital.

Patients were eligible when they had histologically confirmed squamous cell carcinoma of the head and neck, received pembrolizumab-based therapy for recurrent or metastatic disease, and subsequently received evaluable SCT after pembrolizumab. Patients were excluded when they received pembrolizumab in a clinical trial, had non-squamous cell carcinoma, continued pembrolizumab without subsequent SCT at the time of data collection, or received best supportive care alone after pembrolizumab discontinuation.

During the study period, 201 patients received pembrolizumab with or without chemotherapy for R/M HNSCC. One patient was excluded because they were treated in a clinical trial. Thirty-three patients with non-squamous cell carcinoma were also excluded. Among the remaining 167 patients with squamous cell carcinoma, 111 had discontinued pembrolizumab and 56 were still receiving pembrolizumab at the time of data collection. Of the 111 patients who discontinued pembrolizumab, 36 received SCT and were included in the present analysis (Figure 1).

Figure 1.

Study flow diagram. Among 201 patients who received pembrolizumab with or without chemotherapy for head and neck cancer, 34 were excluded. Of 167 patients with squamous cell carcinoma (SCC), 111 had discontinued pembrolizumab. Among these 111 patients, the 36 who received salvage chemotherapy were included in the present analysis.

Treatment. Pembrolizumab-based therapy was defined as pembrolizumab monotherapy or pembrolizumab combined with platinum-based chemotherapy. Pembrolizumab was administered at a dose of 200 mg every 3 weeks or 400 mg every 6 weeks according to routine clinical practice.

SCT was defined as systemic anticancer therapy started after discontinuation of pembrolizumab. SCT included cetuximab plus paclitaxel, cetuximab plus cisplatin and 5-fluorouracil, S-1, and other systemic regimens selected by the treating physicians.

Treatment selection was based on the patient’s condition, disease extent, prior treatment history, organ function, and physician judgment at each institution. Dose modification, treatment interruption, and treatment discontinuation were determined by the treating physicians according to clinical status and toxicity.

Data collection. Clinical data were collected retrospectively from medical records. The collected variables included age, sex, primary tumor site, Eastern Cooperative Oncology Group performance status, stage at initial diagnosis, initial M1 status, surgery before pembrolizumab, recurrent or metastatic site, pembrolizumab regimen, best response to pembrolizumab, SCT regimen, radiotherapy before pembrolizumab, platinum exposure before SCT, programmed cell death-ligand (PD-L1) CPS, best response to SCT, PFS, OS, follow-up duration after the start of SCT, and irAEs recorded during pembrolizumab-based therapy.

CPS was categorized as less than 1, 1 to less than 20, 20 or higher, or unknown. Prior platinum exposure was defined as any platinum-containing systemic therapy given before SCT, including platinum used before or with pembrolizumab. Missing values were handled descriptively and were not imputed.

Outcomes. The primary endpoint was the ORR after SCT. Secondary endpoints were DCR, PFS, OS, and exploratory clinical factors associated with treatment outcomes. irAEs recorded during pembrolizumab-based therapy were summarized descriptively.

PFS was defined as the time from the start of SCT to disease progression or death from any cause. OS was defined as the time from the start of SCT to death from any cause. Patients without an event were censored at the date of last follow-up. The start date of SCT was used as the index date for all survival analyses.

Response assessment. Tumor response was assessed as complete response, partial response, stable disease, or progressive disease according to the Response Evaluation Criteria in Solid Tumors version 1.1 (16). The best overall response was defined as the best response recorded after the start of SCT. The ORR was defined as the proportion of patients with complete or partial response. The DCR was defined as the proportion of patients with a complete or partial response, or stable disease.

Imaging evaluation was performed as part of routine clinical practice at each institution. The timing of imaging assessment was determined by the treating physicians. Patients with clinically apparent disease progression were classified as having progressive disease even when imaging evaluation was not available.

Statistical analysis. Patient characteristics and treatment patterns are summarized using descriptive statistics. Continuous variables are reported as the median and range. Categorical variables are reported as a number and percentage.

Associations between clinical factors and the ORR were evaluated using Fisher’s exact test. PFS and OS were estimated using the Kaplan-Meier method. Subgroup comparisons for survival outcomes were performed using the log-rank test. Cox proportional hazards models were not constructed because of the limited number of patients and events. Therefore, all factor analyses were interpreted as exploratory and hypothesis-generating.

A two-sided p-value of less than 0.05 was considered statistically significant. All statistical analyses were performed using EZR (Saitama Medical Center, Jichi Medical University, Saitama, Japan), a graphical user interface for R (R Foundation for Statistical Computing, Vienna, Austria) (17).

Ethics. This study was approved by the Institutional Review Board of the participating institutions (approval numbers: T2022-0048, 5-22-3, and C22-007). The requirement for written informed consent was waived because of the retrospective observational design. Patients were able to refuse participation through an opt-out announcement on the institutional websites. This study was conducted in accordance with the principles of the Declaration of Helsinki.

Results

Patient selection and characteristics. The baseline characteristics of the 36 patients are summarized in Table I. The median age was 64.5 years (range= 36-78 years), and 30 patients were male. Initial status was M1 for 9 patients, and 12 patients had a history of surgery before therapy with pembrolizumab. Pembrolizumab was administered as monotherapy to 23 patients and in combination with cisplatin and 5-fluorouracil to 13 patients. Cetuximab plus paclitaxel was the most common SCT regimen and was administered to 21 patients.

Table I. Patient characteristics and treatment background (N=36).

5-FU: 5-Fluorouracil; CDDP: cisplatin; CPS: combined positive score; CR: complete response; ECOG PS: Eastern Cooperative Oncology Group performance status; PD: progressive disease; PD-L1: programmed cell death-ligand 1; PR: partial response; SCT: salvage chemotherapy; SD: stable disease. ‘Other’ sites included primary tumor sites not otherwise listed in the table.

Primary endpoint: ORR after SCT. The best overall response to SCT is summarized in Table II. A complete response was observed in 2 patients, partial response in 8, stable disease in 12, and progressive disease in 14. The ORR was 27.8%, and the DCR was 61.1%.

Table II. Treatment response and survival outcomes after salvage chemotherapy.

CR: Complete response; PD: progressive disease; PR: partial response; SD: stable disease.

The ORR was greater for patients with surgery before pembrolizumab than for those without prior surgery (58% vs. 12%, p<0.01). The ORR was also greater for patients without platinum exposure before SCT than for those with prior platinum exposure (42% vs. 0%, p=0.01). These analyses were exploratory because of the limited sample size.

Survival outcomes. The median follow-up period after the start of SCT was 6.6 months (range=0.4-15.8 months). Median PFS and OS after SCT were 4.6 months and 11.5 months, respectively (Figure 2).

Figure 2.

Kaplan-Meier curves for survival after salvage chemotherapy. (A) Progression-free survival from the start of salvage chemotherapy. (B) Overall survival from the start of salvage chemotherapy.

Exploratory factors associated with survival. Exploratory subgroup analyses for PFS are summarized in Table III. Patients with a PD-L1 CPS of 20 or higher had significantly longer PFS than those with a CPS of less than 20 (median=7.0 vs. 1.8 months, p<0.0001). Patients with initial M1 status had significantly longer PFS than those without initial M1 status (median=5.0 vs. 2.7 months, p=0.03). Patients who had undergone surgery before pembrolizumab also had significantly longer PFS than those without prior surgery (median=9.8 vs. 3.7 months, p<0.01).

Table III. Selected exploratory factors associated with response and survival outcomes.

CR: Complete response; CPS: combined positive score; ORR: objective response rate; OS: overall survival; PD: progressive disease; PD-L1: programmed cell death-ligand 1; PFS: progression-free survival; PR: partial response; SCT: salvage chemotherapy; SD: stable disease. Only clinically relevant factors and factors with p<0.10 in univariate analyses are shown. All analyses were exploratory because of the small sample size and limited number of events. Statistically significant p-values are shown in bold.

No clinical factor reached statistical significance for OS. Surgery before pembrolizumab showed a borderline association with longer OS (p=0.051). Initial M1 status and response to pembrolizumab also showed nonsignificant trends toward being associated with longer OS (p=0.09 and p=0.08, respectively). Multivariate models were not constructed because of the small number of patients and events.

irAEs during pembrolizumab-based therapy. irAEs recorded during pembrolizumab-based therapy occurred in six patients (16.7%). The recorded events were grade 2 pneumonitis, grade 2 thyroid dysfunction, grade 1 adrenal dysfunction, grade 1 erythema multiforme, grade 1 pemphigoid, and grade 3 mucositis in one patient each. No grade 4 or 5 irAEs was recorded in the available medical records.

Discussion

In this multicenter retrospective study, SCT after pembrolizumab showed clinically meaningful antitumor activity in selected patients with R/M HNSCC. The primary endpoint was the ORR, and the observed response suggests that systemic therapy can remain active after pembrolizumab in patients who are still able to receive further treatment. This finding is clinically relevant because the number of patients who require treatment after pembrolizumab is increasing. However, this cohort only included patients who proceeded to SCT after pembrolizumab discontinuation, and the results should be interpreted as outcomes in a selected population.

This finding is consistent with previous reports of chemotherapy after ICI therapy. Saleh et al. reported an ORR of approximately 30% after SCT following progression with ICI (9). Other studies also reported activity of post-ICI systemic therapy, including cetuximab- and taxane-containing regimens (10, 11, 13). Our previous study also suggested that SCT after nivolumab was associated with better outcomes than best supportive care (12). These data support active systemic treatment after anti-PD-1 therapy for selected patients but they do not prove that prior therapy with pembrolizumab directly enhanced subsequent sensitivity to chemotherapy.

The exploratory response analysis provides clinically useful but non-definitive information. The ORR was significantly higher for patients who had undergone surgery before therapy with pembrolizumab. This association should not be interpreted as evidence that surgery itself increased sensitivity to SCT; rather, prior surgery may be a clinical selection marker. Patients who underwent surgery were likely to have had disease that was once amenable to curative-intent local treatment, a preserved general condition, and a clinical course that allowed sequential systemic therapy. They may also have had less uncontrolled locoregional disease or lower symptomatic tumor burden at the start of SCT. Because tumor burden, the disease-free interval, locoregional symptom burden, and the number of prior treatment lines were not quantified, this finding should be regarded as hypothesis-generating.

The ORR was also higher for patients without platinum exposure before SCT. This result is clinically plausible but requires careful interpretation. Lack of prior platinum exposure may indicate less heavily pretreated disease, fewer treatment-resistant tumor clones selected by previous cytotoxic therapy, and better preserved organ function or bone marrow reserve. Conversely, prior platinum exposure may reflect more aggressive or treatment-resistant disease biology. Because the SCT regimens used were heterogeneous, this finding does not indicate that platinum-naive patients should preferentially receive platinum-based SCT. It suggests that previous cytotoxic treatment history may help identify patients who are more likely to respond to SCT post-pembrolizumab.

Several factors were associated with longer PFS in exploratory analyses. A high CPS may indicate a tumor-immune environment that was more responsive to pembrolizumab and which may remain clinically relevant during subsequent treatment. However, this interpretation is indirect. The favorable outcome in patients with initial M1 status is more difficult to explain and may reflect selection bias, differences in disease distribution, lower locoregional symptom burden, or greater ability to continue systemic therapy. Because data on the tumor volume, target lesion diameter, and treatment-free interval were not available, no mechanistic conclusion can be drawn from this finding.

irAEs recorded during pembrolizumab-based therapy were observed in a small number of patients and were mainly grade 1 or 2 events, except for one grade 3 mucositis. These data show that immune-related toxicity occurred during the treatment sequence but they do not define the safety of SCT. Standardized adverse-event data during SCT were not available; therefore, it was not possible to establish the safety profile of SCT after pembrolizumab.

Study limitations. Firstly, this study was retrospective and included only 36 patients who received SCT after pembrolizumab, which creates substantial selection bias. Secondly, SCT regimens were heterogeneous and were selected by individual physicians. Thirdly, multivariate analysis was not performed because of the limited number of patients and events. Fourthly, data regarding important confounders such as tumor burden, disease-free interval, number of previous treatment lines, time from pembrolizumab discontinuation to SCT, and standardized AEs during SCT were not available.

These findings have practical implications for post-pembrolizumab treatment planning. SCT may be considered when patients remain fit for systemic therapy, but prior surgery and platinum-naive status should be used only as exploratory clinical markers. Future studies should collect standardized treatment histories, tumor burden, treatment-free intervals, and SCT-related AEs to clarify which patients benefit most from SCT after pembrolizumab.

Conclusion

SCT after pembrolizumab showed meaningful real-world activity for selected patients with R/M HNSCC. Surgery before pembrolizumab and platinum-naive status before SCT may help identify patients with a higher probability of response, but these findings remain exploratory. Larger studies with standardized SCT safety assessment and appropriate comparator groups are needed to clarify the optimal treatment sequence after pembrolizumab.

Conflicts of Interest

The Authors declare no conflicts of interest in relation to this study.

Authors’ Contributions

Chihiro Fushimi designed the study, analyzed the data, and drafted the manuscript. Yuri Ueda, Takuro Okada, Isaku Okamoto, Takashi Matsuki, Hideaki Takahashi, Hiroki Sato, Takahito Kondo, Kunihiko Tokashiki, Kenji Hanyu, Tatsuya Ito, Tatsuo Masubuchi, Kushihashi Yukiomi, Yuichiro Tada, Go Omura, Taku Yamashita, Nobuhiko Oridate, and Kiyoaki Tsukahara collected clinical data and interpreted the results. All Authors reviewed and approved the final manuscript.

Acknowledgements

The Authors thank all participating institutions for their contribution to data collection.

Artificial Intelligence (AI) Disclosure

During the preparation of this manuscript, generative AI was used to support language editing and formatting. All scientific content, data interpretation, and final wording were reviewed and verified by the Authors, who take full responsibility for the content of the manuscript.