Safety and Efficacy of Cetuximab‐Based Salvage Chemotherapy After Checkpoint Inhibitors in Head and Neck Cancer

Abstract

Background

There are still few data on the activity and safety of cetuximab‐based salvage chemotherapy after immunotherapy (SCAI) in patients with squamous cell cancer of the head and neck (SCCHN).

Materials and Methods

This was a retrospective study of patients with SCCHN who received cetuximab‐based SCAI after programmed cell death protein 1 or programmed cell death ligand 1(PD[L]1) inhibitors. Overall response rate (ORR) and disease control rate (DCR) with SCAI and with last chemotherapy before immunotherapy (LCBI) by RECIST 1.1, percentage change from baseline in target lesions (PCTL), progression‐free survival (PFS), overall survival (OS), treatment compliance, and toxicity were evaluated.

Results

Between March 2016 and November 2019, 23 patients were identified. SCAI consisted of cetuximab‐based combinations (3‐weekly cisplatin‐5FU‐cetuximab [n = 2], weekly paclitaxel‐cetuximab [n = 17], weekly cisplatin‐cetuximab [n = 2], weekly carboplatin‐paclitaxel‐cetuximab [n = 2]). ORR was 56.5% (11 partial response, 2 complete response). DCR was 78.3%. Among 13 objective responders, median best PCTL was −53.5% (range, −30% to −100%). Median OS and PFS were 12 months and 6 months, respectively. In 10 patients receiving LCBI, ORR to LCBI was 40%, whereas ORR to SCAI achieved 60%. In LCBI‐treated patients, median PFS with LCBI was 8 months and median PFS and OS with SCAI were 7 months and 12 months, respectively. Reduced dose intensity of the chemotherapy and cetuximab components occurred in 82.6% and 52.2% of the patients. Grade 1 or 2 adverse events (AEs) occurred in all patients. Grade 3 or 4 AEs developed in 65%, being grade 3 in all of them except in one patient (grade 4 neutropenia). There were no treatment‐related deaths.

Conclusion

Cetuximab‐based salvage chemotherapy after PD(L)1 inhibitors associated with high response rates and deep tumor reductions with a manageable safety profile. Subsequent lines of therapy may explain the long survival achieved in our series. These results invite to design studies to elucidate the best therapeutic sequence in patients with SCCHN in the immunotherapy era.

Implications for Practice

Cetuximab‐based salvage chemotherapy (SCAI) achieved high response rates in patients with recurrent/metastatic squamous cell cancer of the head and neck (SCCHN) after progression to PD‐1/PD‐L1 inhibitors. Objective response rate was higher than and progression‐free survival was comparable to that of chemotherapy administered before immunotherapy (IO). In most patients, SCAI consisted of weekly, well‐tolerated regimens. These observations have implications for current practice because of the limited evidence to date in SCCHN and the scant therapeutic options in this disease and invite to elucidate which may be the best treatment sequence for patients with head and neck cancer in the IO era.

Short abstract

This study explored the safety and efficacy of salvage chemotherapy combined with cetuximab after progression to immune checkpoint inhibitors in patients with recurrent/metastatic squamous cell cancer of the head and neck.

Introduction

Several immune checkpoint inhibitors (ICIs) have shown activity in the recurrent/metastatic (R/M) setting of squamous cell cancer of the head and neck (SCCHN). Reported objective response rates (ORRs) with programmed cell death protein 1 or programmed cell death ligand 1 (PD[L]1) inhibitors in platinum‐refractory R/M SCCHN fall below 15%, with median overall survival (OS) times between 6 and 9 months [1, 2, 3, 4, 5]. Among nonplatinum‐refractory patients with a Combined Positive Score (CPS) ≥20 treated with pembrolizumab alone in the first‐line setting, ORR achieved 23% and a median OS of 14.9 months [6]. Chemotherapy is able to modify the highly immunosuppressive tumor microenvironment, this possibly explaining in part the additive effect seen with chemoimmunotherapy combinations across different tumors, including SCCHN [6, 7]. In addition, as reported in the KEYNOTE‐048 study, progression‐free survival measured from the start of first‐line until progression with second‐line therapies has been shown to be prolonged in patients treated with ICIs in first line [8]. Therefore, sequencing ICIs and chemotherapy could be another promising strategy. Retrospective series and case reports in lung cancer and in other entities have shown increased responses to salvage chemotherapy after immunotherapy (SCAI) [9, 10, 11, 12, 13, 14, 15, 16]. However, data demonstrating such responses in SCCHN are still limited, particularly when using anti–epidermal growth factor receptor (EGFR)‐based chemotherapy [13, 15, 18, 19, 20, 21, 22, 23]. Our aim was to retrospectively study the safety and efficacy of salvage chemotherapy combined with cetuximab after progression to ICIs in patients with R/M SCCHN.

Materials and Methods

Study Design and Patients

This was a retrospective study of patients with R/M SCCHN that showed overt radiological and clinical progression during first‐, second‐, or third‐line treatment with programmed cell death protein (PD‐1) or programmed cell death ligand 1 (PD‐L1) inhibitors and were subsequently treated with SCAI.

A descriptive study of the general baseline characteristics of the patients was performed, including geriatric assessment employing the modified Charlson Comorbidity Index (mCCI), the Adult Comorbidity Evaluation‐27 (ACE‐27), and the Generalized Competing Event Composite Omega Score (GCE‐COS), the latter using an online tool developed to summarize different comorbidity scores to predict relative cancer versus noncancer risk, as previously described [24]. mCCI ≥2 and ACE‐27 ≥3 indicate moderate‐to‐severe comorbidity. A GCE‐COS ≥0.6 indicates a higher overall survival with oncological treatment. For mCCI calculation in the present series, patients with SCCHN were considered as localized disease if they were only locoregionally recurrent and as metastatic if they had distant disease.

ORR according to RECIST 1.1, ORR in target lesions only (ORRTL), percentage of tumor change from baseline in TL (PCTL), progression‐free survival (PFS), and OS from the start of SCAI and from the start of first‐line therapy were studied [25]. ORR and PFS of last chemotherapy given before immunotherapy (LCBI) were also analyzed. The primary endpoint was ORR and PCTL during SCAI. Secondary endpoints were safety; treatment compliance; PFS during SCAI and LCBI, defined as the time from the start of SCAI or LCBI until disease progression or death due to any cause; OS since SCAI, defined as the time from the start of SCAI until death from any cause; and OS since the first line of treatment, defined as the time from the start of first‐line therapy until death from any cause. Efficacy results are presented using different plots (waterfall plots, spider plots and swimmer plots) to better report the kinetics of response, as stated in the Trial Reporting in Immuno‐Oncology guidelines [26]. Adverse events (AEs) were recorded using the National Cancer Institute Common Terminology Criteria for Adverse Events, version 4.0.

Evaluation of Response

Tumor response was assessed according to RECIST 1.1 criteria with computed tomography (CT) scans performed every 8 to 12 weeks, according to the local protocol. Response evaluation was undertaken independently by two expert radiologists in head and neck cancer. In case of disagreement, the case was reassessed and a consensus was reached.

Partial response (PR) was defined as a ≥ 30% decrease in the sum of the diameters of target lesion (TL), progressive disease (PD) was defined as a ≥ 20% increase in the sum of diameters of TL or the appearance of new metastatic lesions, and stable disease (SD) was defined as a < 30% decrease to a 20% increase of tumor size.

PD‐L1 Expression

PD‐L1 membrane expression was measured in tumor cells using an immunohistochemistry assay (PD‐L1 IHC 22C3 pharmDx; Dako North America, Carpinteria, CA). PDL1 was considered positive when ≥1% of cells showed partial membrane staining [1].

Statistical Analyses

A descriptive analysis of demographic and clinicopathological data was performed. PFS and OS were estimated using the Kaplan‐Meier method. The χ² test was used for comparing qualitative non‐normally distributed variables. Statistical analyses were performed using R v3.6.3 under R‐Studio 1.1.383 (R Development Core Team Vienna, Austria; https://www.r‐project.org). The study was approved by the Institutional Review Board of Hospital Clínico Universitario San Carlos, and was conducted in accordance with the principles of the Declaration of Helsinki. The requirement of informed consent was waived as the study was based on a retrospective analysis of existing administrative and clinical data.

Results

Baseline Characteristics

Between March 2016 and November 2019, 23 patients were included and were followed until August 2020. Baseline characteristics of the patients are summarized in Table 1.

Table 1.

Patient and tumor characteristics

Patients	n = 23, n (%)
Age at SCAI, years
Median (min–max)	65 (36–89)
≥70	7 (30.4)
<70	16 (69.6)
ECOG at SCAI
0	1 (4.3)
1	6 (26.1)
2	15 (65.2)
3	1 (4.3)
mCCI, median (range: min–max)	8 (4–10)
ACE‐27, median (range: min–max)	3 (3–3)
GCE‐COS a
Median (range: min–max)	0.77 (0.51–0.93)
COS <0.6	3 (13)
COS ≥0.6	20 (87)
Smoking history	16 (69.6)
Sex
Male	17 (73.9)
Female	6 (26.1)
Anatomic subsite
Oral cavity	13 (56.5)
Oropharynx	3 (1 HPV+) (13)
Larynx	4 (17.4)
Hypopharynx	2 (8.7)
Unknown primary	1 (4.3)
Stage at initial diagnosis (AJCC 8th Ed)
II	2 (8.7)
III	2 (8.7)
IVA	13 (56.5)
IVB	2 (8.7)
IVC	4 (17.4)
Treatment at initial diagnosis
Surgery	15 (2 salvage Sx post‐CRT) (65.2)
Adjuvant RT	4 (17.4)
Adjuvant CRT	9 (39.1): 3wkCDDP: 4;Cetuximab: 5
Radical CRT	5 (21.7): wkCDDP: 2 Cetuximab: 3
Induction CT	7 (30.4): ERBITAX: 3;TPF: 4
Upfront TX for R/M	4 (17.4); anti‐PD‐L1 + anti‐CTLA4 (CT);anti‐PD‐L1 + iSTAT3 (CT);pembrolizumab;extreme
R/M disease at SCAI
Locoregional only	4 (17.4)
Distant only	3 (13)
Locoregional + Distant	16 (69.6)
No. of lines for R/M disease	4 (2–6)
No. of lines before IO
0	13 (56.5)
1	9 (39.1)
2	1 (4.3)
Platinum before IO b
No	15 (65.2)
Yes
PRf	7 (30.4)
Non‐PRf	1 (4.3)
LCBI (n = 10)
EXTREME	6 (60)
ERBITAX	4 (40)
IO
Within clinical trials
Anti‐PD‐1	1 (4.3)
Anti‐PD‐L1	3 (13)
Anti‐PD‐L1 + STAT3 inhibitor	7 (30.4)
Anti‐PD‐L1 + CXCR2 inhibitor	1 (4.3)
Regular practice
Nivolumab	10 (43.5)
Pembrolizumab	1 (4.3)
SCAI
EXTREME	2 (8.7)
ERBITAX	17 (74)
CARBITAX	2 (8.7)
wkCDDP‐Cetuximab	2 (8.7)

^aGCE‐COS ≥0.6 indicates significantly higher overall survival with treatment intensification (i.e., chemotherapy).

^bPlatinum received either in the locally advanced and/or R/M settings. Platinum‐refractory disease was defined as patients progressing to platinum‐based regimen for advanced disease or relapsing within 6 months of platinum‐based therapy with curative intent.

Abbreviations: ACE‐27, Adult Comorbidity Evaluation‐27; AJCC, American Joint Committee on Cancer; CARBITAX, weekly carboplatin‐paclitaxel‐cetuximab; CRT, concurrent chemoradiotherapy; CT, chemotherapy; ECOG, Eastern Cooperative Oncology Group performance status; ERBITAX, paclitaxel plus cetuximab; EXTREME, 3‐weekly cisplatin‐5FU‐cetuximab; GCE‐COS, Generalized Competing Event Composite Omega Score; IO, immunotherapy; HPV, human papillomavirus; LCBI, last chemotherapy before immunotherapy; mCCI, modified Charlson Comorbidity Index; PRf, platinum‐refractory; R/M, recurrent/metastatic; RT, radiotherapy; SCAI, salvage cetuximab‐based chemotherapy after immunotherapy; Sx, surgery; TPF, 3‐weekly docetaxel‐platinum‐5FU; TX, therapy; wkCDDP, weekly cisplatin; 3wkCDDP, 3‐weekly cisplatin.

The most common primary tumor location was the oral cavity (56.5%). Eighty‐two percent were stage IV at initial diagnosis, 65.2% had been treated with surgery, 56.5% with adjuvant radiotherapy (RT) or chemoradiotherapy (CRT), 30.4% with induction chemotherapy, and 21.7% with radical CRT. In the R/M setting, 43.4% of the patients had received at least one line of chemotherapy prior to immunotherapy (IO), and up to 30.4% were platinum‐resistant.

Median age was 65 years, and 30.4% were ≥ 70 years. At SCAI, a majority of patients had an ECOG performance status ≥2 (69.5%). Median mCCI score was 8, ACE‐27 was 3, and GCE‐COS was 0.77.

ICIs and cetuximab‐based chemotherapy administered prior and after immunotherapy were as follows.

PD‐1/PD‐L1 Inhibitors Used

Twelve patients received PD‐L1 or PD‐1 inhibitors within clinical trials (NCT02252042, NCT02551159, NCT02499328): PDL1 inhibitors (n = 11) and PD‐1 inhibitors (n = 1); 11 patients received PD‐1 inhibitors as per standard practice: nivolumab (n = 10) and pembrolizumab (n = 1); PD‐1/PD‐L1 inhibitors were always administered either as single agents or in combination with other ICIs (anti‐CTLA‐4) or immunomodulatory agents (STAT3 or CXCR2 inhibitors).

Cetuximab‐Based Chemotherapy Regimens

Four chemotherapy regimens (all in combination with the anti‐EGFR agent cetuximab) were used either as SCAI, LCBI, or both.

• Weekly cetuximab (400 mg/m² [load], 250 mg/m² [maintenance]) plus paclitaxel (80 mg/m²; ERBITAX regimen): LCBI (n = 4), SCAI (n = 17).

• Weekly cetuximab (400 mg/m² [load], 250 mg/m² [maintenance]) plus paclitaxel (80 mg/m²) plus carboplatin (AUC 2; CARBITAX regimen): SCAI (n = 2).

• Weekly cisplatin (50 mg/m²), and weekly cetuximab (400 mg/m² [load], 250 mg/m² [maintenance]): SCAI (n = 2).

• Three‐weekly cisplatin (100 mg/m²), infusional 5FU (750 mg/m² per day on days 1 to 5) and weekly cetuximab (days 1, 8 and 15; EXTREME regimen): LCBI (n = 6), SCAI (n = 2).

Response During SCAI

Objective Responses and Percentage Change from Baseline in Target Lesions in Objective Responders

ORR was 56.5% (13/23), with 11 PRs and 2 complete responses (CRs). Eight of these responders received SCAI in second line and five received it in third line. Five patients showed SD, and five showed PD as best responses during SCAI. Disease control rate was 78.3%. Median best PCTL in objective responders was −57.5% (range, −30 to −100).

Responses to SCAI occurred in one patient treated with EXTREME and in 12 patients treated with weekly regimens (ERBITAX [n = 10], CARBITAX [n = 1], weekly cisplatin [wkCDDP]‐cetuximab [n = 1]; Figs. 1, 2; Table 2).

Figure 1.

Efficacy of SCAI in the whole population and in LCBI‐pretreated patients. Waterfall plot showing objective response in TL with SCAI (A). Spider plot showing best objective response in TL and PCTL during treatment with SCAI (B). Swimmers plot depicting treatment exposure and overall response duration per RECIST 1.1 with SCAI (C). Waterfall and spider plots showing best objective responses in TL with LCBI (D) and with SCAI (E). ORR was higher for SCAI (60%) than for LCBI (40%). In addition responses were deep and durable. Note that scales are different between D and E. Kaplan‐Meier estimates of PFS and OS in LCBI‐treated patients (F) and in the whole population since the start of SCAI and since first line (G, H). Comparison of OS since first line in patients treated with first‐ versus second‐ or further IO lines (H). *These patients achieved PR in TL but were classified as PD due to the appearance of pulmonary metastases [(A) patients 6 and 10; (E) patient 10]. +This patient achieved a CR to SCAI [(E) patient 21]. Dashed (A, D, E) and solid (B) black lines mark the limits that define objective responses. Abbreviations: CR, complete response; IO, immunotherapy; LCBI, last chemotherapy before immunotherapy; ORR, objective response rate; OS, overall survival; PCTL, percentage change from baseline in target lesions; PD, progressive disease; PR, partial response; SCAI, cetuximab‐based salvage chemotherapy after immunotherapy; SD, stable disease; TL, target lesions.

Table 2.

Response rate, best percentage change from baseline in target lesions, and Kaplan‐Meier estimates of progression‐free and overall survival

A. Response rate
Tumor response	SCAI	LCBI	SCAI (LCBI pretreated)
n	23	10	10
ORR, n (%)	13/23 (56.5)	4/10 (40)	6/10 (60)
Type of response, n (%)
CR	2/23 (8.7)	0	1/10 (10)
PR	11/23 (47.8)	4/10 (40)	5/10 (50)
SD	5/23 (21.7)	5/10 (50)	2/10 (20)
PD	5/23 (21.7)	1/10 (10)	2/10 (20)
Best PCTL among objective responders, % (range)	−57.5 (−30 to −100)	‐	−35.5 (−30.9 to −100)
ORR, n (%)
IO in first line	7/12 (58.3)	‐	‐
IO in second/third line	6/11 (54.5)	‐	‐
DCR	18/23 (78.3)	9/10 (90)	8/10 (80)
Response in TL only
ORR	15/23 (65.2)	‐	7/10 (70)
PCTL among objective responders, % (range)	−57.5 (−30 to −100)
PCTL among objective responders in TL only, % (range)	−53.5 (−30 to −100).	‐	−36 (−30.9 to −100)
Patients with tumor reduction in TL, % (range)	82.6 (19/23)	‐	80 (8/10)
PCTL among patients with tumor reduction in TL, % (range)	−45 (−4 to −100)	‐	−35.5 (−100 to −7.5)

B. Follow‐up and survival
	SCAI	LCBI	SCAI (LCBI pretreated)	First line a	First line vs. second line IO b
n	23	10	10	23	13 vs. 10
Follow‐up c	12 mo (4–38)	6 mo (1–12)	5 (2.5–15)	22 (9–51)	22 (9–51)
PFS	6 mo (95% CI, 5–8)	8 mo (95% CI, 4–NR)	7 mo (95% CI, 4–NR)	‐	‐
OS	12 mo (95% CI, 9‐NR)	‐	12 mo (95% CI, 6–NR)	28 (95% CI, 23–NR)	First‐line IO: 27 mo (95% CI, 15–NR) vs. second‐line IO 28 mo (95% CI, 17–NR) p = .94

^aSurvival times calculated from the start of first‐line therapy in the whole population.

^bSurvival times calculated from the start of first‐line therapy in patients treated with first‐line vs. second‐line IO.

^cFollow‐up times are expressed as median (range: min–max).

Abbreviations: (‐), not applicable; CI, confidence interval; CR, complete response; IO, immunotherapy; LCBI, last chemotherapy before immunotherapy; NR, not reached; ORR, objective response rate; OS, overall survival; PCTL, best percentage change from baseline in target lesions; PD, progressive disease; PFS, progression‐free survival; PR, partial response; RR, response rate; SCAI, cetuximab‐based salvage chemotherapy after immunotherapy; SD, stable disease; TL, target lesion.

Objective responses to SCAI occurred in 4 of 12 patients previously treated with PD‐1 or PD‐L1 inhibitors within clinical trials and in 10 of 11 patients treated with PD1 inhibitors outside clinical trials, this difference being statistically significant (p = .019). In addition, ORR to SCAI was higher in patients treated with PD1 inhibitors (most outside clinical trials) (anti‐PD‐1 vs. anti‐PD‐L1: 83% vs. 27%, p = .007). There were no differences in ORR to SCAI among patients treated with IO in first line versus those treated with IO in second or third lines (58.3% vs. 54.5%, p = 0.768; Table 2).

Response Rate in Target Lesions and Percentage Change from Baseline in Target Lesions

There were two patients with paradoxical responses, which achieved a major response in TL but suffered the appearance of new metastatic lesions. ORRTL was 65.2% (15/23). Among responders in TL, median best PCTL was −53.5% (range, −30 to −100). Nineteen patients (82.6%) showed some degree of tumor reduction in TL, with a median best PCTL of −45% (range: −4% to −100%; Figs. 1, 2; Tables 2, 3).

Table 3.

Clinical history and lines of treatment received before, during, and after salvage cetuximab‐based chemotherapy after immunotherapy (SCAI) for each individual patient

Patient order No.	Age/primary tumor /AJCC 8th Ed stage/PDL1/treatment at initial DX	Age/PR/ECOG/ mCCI/ACE‐27/GCE‐COS at first SCAI	Location of disease at first SCAI	First line (ORR/PCTL /TD)	Second line (ORR/PCTL/TD)	Third line (ORR/PCTL/TD)	Fourth line (ORR/PCTL /TD)	Fifth line (ORR/PCTL /TD)	Sixth line (ORR/PCTL /TD)	OS SCAI, OS first line, status
1.	M/54 y/OC/IVA/NA/iCT➔SX➔3wkCDDP‐RT	58/PRf/ECOG 1/mCCI = 7/ACE27 = 3/COS = 0.779	Mx (LN, Lung)	EXTREME (7.5 m)	Anti‐PD1 (CT)	ERBITAX PR/‐45% (6.5 mo)	–	–	–	OS SCAI: 12 m, OS first L: 31 mo, dead
2.	M/69 y/HP/II/80%/BRT➔SX	71/PNRf/ECOG 1/ mCCI = 10/ACE27 = 3/COS = 0.51	Mx (Lung)	Anti‐PDL1 + Anti‐CTLA4 (CT)	ERBITAX, PR/−30%/7 mo	Nivo PR/‐80/6 mo	ERBITAX PD/+20%/1 mo	Nivo ‐/‐/1 mo	EXTREME SD/‐10%/5 mo	OS SCAI: 20 mo, OS first L: (CT), dead
3.	M/62 y/HP/IVA/<1%/BRT➔SX	65/PRf/ECOG 2/ mCCI = 9/ACE27 = 3/COS = 0.775	L + R + Mx (LN)	EXTREME	Anti‐PDL1 + iSTAT3 (CT)	ERBITAX SD/‐7.5%/3 mo	–	–	–	OS SCAI: 3 m, OS first L: 15 mo, dead
4.	F/70 y/HP/IVB/NA/BRT➔SX	75/PNRf/ECOG 2/ mCCI = 8/ACE27 = 3/COS = 0.741	L + R + Mx (Liver)	Anti‐PDL1 (CT)	ERBITAX, PD/+22%/9 mo	–	–	–	–	OS SCAI: 10 mo, OS first L: (CT), Dead
5.	M/59 y/OC/IVA/<1%/SX➔3wkCDDP‐RT	61/PRf/ECOG 2/ mCCI = 8/ACE27 = 3/COS = 0.821	L + R + Mx (LN)	Anti‐PDL1 + iSTAT3 (CT)	ERBITAX, PR/−50%/5 mo	Nivo PD/+50%/5 mo	EXTREME PR/‐60%/3 mo	Nivo CR/3 m	–	OS SCAI: 37 mo, OS first L: (CT), AWOD
6.	M/68 y/LX/IVC/NA/Anti‐PDL1 + iSTAT3 (CT)	68/PNRf/ECOG 2/ mCCI = 9/ACE27 = 3/COS = 0.716	L + R + Mx (Lung)	Anti‐PDL1 + iSTAT3 (CT)	ERBITAX, PD a /−85%/4 mo	Nivo (‐/‐/1 mo)	–	–	–	OS SCAI: 5 mo, OS first L: (CT), dead
7.	F/87 y/OC/IVA/NA/iCT (C‐Px)➔BRT	89/PNRf/ECOG 2/ mCCI = 7/ACE27 = 3/COS = 0.644	L + R	Anti‐PDL1 + iSTAT3 (CT)	ERBITAX, SD/−4%/7 mo	Nivo ‐/‐/ 1 mo	C‐wkCDDP ‐/‐/5m	–	–	OS SCAI: 12 mo, OS first L: (CT), dead
8.	M/35 y/OC/IVA/<1%/iCT (TPF x3)➔SX➔3wkCDDP + RT	36/PRf/ECOG 2/ mCCI = 6/ACE27 = 3/COS = 0.930	L + R + Mx (Lung, LN, bone)	ERBITAX PR/‐/9m	EXTREME, SD/‐/3 mo	Anti‐PDL1 + iSTAT3 (CT)	EXTREME PD/+64%/2m	ERBITAX PD/‐/2m	–	OS SCAI: 3 mo, OS first L: 15 mo, dead
9.	F/50 y/OC/IVA/<1%/SX➔BRT	51/PNRf/ECOG 2/ mCCI = 9/ACE27 = 3/COS = 0.843	L	ERBITAX PR/‐/12m	Nivo, PD/+60%/1 mo	EXTREME PR/‐35.5%/5 mo	Nivo PD/+50%/1 mo	C‐wkCDDP SD/‐10%/8 mo	–	OS SCAI: 14 mo, OS first L: 28 mo, dead
10.	M/60 y/LX/IVC/<1%/C‐PF	63/PRf/ECOG 1/ mCCI = 8/ACE27 = 3/COS = 0.756	L + R + Mx (Lung, LN, bone)	EXTREME SD/+10%/6 mo	Nivo, PD/‐/11 mo	ERBITAX PD a /‐46%/4 mo	Nivo ‐/‐/2 mo	–	–	OS SCAI: 6 mo, OS first L: 25 mo, dead
11.	M/61 y/OP/IVB/60%/BRT	62/PNRf/ECOG 2/ mCCI = 6/ACE27 = 3/COS = 0.769	L	Nivo PD/‐/1 mo	ERBITAX, PR/−100%/3 mo	C‐wkCDDP PR/‐/17 mo	ERBITAX PD/‐/5.5 mo	Pembro‐wkCBDCA‐Px ‐/‐/2 mo	–	OS SCAI: 26 mo, OS first L: 28 mo, AWD
12.	M/72 y/OP/IVC/20%/Anti‐PDL1 + anti‐CTLA4 (CT)	74/PNRf/ECOG 2/ mCCI = 10/ACE27 = 3/COS = 0.534	L + R + Mx (Bone) at diag	Anti‐PDL1 + Anti‐CTLA4 (CT)	Anti‐PDL1 + iCXCR2 (CT)	ERBITAX SD/‐14%/4 mo	–	–	–	OS SCAI: 5 mo, OS first L: (CT), dead
13.	F/64 y/OP/IVB/20%/iCT (TPF × 3)➔wkCDDP + RT	67/PNRf/ECOG 2/ mCCI = 9/ACE27 = 3/COS = 0.770	L + R + Mx (Lung)	Anti‐PDL1 (CT)	ERBITAX, SD/−11%/14 mo	Nivo SD/‐/13 mo	C‐wkCDDP SD/+10% 4 mo	–	–	OS SCAI: 36 mo, OS first L: (CT), AWD
14.	M/54 y/LX/IVA/5%/SX➔RT	57/PNRf/ECOG 0/ mCCI = 10/ACE27 = 3/COS = 0.716	Mx (Lung, liver)	ERBITAX PR/‐81.5%/	Nivo, PR/−70%/11 mo	ERBITAX PR/‐30.9%/20 mo	Nivo PD/‐/1 mo	–	–	OS SCAI: 20 mo, OS first L: 43 mo, AWD
15.	M/68 y/OC/IVB/NA/SX➔3wkCDDP + RT	69/PRf/ECOG 2/ mCCI = 8/ACE27 = 3/COS = 0.828	L + R + Mx (Lung, liver, LN, bone)	Nivo PD/‐/1 mo	ERBITAX, PR/−62.4%/8 mo	Nivo PD/‐/3 mo	ERBITAX SD/+15% /8.5 mo	–	–	OS SCAI: 18 mo, OS first L: 20 mo, AWD
16.	M/57 y/OC/III/90%/SX➔RT	58/PRf/ECOG 1/ mCCI = 4/ACE27 = 3/COS = 0.740	R	EXTREME PR/‐58%/3 mo	Nivo PD/‐/2 mo	ERBITAX PR/‐85%/17 mo	–	–	–	OS SCAI: 15 mo, OS first L: 22 mo, AWD
17.	M/55 y/OC/III/5%/SX➔BRT	65/PNRf/ECOG 1/ mCCI = 8/ACE27 = 3/COS = 0.796	L + Mx (LN)	Nivo SD/‐/13.5 mo	ERBITAX PR/−70%/14 mo	–	–	–	–	OS SCAI: 13 mo, OS first L: 27 mo, AWD
18.	F/59 y/OC/II/5%/SX➔RT	61/PNRf/ECOG 2/ mCCI = 8/ACE27 = 3/COS = 0.844	L + R + Mx (LN)	ERBITAX SD/‐20.5%/6 mo	Nivo PD/‐/3 mo	C‐wkCDDP SD/+6.9%/8 mo	–	–	–	OS SCAI: 8 mo, OS first L: 17 mo, dead
19.	M/43 y/LX/IVA/10%/SX➔RT	48/PRf/ECOG 2/ mCCI = 6/ACE27 = 3/COS = 0.824	L + R + Mx (Bone)	Anti‐PDL1 + iSTAT3 (CT)	CARBITAX PD/+21%/5 mo	Nivo PD/‐/1 mo	C‐wkCDDP ‐/‐/1 mo	–	–	OS SCAI: 7 mo, OS first L: ‐(CT), dead
20.	M/59 y/OC/IVC/60%/Anti‐PD1 (Pembro)	59/PNRf/ECOG 3/ mCCI = 8/ACE27 = 3/COS = 0.845	L + Mx (Lung)	Pembro PD/‐/1 mo	CARBITAX PR/−53.5%/6.5 mo	Nivo PD/‐/1 mo	C‐wkCDDP PD/‐/ 1 mo	–	–	OS SCAI: 9 mo, OS first L: 10 mo, dead
21.	M/74 y/UPHNC/IVA/<1%/SX➔BRT	76/PRf/ECOG 1/ mCCI = 9/ACE27 = 3/COS = 0.586	R + Mx (Bone)	EXTREME SD/‐20%/4 mo	NIvo PD/‐/1.5 mo	ERBITAX PR/‐100%/10 mo	–	–	–	OS SCAI: 9 mo, OS first L: 18 mo, AWD
22.	F/75 y/OC/IVA/80%/SX➔BRT	77/PNRf/ECOG 2/ mCCI = 9/ACE27 = 3/COS = 0.757	L + Mx (Pleura)	ERBITAX PR/‐50%/4.5 mo	Nivo PR/‐/11 mo	C‐wkCDDP PR/‐53%/3 mo	Nivo SD/‐/5 mo	C‐wkCDDP ‐/‐/ 1 mo	–	OS SCAI: 20 mo, OS first L: 43 mo, AWD
23.	M/70 y/OC/IVB/<1%/SX➔BRT	71/PNRf/ECOG 2/ mCCI = 9/ACE27 = 3/COS = 0.780	L + R + Mx (Skin)	Anti‐PDL1 + iSTAT3 (CT)	ERBITAX PR/−70%/8 mo	Pembro PD/‐/1 mo	C‐wkCDDP ‐/‐/ 3 mo	–	–	OS SCAI: 13 mo, OS first L: (CT), AWD

Blue cells represent first IO and adjacent brown cells indicate first SCAI. The table also shows that some patients were rechallenged with IO followed by salvage chemotherapy. Data on efficacy or toxicity occurring during treatment within a clinical trial are not given. In those cases it is indicated as “(CT)”.

^aPartial response in target lesions but overall progressive disease due to appearance of new lesions.

^bPDL1 expression on tumor cells.

Abbreviations: AJCC, American Joint Committee on Cancer; (–), not applicable or not available; ACE‐27, Adult Comorbidity Evaluation‐27; BRT, bioradiotherapy (radiotherapy combined with cetuximab); CARBITAX, weekly cetuximab; carboplatin and paclitaxel; CR, complete response; CT, clinical trial; C‐wkCDDP, cetuximab plus weekly cisplatin; ECOG, Eastern Cooperative Oncology Group; ERBITAX, weekly cetuximab plus paclitaxel; EXTREME, three‐weekly cisplatin‐5FU and weekly cetuximab; F, female; GCE‐COS, Generalized Competing Event Composite Omega Score; HP, hypopharynx; iCT, induction chemotherapy; LCBI, last chemotherapy before immunotherapy; L, local; LN, lymph node; Lx, larynx; M, male; Mx, metastatic; mCCI, modified Charlson Comorbidity Index; N, No. of patients; Nivo, nivolumab; NR, not reached; OC, oral cavity; OP, oropharynx; ORR, overall response rate in target and nontarget lesions; OS, overall survival; PCTL, best percentage change from baseline in target lesions; PD, progressive disease; Pembro, pembrolizumab; PFS, progression‐free survival; PNRf, platinum nonrefractory; PR, partial response; PRf, platinum‐refractory; R, regional; RT, radiotherapy; SCAI, salvage chemotherapy after immunotherapy; SD, stable disease; SX, surgery; TD, treatment duration; TPF, three‐weekly docetaxel‐platinum‐5FU; UPHNC, unknown‐primary head and neck cancer.

Objective Response to SCAI According to PD‐L1 Expression

PD‐L1 expression could be tested in 18 patients: 11 objective responders to SCAI and 7 nonresponders (Table 3). PDL1 expression ≥1% occurred in 7 or 11 (64%) responders and in 4 of 7 nonresponders (57%). PD‐L1 < 1% occurred in 4 of 11 (36%) responders and 3 of 7 (43%) of nonresponders. There were no statistically significant differences in ORR to SCAI according to PD‐L1 expression (p = .783).

Clinical Response to SCAI

At the start of SCAI, 69.5% of the patients had ECOG ≥2. After starting SCAI, ECOG improved by 1 point in 14 patients (61%). Median time to ECOG improvement was 7 weeks (min–max: 3–20).

Response During LCBI and SCAI in LCBI‐Treated Patients

Among 10 patients treated with IO in second (n = 9) or third line (n = 1), LCBI consisted of EXTREME (n = 6) and ERBITAX (n = 4), whereas in these same patients SCAI consisted of ERBITAX (n = 6), wkCDDP‐cetuximab (n = 2), or EXTREME (n = 2; Table 1).

ORR to LCBI was 40%, whereas ORR to SCAI achieved 60% (if considering ORR in TL only, it achieved 70%, as one patient showed response in TL with appearance of new metastatic lesions). Two of the patients received the same regimens as LCBI and SCAI. In one of these patients treated with ERBITAX pre‐ and post‐IO, ORR was PR both as LCBI (−81.5%) and as SCAI (−30.9%; Figs. 1, 2; Tables 2, 3).

Progression‐Free Survival and Overall Survival

In the whole population, after a median follow‐up of 12 months (range, 4–38), median OS since SCAI was 12 months (95% confidence interval [CI], 9–not reached [NR]) and PFS was 6 months (95% CI, 5–8). Median OS since first line was 28 months (95% CI, 23–NR). OS since first‐line was similar between patients receiving ICIs in first line versus second line (27 vs. 28 months, p = .94).

In LCBI‐treated patients, median PFS during LCBI was 8 months (95% CI, 4–NR) and during SCAI was 7 months (95% CI, 4–NR). Median OS with SCAI in LCBI‐treated patients achieved 12 months (95% CI, 6–NR; Fig. 1, Table 2).

Objective Response Rate and Treatment Compliance During PD(L)1 Inhibitor Treatment and Time Lapse Between PD(L)1 Inhibitors and SCAI

ORR to PD(L)1 inhibitors in the total population was 17%. Median treatment duration during IO was 3 months (0.5–24). Median number of IO doses was 5 (1–30). Median time from last dose of PD(L)1 inhibitors to SCAI was 16 days (range, 7–75). All patients discontinued PD(L)1 inhibitors and started SCAI because of radiological progression with symptomatic and ECOG performance status deterioration.

Median number of PD‐1 inhibitor doses in patients treated outside clinical trial (all treated with nivolumab) was 6 (2–26). Median treatment duration in these latter group was 3 months (1–13.5).

Treatment Compliance During SCAI

Total Population

In the whole series, 82.6% and 52.2% of the patients had a reduced dose intensity of the chemotherapy and cetuximab components, respectively. Median treatment delay of chemotherapy was 2 weeks (min to max: 0–10) and median dose reductions was 1 (min to max: 0–2). Single‐agent maintenance cetuximab was started in 7 of 23 patients (30.4%; Table 4; supplemental online Table 1).

Table 4.

Summary of treatment compliance and toxicity during cetuximab‐based salvage chemotherapy after immunotherapy (SCAI)

A. Treatment compliance
Regimen and agent	Chemotherapy + cetuximab				Maintenance cetuximab
	Median No. of cycles	Median No. weeks delay	Median dose reductions	Median dose/cycle	Median No. of cycles	Median No. weeks delay	Median dose reductions	Median dose quantity/cycle
ERBITAX (n = 17)
Paclitaxel	12 (Min–max: 4–38)	2 (Min–max: 0–10)	1 (Min–max: 0–2)	75 mg/m2/qwk (Min–max: 50–80)	‐	‐	‐	‐
Cetuximab	15 (Min–max: 4–38)	1 (Min–max: 0–10)	0 (Min–max: 0–2)	Load: 400 mg/m2; follow: 225 mg/m2/qwk (Min–max: 221–250)	14 (Min–max: 7–30)	0 (Min–max: 0–3)	0 (Min–max: 0–1)	500 mg/m2/q2wk (Min–max: 300–500)
EXTREME (n = 2)
Cisplatin	3.5	1.5	1	70 mg/m2/q3wk	‐	‐	‐	‐
5FU	3.5	1.5	1	3,378 mg/m2/q3wk	‐	‐	‐	‐
Cetuximab	3.5	1.5	1	Load: 400 mg/m2; follow (mean): 237 mg/m2/qwk	‐	‐	‐	‐
CARBITAX (n = 2)
Carboplatin	16.5	5	0	100.5 mg/m2/qwk	‐	‐	‐	‐
Paclitaxel	18	3.5	0	74 mg/m2/qwk	‐	‐	‐	‐
Cetuximab	21.5	0	0	Load: 400 mg/m2; Follow (mean): 250 mg/m2/qwk	‐	‐	‐	‐
wkCDDP‐Cetuximab (n = 2)
wkCDDP	11	4.5	2	39 mg/m2/qwk	‐	‐	‐	‐
Cetuximab	13	3.5	0	Load: 400 mg/m2; follow (mean): 237 mg/m2/qwk	‐	‐	‐	‐

B. Toxicity (ERBITAX)
SCAI	ERBITAX (n = 17), n (%)
Event Grade (CTCAE v4.0)	G1–2	G1	G2	G3–4	G3	G4
Event	17/17 (100)	17/17 (100)	17/17 (100)	11/17 (64.7)	11/17 (64.7)	1/17 (6)
General
Asthenia	(100)	7/17 (41)	10/17 (59)	‐	‐	‐
Nutrition and metabolic disorders
Decreased appetite	(88.5)	11/17 (65)	4/17 (23.5)	‐	‐	‐
Hypomagnesemia	(47)	2/17 (12)	6/17 (35)	6	1/17 (6)	‐
Gastro‐intestinal
Nausea	(30)	3/17 (18)	2/17 (12)	‐	‐	‐
Diarrhea	(6)	‐	1/17 (6)	‐	‐	‐
Mucositis	(24)	1/17 (6)	3/17 (18)	‐	‐	‐
Skin
Acneiform rash	(76.5)	4/17 (23.5)	9/17 (53)	23.5	4/17 (23.5)	‐
Hand‐foot fissures	(59)	4/17 (23.5)	6/17 (35)	‐	‐	‐
Onycolisis		‐	10/17 (59)	‐	‐	‐
Onycodystrophy		‐	6/17 (35)	‐	‐	‐
Paronichia	(18)	‐	3/17 (18)	6	1/17 (6)	‐
Eye disorders
Conjunctivitis	(12)	‐	2/17 (12)	‐	1/17 (6)	‐
Ectropion	(6)	‐	1/17 (6)	‐	‐	‐
Infections
URT	(12)	‐	2/17 (12)	‐	‐	‐
Pneumonia	‐	‐	‐	23.5	4/17 (23.5)	‐
Diverticulitis	‐	‐	‐	6	1/17 (6)	‐
Soft tissue	(18)	‐	3/17 (18)	‐	‐	‐
Blood and lymphatic system disorders
Anemia	(23.5)	1/17 (6)	3/17 (18)	12	2/17 (12)	‐
Neutropenia	(29.4)	‐	5/17 (29.4)	12	1/17 (6)	1/17 (6)
Lymphopenia	(23.5)	‐	4/17 (23.5)	12	2/17 (12)	‐
Thrombopenia	(23.5)	2/17 (12)	2/17 (12)	6	1/17 (6)	‐
Nervous system disorders
Peripheral neuropathy	(41.5)	3/17 (18)	4/17 (23.5)	‐	‐	‐
Vascular disorders
Tumor bleeding	(18)	‐	3/17 (18)	12	2/17 (12)	‐

C. Toxicity (Other regimens)
Event Grade (CTCAE v4.0)	G1–2	G1	G2	G3–4	G3	G4
No. (%)	6/6 (100)	5/6 (83)	6/6 (100)	4/6 (67)	4/6 (67)	‐
SCAI (Summary of events for Other regimens)	G1		G2		G3	G4
Event type	Asthenia (1), decreased appetite (1), nausea (1), acneiform rash (1)		Asthenia (1), decreased appetite (1), acneiform rash (1), lymphopenia (1), onycolisis (1), onycodystrophy (1), hand‐foot fissures (1), thrombopenia (1)		Neutropenia (1), anemia (1), hemoptisis (lung tumor bleeding) (1)	‐
wkCDDP‐Cetuximab (n = 2)
Event type	Acneiform rash (1), hand‐foot fissures (1)		Nausea (1)		Acneiform rash (1), hand‐foot fissures (1)	‐
EXTREME (n = 2)
Event Type	Acneiform rash (1)		Asthenia (2), decreased appetite (2), acneiform rash (1), lymphopenia (1), nausea (2), onycolisis (1), onycodystrophy (1), hand‐foot fissures (1), peripheral neuropathy (1)		Pneumonia (1)	‐

(‐): not applicable. Data for ERBITAX cohort are presented as medians (range: min–max). Event grading according to CTCAE version 4.0.

Data for the remnant patients are presented as means due to the low number of cases.

Abbreviations: CARBITAX, weekly cetuximab; paclitaxel and carboplatin; CTCAE, Common Terminology Criteria for Adverse Events; ERBITAX, weekly cetuximab plus paclitaxel; EXTREME, 3‐weekly platinum‐5FU plus weekly cetuximab; q3wk, every 3 weeks; qwk, every week; SCAI, salvage chemotherapy after immunotherapy; wkCDDP‐Cetuximab, weekly cisplatin plus cetuximab.

Table 5.

Summary of studies and case reports on salvage chemotherapy after immunotherapy (SCAI) in squamous cell cancer of the head and neck

Studies
Author (year)	n/baseline characteristics	LCBI			IO				SCAI
		Agents	ORR	PFS/OS	Agents		ORR	PFS/OS	Agents/ECOG	ORR	PFS/OS
Saleh et al. (2019) [18]	n = 82	≥1 line pre‐IO: 76%; median No. lines: 3 (2–8); median No. lines pre‐SCAI: 2 (1–6); LCBI: platinum (61%), taxane (11%), other (4%)	ORR (CR/PR): 38%	–	Anti‐PD1: 30; anti‐PDL1: 5; anti‐CTLA4: 2; anti‐PDL1 + anti‐CTLA4: 16; anti‐PD1 + anti‐KIR: 29		ORR (PR): 13.6%	–	Agents: IO in first L: 40%, IO in second L: 27%; ECOG: 0: 14%, 1: 59%, 2–3:27%	CR/PR: 30% (3CR, 22 PR), DCR: 57%; CR/PR SCAI ± cetuximab: 53% vs. 25% (p = 0.024)	PFS: 3.6 mo (95% CI, 2.6–5.1); OS: 7.8 mo (95% CI, 5.3–10.8); PFS IO first vs. second L: 5.2 vs. 3.4 mo, p = .14); OS IO first vs. second L: 12 vs. 7.6 mo, p = .36)
Pestana et al. (2019) [19]	n = 43; median age Dx: 58 (23–83); median No. lines: 3 (2–8)	LCBI agent: platinum (61%), taxane (11%), other (4%); lines pre‐IO: ≥1 line pre‐IO: 76%; agents pre‐IO: platinum prior to IO: 90.7%, platinum within 6‐mo of IO: 60.5%; Anti‐EGFR prior to IO: 34.8%, anti‐EGFR within 6‐m of iO: 27.9%	‐ORR: 47% (7/15)	‐PFS: 3.89 m (95%CI 2.56–6.97)	Agent: pembro: 67.5%; nivo: 32.5%; IO line: IO first L: 58.1%; IO second L: 39.5%; IO third L: 2.3%		ORR (PR): 20.9%. No difference in ORR based on prior anti‐EGFR or not (p = 0.24)	PFS: 2.79 m (95%CI: 2.23–4.11); PFS (OP): HR 0.34 (95%CI: 0.18–0.67); No survival difference between first L IO vs. CT ± anti‐EGFR	Agent: anti‐EGFR alone: 37.2%, single‐agent CT: 32.5%, anti‐EGFR + CT: 18.6%, CT + other: 11.6%; median No. l lines pre‐SCAI: 2 (1–6); ECOG: 0: 14%, 1: 59%, 2–3: 27%	ORR Anti‐EGFR alone: 37.5% (similar to ORR with CT); Response to SCAI associated with prior response to IO	PFS: 4.24 m (95%CI 2.63–5.19); OS: 8.41 m (95%CI 7.62–11.07); No difference between single‐agent cetuximab and CT in PFS (p = .78) and OS (p = .22); OS from first L: 16.5 m (95%CI: 13.8–28.9; p = 0.16)
Aspeslagh et al. (2017) [17]	N = 118 (n = 13 HNC) median age Dx: 54 (20–78)	–	RR: 31%	PFS: 4.1 mo	51/118: anti‐PD(L)1; 67/118: IO w/o anti‐PD(L)1		–	–	–	ORR post‐IO: 8%	PFS pre‐IO: 4.1 mo; PFS post‐IO: 3.6 mo; ORR pre‐IO: 31%; ORR post‐IO: 8%
Harrington et al. (KN‐048 ASCO 2020) [8]	n = 882	–	–	–	Pembro (n = 301) vs. EXTREME (n = 300)	CPS ≥20	23.3% vs. 36.1%	PFS: 3.4 vs. 5.3; OS: 14.8 vs. 10.7	Pembro arm (49.2%): CT: 44.9%, EGFRi: 19.6%, ICIs: 2%, Other: 1.3%	–	Pembro arm PFS2a longer in:CPS ≥20: 11.7 vs. 9.4; CPS ≥1: 9.4 vs. 8.8
						CPS 1–19	14.5% vs. 33.8%	PFS:2.2 vs. 4.9; OS: 10.8 vs. 10.1
						TP	16.9% vs. 36%	PFS: 2.3 vs. 5.2; OS: 11.5 vs. 10.7	Pembro+CT arm (40.9%): CT: 31.3%, EGFRi: 13.2%, ICIs:4.3%, other: 2.9%	–
					Pembro + CT (n = 281) vs. EXTREME (n = 278)	CPS ≥20	42.9% vs. 38.2%	PFS: 5.8 vs. 5.3; OS: 14.7 vs. 11			Pembro + CT arm PFS2a longer in: CPS ≥20: 11.3 vs. 9.7; CPS ≥1: 10.3 vs. 8.9; TP: 10.3 vs. 9
						CPS 1–19	29.3% vs. 33.6%	PFS: 4.9 vs. 4.9; OS:12.7 vs. 9.9	EXTREME arm (53%): CT: 34%, EGFRi: 6.3%, ICIs: 16.7%, other: 3%	–
						TP	35.6 vs. 36.3%	PFS: 4.9 vs. 5.1; OS: 13 vs. 10.7
Current series	n = 23	N = 10, EXTREME: 6, ERBITAX: 4	ORR:40%	PFS: 6 m (IC95% 3‐NR)	Anti‐PD1: 52.1%; anti‐PDL1: 13%; anti‐PDL1 + iSTAT3: 30.4%; anti‐PDL1 + iCXCR2: 4.3%		–	–	n = 23; ERBITAX: 17; EXTREME: 2; CARBITAX: 2; wkCDDP‐Cetuximab: 2	ORR (all): 56.5% (13/23); ORR (LCBI pretreated): 60% (6/10)	PFS: 6 mo (95% CI, 5–8); OS: 12 mo (95% CI, 9‐NR)

Case Reports
Author (year)	N/Location/Stage/Tx IDx	LCBI			First IO				First SCAI	Second IO	Second SCAI
Daste et al. (2017) [20]	64 y/OC/ local relapse	EXTREME × 6 cycles ➔ PR ➔ Cetuximab × 12 mo			Anti‐PD1 (Nivolumab) × 6 ➔ PD ➔ × 5 ➔PD				CBDCA AUC5 d1 + Pac d1, 8, 15 × 2 cycles ➔ major PR	–	–
Daste et al. (2018) [21]	‐/HP/IVA/TPF ➔BRT	Relapsed after 6 mo ➔ EXTREME × 6 cycles ➔ PR ➔ PD ➔ EXTREME ➔ SD			Anti‐PD1 (Nivolumab) × 4 ➔ PD				wkPac × 3 m ➔ major PR	–	–
Daste et al. (2018) [21]	62 y/L/IVA/BRT ➔ SX	C‐PF × 6 wk ➔ PR ➔EXTREME × 12 wk ➔ PD ➔ Cetuximab × 6 m ➔ PD; PR			Anti‐PD1 (Nivolumab) × 4 ➔ PD				wkPac x 3 m ➔ major PR	–	–
Larroquette et al. (2018) [22]	52 y/MS/IVA/SX ➔ RT + CDDP	–			LR relapse 3 mo after POCRT ➔ Anti‐PD1 (CT) × 4 (2 mo) ➔ PD				EXTREME x 6 cycles ➔ major PR	–	–
Lasserre et al. (2019) [23]	51 y/HP/IVC at IDx	–			Anti‐PD(L)1 + Anti‐CTLA4 (CT) ➔ SD ➔ PD				EXTREME × 6 cycles ➔ major PR	Anti‐PD1 (Nivolumab) × 3 ➔ PD	CBDCA‐Pacli × 3 cycles ➔ PR (ongoing after 3 additional cycles)

Blue cells represent IO lines and adjacent brown cells indicate either LCBI or SCAI.

^aPFS2 indicates PFS calculated from initial randomization in KEYNOTE‐048 (from the start of first‐line therapy).

Abbreviations: (–), not applicable; AUC, area under the concentration curve; BRT, bioradiotherapy; CBDCA, carboplatin; CT, clinical trial; ECOG, Eastern Cooperative Oncology Group; HP, hypopharynx; IDX, initial diagnosis; IO, immunotherapy; LCBI, last chemotherapy before immunotherapy; MS, maxillary sinus; NR, not reached; OC, oral cavity; OS, overall survival; PD, progressive disease; PR, partial response; PFS, progression‐free survival; RT, radiotherapy; SCAI, salvage chemotherapy after immunotherapy; SD, stable disease; SX, surgery.

Erbitax

Among 17 patients treated with ERBITAX, median number of paclitaxel and cetuximab weekly cycles were 12 and 15, and median dose per cycle was 75 mg/m² and 400 mg/m² (load) and 225 mg/m² (follow), respectively. Median number of maintenance cetuximab was 14, with a median dose per 2‐weekly cycles of 500 mg/m² (Table 4; supplemental online Table 1).

Other Regimens

Among six patients treated with other regimens, mean doses of chemotherapy were also reduced to a variable extent depending on the regimen used.

Table 4 summarizes data on dose intensity and compliance. Supplemental online Table 1 shows data for each individual patient.

Toxicity During SCAI

In the whole series of 23 patients, grade 1 or 2 AEs occurred in 100%. Grade 3 or 4 AEs developed in 65%, being grade 3 in all of them except in one patient (grade 4 neutropenia). There were no treatment‐related deaths.

Among patients treated with ERBITAX (n = 17), AEs of grades 1 or 2 occurred in 100% of the patients, the most common being asthenia, decreased appetite, and acneiform rash. Grade 3 or 4 AEs occurred in 64.7% of the patients, the most common being acneiform rash (23.5%), pneumonia (23.5%), anemia, neutropenia, lymphopenia, and tumor bleeding (12% each). One patient experienced a grade 4 AE (neutropenia).

Among patients treated with other regimens of SCAI (n = 6), all patients experienced grade 1 or 2 AEs, and four (67%) patients suffered grade 3 AEs. A patient treated with CARBITAX suffered grade 3 hemoptysis from a lung metastasis bleeding. There were no grade 4 AEs.

Therefore, grade 3 bleeding occurred in 3 (ERBITAX [n = 2], CARBITAX [n = 1]) of the 23 patients (13%). In the two patients treated with ERBITAX, both with bulky and friable tumors, bleeding was judged to be due to rapid tumor reduction.

Table 4 and supplemental online Table 2 in the supplementary appendix summarize adverse events during SCAI in the whole population and in each individual patient, respectively.

Case Reports

Case A (Patient 15)

A 69‐year‐old man diagnosed with an oral cavity primary suffered a local and distant (lung, liver, lymph node, and bone) early relapse only 2 months after ending 3‐weekly platinum‐based CRT. He received two doses of nivolumab with rapid radiological and symptomatic progression (intense laterocervical pain requiring opioids) suggesting hyperprogressive disease. Treatment with weekly ERBITAX was started 3 weeks after the last dose of nivolumab, with an overall PR after seven doses and a maximum reduction in TL of 62%. After 8 months, the patient progressed and, because of the limited therapeutic options, he was rechallenged with nivolumab for 3 months with PD as best response, after which ERBITAX was reintroduced with SD as best response and treatment was maintained for 8.5 months (Fig. 2A; Table 3) .

Figure 2.

Examples of patients treated with SCAI. Each case (A‐E) corresponds to each of the cases described in the main text.Abbreviations: CMR, complete metabolic response; CR, complete response; CRT, chemoradiotherapy; CT, chemotherapy; ECOG PS, Eastern Cooperative Oncology Group Performance Status; PD, progressive disease; PR, partial response; RT, radiotherapy; SCAI, cetuximab‐based salvage chemotherapy after immunotherapy; SCCHN, squamous cell cancer of the head and neck; SD, stable disease; TL, target lesions.

Case B (Patient 6)

A 68‐year‐old man initially diagnosed with a large laryngeal mass, bilateral neck lymphadenopathies, and lung metastases started treatment with a PD‐L1 inhibitor combined with a STAT3 inhibitor within a clinical trial. Although pseudoprogression (PsPD) cannot be disregarded, the patient suffered an overt radiological and clinical progression (ECOG deterioration, intense pain in the neck, and grade 2 asthenia), and thereby he started treatment with weekly ERBITAX. He received 10 doses, achieving major locoregional and pulmonary responses. However, it was considered as PD because of appearance of new hepatic and bone metastases (Fig. 2B; Table 3).

Case C (Patient 14)

A 54‐year‐old man was diagnosed with stage IVA laryngeal cancer. Six months after laryngectomy and adjuvant RT, he suffered a pulmonary and hepatic relapse. Treatment with weekly ERBITAX (with paclitaxel reduced to 50 mg/m² because of hypersplenism in the context of alcoholic liver disease) achieved PR with a maximum PCTL of −81.5%. After 6 months, the patient suffered PD and started nivolumab, with a PR as best response. After 11 months of nivolumab, the patient suffered PD. Weekly ERBITAX was reintroduced (maintaining paclitaxel reduced to 50 mg/m²) achieving a PR (−30.9%) as best response, eventually starting maintenance cetuximab achieving 20 months under this line of therapy. After progression, he restarted treatment with nivolumab (Figure 2C; Table 3).

Case D (Patient 4)

A 70‐year‐old woman received induction CT with weekly ERBITAX for a stage IVB oral cavity cancer. After radical concurrent RT plus cetuximab, the patient suffered a local relapse, starting treatment with a PD‐L1 inhibitor within a clinical trial. After an overt local and distant radiological and clinical progression (intense local pain, macroglossia, grade 2 asthenia, and ECOG deterioration), weekly ERBITAX was started (with paclitaxel reduced to 70 mg/m² because of ECOG 2), achieving a major local response and rapid symptomatic improvement (pain disappeared) but with hepatic progression (PD as overall response). Because of the symptomatic improvement treatment with ERBITAX was maintained for 9 months (Fig. 2D; Table 3).

Case E (Patient 21)

A 74‐year‐old man diagnosed with an SCCHN of unknown primary relapsed 18 months after surgery and adjuvant RT. Treatment with EXTREME regimen was started with PD after two cycles. Nivolumab was started, but the patient suffered PD after three doses (growth of right neck mass, appearance of C5–C6 bone metastases, and right neck intense pain, which made PsPD less likely). Three weeks after the last dose of nivolumab, ERBITAX was started. After four doses of weekly ERBITAX (second dose without paclitaxel because of G2 neutropenia) a CR with complete metabolic response (CMR) was achieved. The patient continued with maintenance 2‐weekly cetuximab for 14 cycles, achieving a sustained CR and CMR (Fig. 2E; Table 3).

Discussion

In this retrospective series of 23 patients with R/M SCCHN, in which 56.5% had received IO in the first line, 30.4% were platinum‐refractory, and 74% received weekly ERBITAX as SCAI, we show a 53% ORR to salvage cetuximab‐based chemotherapy after an overt clinical and radiological progression to PD(L)1 inhibitors. This 53% ORR to SCAI in second or third line, is notable, considering that cetuximab‐based 3‐weekly regimens such as EXTREME and TPEx, or weekly regimens like ERBITAX, achieve ORR in the first‐line setting of 36%, 46% and 55%, respectively [27, 28, 29]. In a French series of 82 heavily pretreated patients with R/M SCCHN, in which 76% had received at least one line of chemotherapy pre‐IO and a majority were platinum‐refractory, Saleh et al. [18], found a 30% ORR to SCAI, which raised to 53% with cetuximab‐based chemotherapy. Likewise, in a series from the U.S. of 43 patients with R/M SCCHN, in which 58% had received first‐line IO and 60% were platinum‐refractory, Pestana et al. [19], found a 42% ORR to SCAI and, interestingly, a 37% ORR with single‐agent cetuximab. In our series, SCAI consisted, in all patients, of cetuximab‐based chemotherapy, and thus, efficacy could not be compared with a control group receiving chemotherapy alone. However, ORR was comparable to that of patients treated with cetuximab‐based regimens in the French series (ORR, cetuximab‐based vs. noncetuximab‐based: 53% vs. 25%) [18].

Our results and those from the other series suggest a possible additive effect between IO and subsequent SCAI in terms of ORR, which may also be illustrated by comparing responses to chemotherapy before and after IO. In our study, ORR and PFS with LCBI were 40% and 8 months, whereas ORR and PFS with SCAI in LCBI‐treated patients achieved 60% and 7 months, respectively. These results agree with those reported by other authors in patients with lung cancer. Schvartsman et al. [9], in a series of 28 patients with non‐small cell lung cancer (NSCLC), reported an ORR and PFS to LCBI (platinum‐doublet) of 37% and 7.3 months, whereas ORR and PFS to single‐agent SCAI were 39.3% and 4.7 months, respectively. Likewise, in another series of 78 patients with NSCLC by Park et al. [10], ORR with SCAI was higher than with LCBI (53.4% vs. 34.9%), and PFS with LCBI and SCAI were similar (4.2 vs. 4.7 months, p = .84).

Despite the few existing data regarding the depth of responses with chemotherapy in SCCHN in patients both pretreated and non‐pretreated with IO, responses to SCAI in our series showed a median PCTL of −57.5% (range, −30 to −100), which is deeper than the median PCTL of −40% reported by Enokida et al. [30], in a Japanese series of 23 platinum‐refractory patients receiving first‐line ERBITAX, the most frequently used SCAI in our series. Although only 30.4% of the patients in our series were platinum‐refractory, all, per definition, were more pretreated because they received SCAI as second or further line.

Interestingly, we found that ORR to SCAI was significantly higher in patients pretreated with PD‐1 than with PD‐L1 inhibitors. Although potential differences in efficacy between PD1 and PDL1 inhibitors cannot be disregarded in the absence of head‐to‐head comparisons, the differences in ORR observed in our series may probably be explained because all patients receiving PDL1 inhibitors were treated within clinical trials, which usually constitute a highly selected population based on strict inclusion and exclusion criteria and are subjected to a generally more intensive surveillance [2, 3, 4, 31, 32]. This might lead to a longer time on treatment and to the emergence of tumor‐cell resistance mechanisms as well as of an immunosuppressive “milieu” that renders the tumor less responsive to subsequent therapies [33]. Indeed, in our series, patients treated with PD‐1 inhibitors outside clinical trials (most treated with nivolumab) received a median of six cycles (range, 2–26), which indicates a relatively early switch to SCAI, increasing the likelihood that PsPD may go undetected. However, PsPD is infrequent in SCCHN (only 4%–8% of patients achieving a partial response) [34]. In contrast, median time to the start of SCAI in the whole population was only 16 days, a rather short time in which, because of their long half‐life, most PD(L)1 inhibitors may still be rendering direct therapeutic effects: median half‐life ranges from 17 to 27 days [35, 36, 37]. Finally, data from CheckMate‐141 showed that patients progressing to nivolumab without ECOG deterioration and who were nonrapid progressors treated beyond progression may experience a longer survival and in some cases even late responses, suggesting that beneficial effects of IO may persist long time, and thereby explaining some of the responses seen with subsequent treatments [38]. However, all of our patients were switched to SCAI because of a radiological progression with ECOG deterioration, suggesting a true progression and rendering the possibility of PsPD less likely [34]. Although at the start of SCAI, nearly 70% of our patients had ECOG ≥2, most patients showed an improvement in the performance status during SCAI because of clinical benefit. In addition, comorbidity assessment using GCE‐COS at the start of SCAI revealed that 87% of patients had a score that justified to start antitumor treatment [24].

Although in the French series, patients treated with first‐line IO showed a higher ORR to SCAI than those treated in second line (40% vs. 27%), we found no differences in our series between those treated with first‐ versus further lines (58.3% vs. 54.5%). This may be explained because our patients were less heavily pretreated and thus more prone to respond to SCAI.

Finally, although limited by the reduced sample size, PD‐L1 expression in tumor cells (<1 vs. ≥1) was not different between responders and nonresponders. Although the same immunohistochemistry assay was used in all patients, the interval between the biopsy sample and the start of SCAI, and not testing the CPS, a more reliable biomarker in SCCHN, may also limit any conclusions.

In our series, responses and survival were durable, with a median PFS and OS since the start of SCAI of 6 and 12 months, respectively, which are notable considering that patients had received, per definition, at least one line of systemic therapy before SCAI. In the French series, median PFS and OS were 3.6 and 7.8 months, and in the U.S. series, they were 4.2 and 8.4 months, respectively [18, 19]. Although both PFS and OS with SCAI were longer in our series, this may be explained because our patients were less heavily pretreated and a majority received additional lines of therapy after SCAI, with up to 14 patients being rechallenged with IO and 10 of these being treated with a second cetuximab‐based SCAI. In contrast, in the KEYNOTE‐048 study, progression‐free survival 2 (PFS2), calculated from initial randomization, in the intention‐to‐treat population was significantly longer in the pembrolizumab arm, with a median PFS2 of 11.7 and 9 months for CPS ≥20 and CPS ≥1, respectively, as well as in the pembrolizumab plus chemotherapy arm, with a median PFS2 of 11.3, 10.3, and 10.3 months for CPS ≥20, CPS ≥1, and total populations, respectively. These results are even more relevant if we consider that only around half of the patients in KEYNOTE‐048 received second‐line therapy [8]. In contrast with the latter study, in our series, PFS was calculated from the start of SCAI and not from the start of IO and, therefore, is not readily comparable. However, and considering all the limitations of our small retrospective series, the 6‐month median PFS achieved in our study is notable considering that with first‐line regimens such as EXTREME, TPEx, and ERBITAX, median PFS was 5.6 months, 6 months, and 4.2 months, respectively [27, 28, 29]. In KEYNOTE‐048, median OS in the CPS ≥20 population was 14.8 and 14.7 months in the pembrolizumab and pembrolizumab plus chemotherapy arms, respectively. Interestingly, in our series, median OS since first‐line treatment achieved 28 months, with no differences in first‐line OS between those receiving IO in first‐line versus second‐line. Although our results cannot be reliably compared with those from a large phase III trial, the fact that only half of the patients in KEYNOTE‐048 received second‐line therapy after IO, whereas in our series, the median number of treatment lines was four, with up to 60% of our patients receiving additional therapies after SCAI, may explain the prolonged survival times achieved. In contrast, in TPExtreme phase III trial, comparing TPEx with EXTREME in the first‐line setting, survival was longer for patients treated with IO as second‐line compared with those treated with chemotherapy, and it was also longer with IO in second‐line in patients from the TPEx arm compared with the EXTREME arm (21.9 vs. 19.4 months) [39]. Results from the mentioned studies and from our series open the question of whether not only chemoimmunotherapy combinatorial strategies but also a sequential approach of IO followed by chemotherapy or vice versa, and even reintroducing a second IO followed by a second SCAI, may be worth exploring in future clinical trials. In this regard, some trials are studying if reintroducing chemotherapy after progression on immunotherapy and then returning to immunotherapy is a feasible approach [NCT03083808].

It must be noted that the high ORR and survival in our series occurred despite a large majority of patients receiving weekly chemotherapy regimens with doses reduced because of poor performance status or because of advanced age or comorbidities. According to these observation, reducing dose intensity of chemotherapy could be explored in future chemoimmunotherapy trials as a strategy to reduce toxicity while maintaining efficacy.

Several aspects may constitute the biological rationale to explain why IO may favor responses to SCAI in patients with SCCHN. SCCHN is characterized by a highly immunosuppressive tumor microenvironment (TME) mainly composed of myeloid‐derived suppressor cells and T regulatory cells that impede tumor eradication by the immune system and explain the appearance of acquired resistance to ICIs. Besides direct tumor cell killing, chemotherapies like platinums or taxanes may favor tumor‐antigen presentation and reduce the immunosuppressive component of the TME, allowing to emerge a more effective antitumor immune response [40, 41, 42]. Although in our series, during SCAI, some patients received either cisplatin or carboplatin combinations, most patients received ERBITAX, a weekly regimen consisting of paclitaxel and cetuximab [29]. Besides the direct antitumor effects of paclitaxel, the latter is also known for behaving as a toll‐like receptor agonist capable of reprogramming immunosuppressive macrophages toward an immune‐effector phenotype [40, 43, 44]. Another possible contributor to the high ORR with SCAI seen in our study is the use of cetuximab, which is known not only for targeting EGFR but for being able to induce antibody‐dependent cellular cytotoxicity (ADCC) [43, 45, 46]. ADCC is mediated by natural killer (NK) cells, and SCCHN is characterized by a deep infiltration by these cells [47]. In contrast, it is also possible that treatment with ICIs before SCAI could enhance the ADCC effect of cetuximab [43]. Recently, the combination of cetuximab plus monalizumab, an NK cell inhibitory‐checkpoint inhibitor, showed a 27.5% ORR and a median OS of 10.3 months [48]. In addition, the combination of pembrolizumab plus cetuximab showed promising activity in patients with R/M SCCHN, achieving a 45% ORR and a median OS of 18.4 months [49].

Although with limitations due to the retrospective nature of our study, toxicity during SCAI was comparable to that already known for the regimens used. It must be noted, however, that three patients suffered grade 3 tumor bleeding, which was effectively managed with conservative measures, and that in two of the patients, both with a high tumor burden, bleeding was considered to be associated with the rapid tumor reduction observed with SCAI. Therefore, patients should be closely followed at least during the first cycles of treatment, especially in case of large, friable tumors or tumors located near large vessels that may render them more prone for such complications.

In our study, we show that benefit from IO may occur “after” treatment with ICIs, possibly conditioning response rates and survival with subsequent lines of therapy. These findings are relevant considering the importance of achieving high response rates and deep tumor reductions for improving patients’ symptoms and quality of life and possibly prolonging survival.

Although, to our knowledge, this is currently the fourth largest study evaluating the efficacy and the first evaluating toxicity and treatment compliance of salvage chemotherapy after PD(L)1 inhibitors in head and neck cancer, we must acknowledge several limitations. Our results are limited by the retrospective nature and reduced sample size and by the heterogeneity of our population, as some patients were treated with IO in first line whereas others in second or further lines, and different immunotherapy drugs and combinations (PD‐1, PD‐L1, CTLA4, STAT3, CXCR2 inhibitors) as well as different regimens of SCAI were used. In addition, because of the retrospective nature, patients in our series may be exposed to selection bias, because patients able to receive SCAI are highly selected and possibly selected to do better. As already mentioned, early switch to SCAI since the start of IO and a short interval between last IO doses and SCAI could have conditioned the response rate observed during SCAI. In this regard, comparing pre‐ and post‐SCAI tumor biopsies could help to understand the immunobiological mechanisms underlying responses to SCAI. Unfortunately, pre‐ and post‐SCAI biopsies were not performed because of the retrospective nature of our study. Finally, patients received different systemic therapies after SCAI that may have impacted overall survival results.

Conclusion

Our study describes a higher than expected response rate to cetuximab‐based salvage chemotherapy after progression to IO in patients with R/M SCCHN. In addition, response rates were higher and PFS was maintained during SCAI when compared with LCBI. These results should encourage the design of trials to elucidate the best treatment sequence for patients with head and neck cancer in the IO era, until better, rationally based IO combinations are found.

Author Contributions

Conception/design: Santiago Cabezas‐Camarero

Provision of study material or patients: Santiago Cabezas‐Camarero, María Nieves Cabrera‐Martín, Salomé Merino‐Menéndez, Mateo Paz‐Cabezas, Maricruz Iglesias‐Moreno, Almudena Alonso‐Ovies, Vanesa García‐Barberán, Pedro Pérez‐Segura

Collection and/or assembly of data: Santiago Cabezas‐Camarero, María Nieves Cabrera‐Martín, Salomé Merino‐Menéndez, Mateo Paz‐Cabezas

Data analysis and interpretation: Santiago Cabezas‐Camarero, María Nieves Cabrera‐Martín, Salomé Merino‐Menéndez, Mateo Paz‐Cabezas, Vanesa García‐Barberán, Pedro Pérez‐Segura

Manuscript writing: Santiago Cabezas‐Camarero

Final approval of manuscript: Santiago Cabezas‐Camarero, María Nieves Cabrera‐Martín, Salomé Merino‐Menéndez, Mateo Paz‐Cabezas, Maricruz Iglesias‐Moreno, Almudena Alonso‐Ovies, Vanesa García‐Barberán, Pedro Pérez‐Segura

Disclosures

Santiago Cabezas‐Camarero: Bristol Myers Squibb, Sanofi (C/A), Bristol Myers Squibb, AstraZeneca (ET), Bristol Myers Squibb, Merck KGaA, Abbvie (Other [Scientific meeting travel arrangement]). The other authors indicated no financial relationships.

(C/A) Consulting/advisory relationship; (RF) Research funding; (E) Employment; (ET) Expert testimony; (H) Honoraria received; (OI) Ownership interests; (IP) Intellectual property rights/inventor/patent holder; (SAB) Scientific advisory board

Supporting information

See http://www.TheOncologist.com for supplemental material available online.

Table S1 Treatment compliance during salvage cetuximab‐based chemotherapy after immunotherapy (SCAI) for each individual patient.

Table S2. Adverse events during treatment with salvage cetuximab‐based chemotherapy after immunotherapy (SCAI) for each individual patient.