Biomarkers of neoadjuvant immunotherapy for locally advanced head and neck cancer: A systematic review and meta-analysis

Abstract

Background:

Despite the general advantages of neoadjuvant immunotherapy for locally advanced head and neck squamous cell carcinoma (LAHNSCC), many patients do not achieve successful clinical outcomes. Hitherto, there has been a lack of biomarkers to predict the effect of neoadjuvant immunotherapy in LAHNSCC.

Methods:

We systematically searched PubMed, Embase, Web of Science, and Cochrane databases until June 2024 for studies related to neoadjuvant immunotherapy for LAHNSCC. We calculated the pathological response rates after neoadjuvant immunotherapy for LAHNSCC according to biomarkers. Statistical analysis included correlation analysis with Spearman coefficient and the Mann–Whitney U test.

Results:

This meta-analysis included 26 studies with a total of 980 patients enrolled. The results showed a correlation between pretreatment combined positive score (CPS) values and posttreatment pathological responses (Z = 0.289, P < .001). Patients with CPS above the cutoff of 5 had significantly higher posttreatment overall pathological response rates than those with CPS below the cutoff (relative risk [RR] = 1.852, P = .012). Similar results have been found in patients with CPS ≥ 10 (RR = 1.698, P = .015), patients with CPS ≥ 20 (RR = 1.488, P = .035), patients with CPS ≥ 30 (RR = 1.679, P = .028), patients with CPS ≥ 40 (RR = 1.783, P = .02), and patients with CPS ≥ 50 (RR = 1.819, P = .027). In addition, patients with human papillomavirus (HPV) positive status had significantly higher posttreatment pathological complete response rates than those with negative HPV status (RR = 2.15, P = .01). Moreover, there is a significant correlation between CD4+ tumor-infiltrating lymphocytes values and pathological responses (Z = 0.511, P = .01).

Conclusions:

Our findings suggest that CPS, HPV status, and CD4+ tumor-infiltrating lymphocytes values may be predictive biomarkers of the efficacy of neoadjuvant immunotherapy in patients with LAHNSCC. Further clinical studies are needed to validate our results.

1. Introduction

Head and neck squamous cell carcinoma (HNSCC) is among the most prevalent malignant tumors worldwide. More than 60% of patients with HNSCC are diagnosed with a locally advanced (stages III–IVB) tumor at the time of initial diagnosis.^[1] In general, the results of single therapy for locally advanced head and neck squamous cell carcinoma (LAHNSCC) are poor, and treatment is usually a combination of different strategies, including surgery, radiation, and chemotherapy. However, even with local resection and adjuvant radiotherapy or radiochemotherapy, patients with resectable LAHNSCC have a 50% risk of local recurrence and distant metastasis within 3 years after treatment.^[1,2] Therefore, further improvements are needed in the treatment strategy for patients with LAHNSCC.

Immune checkpoint inhibitors (ICIs), including nivolumab and pembrolizumab, have been used as 1st-line treatment options for patients with recurrent or metastatic HNSCC, according to the CheckMate141 and KEYNOTE-048 studies.^[3,4] Several clinical studies are currently ongoing to evaluate the efficacy of neoadjuvant immunotherapy in patients with LAHNSCC, and some promising results have been reported.^[5–10] However, clinical outcomes vary among LAHNSCC patients with similar clinical and pathological features at diagnosis and similar treatments. Therefore, identification of factors that predict treatment efficacy is the 1st step toward identifying patients with LAHNSCC who could benefit from neoadjuvant immunotherapy.

Multiple biomarkers may have predictive value during neoadjuvant immunotherapy for LAHNSCC. Combined positive score (CPS) may serve as a potential biomarker for predicting the efficacy of immunotherapy. Other molecular markers, including human papillomavirus (HPV) status, tumor mutational burden (TMB), CD4+/CD8+ tumor-infiltrating lymphocytes (TILs), and gene mutations, have also been linked to the efficiency of neoadjuvant immunotherapy; however, the results have been inconsistent.

To further explore the relationship between the abovementioned biomarkers and pathological response in patients with LAHNSCC after neoadjuvant immunotherapy, we pooled data from clinical trials on neoadjuvant immunotherapy for LAHNSCC and performed a meta-analysis by assessing the correlation between biomarkers and pathological response in LAHNSCC patients.

2. Methods

2.1. Data sources and search strategy

In this meta-analysis, we searched the PubMed, Embase, Web of Science, and Cochrane databases for literature published until June 19, 2024. A search strategy combining subject terms (e.g., MeSH in PubMed) with abstracts or titles was used, and the search formula included the following terms: neoplasms, head and neck, head, neck neoplasms, head and neck neoplasm, neoadjuvant therapies, therapy, and neoadjuvant (Table S1, Supplemental Digital Content 1).

2.2. Selection criteria

In the evaluation of tumor treatment outcomes, pathological response not only represents the treatment effect but is also an important indicator to predict the prognosis of the tumor. Major pathological response (MPR) and pathological complete response (pCR) are 2 important evaluation indicators in the pathological response, where MPR is usually defined as ≤10% of viable tumor cells on pathology, and pCR is usually defined as 0% tumor cells on pathology.^[11] Studies were included if they met the following criteria: they reported histologically confirmed untreated resectable LAHNSCC; the intervention encompassed neoadjuvant ICIs, including programmed cell death receptor-1 (PD-1) or its ligand PD-L1 inhibitors and cytotoxic T-lymphocyte–associated protein 4 inhibitors; the expression level of PD-L1/TMB/HPV/CD4+/CD8+ TIL and whole-exome sequencing of the local tumors were determined preoperatively; the results of the pathological response of the tumors, MPR/pCR, were obtained postoperatively; and the studies were completed and the corresponding results were available.

2.3. Exclusion criteria

The exclusion criteria were as follows: studies on irrelevant topics, as well as reviews, meta-analyses, case reports, and guidelines; studies with inadequate experimental data; studies that did not produce relevant outcomes; and studies that discussed retrospective studies.

2.4. Data extraction and quality assessment

Two authors independently performed the literature search, study selection, and data extraction. Controversial data were discussed with a 3rd expert when necessary. For each study, the following relevant data were recorded: study characteristics: 1st author, year of publication, clinical trial, NCT number, treatment regimen, study type, study phase, study site, main inclusion criteria, ICI drugs, and sample size; baseline characteristics of the enrolled patients: age, gender, tumor location, CPS, TMB, HPV status, and results of CD4+/CD8+ TIL and whole-exome sequencing; end-point data: including pCR and MPR. In cases of incomplete data, the original authors were contacted for supplementary materials, if possible.

As the majority of clinical studies in the literature were single-arm studies, their quality was evaluated using the methodological index for non-randomized studies checklist^[12] (Table S2, Supplemental Digital Content 2).

2.5. Statistical analysis

The posttreatment pathological responses, including partial pathological response, MPR, and pCR, were calculated in the included studies, as were the MPR rate, pCR rate, and the incidence of overall pathological responses after treatment. Heterogeneity was assessed by the Cochrane Q test and I² statistic. If the P value of the Q test was large (e.g., >.1) and the I² value was small (e.g., <50%), a fixed-effects model was used; otherwise, a random-effects model was used. Data regarding MPR rate and pCR rate were expressed as 95% confidence intervals (CIs), and P values <.05 were considered statistically significant. Statistical analyses included Spearman correlation analysis and the Mann–Whitney U test. Spearman correlation coefficients were converted to Z-scores using Fisher Z-transform, and GraphPad 8.0 (GraphPad Software, Inc.) was used as the software. Main statistical analyses were performed using STATA software, version 17.0 (Stata Corp., LLC), and the results are presented in the form of forest plots. The analysis of variance was also carried out using the dummy variable method in this software. In this meta-analysis, we used the Preferred Reporting Items for Systematic Evaluation and Meta-Analysis Statements 2020 and A Measurement Tool to Assess Systematic Reviews 2 checklists to improve the reporting of our study.^[13,14] In addition, in the sensitivity analyses, single studies were sequentially excluded to assess the stability of the pooled findings. Publication bias was assessed using funnel plots.

3. Results

3.1. Patients’ characteristics

According to the search strategy mentioned above, and identification and exclusion performed with reference to the inclusion criteria, a total of 26 articles on neoadjuvant immunotherapy for LAHNSCC^{[5–10,15–34]} were retrieved, comprising a total of 980 patients (Fig. 1). Seventeen studies reported pathological responses after neoadjuvant immunotherapy (either MPR or pCR), and 15 studies reported both MPR and pCR after neoadjuvant immunotherapy. The treatment modalities included neoadjuvant immune monotherapy (3 studies) and neoadjuvant immune combination therapies (ICIs combined with chemotherapy, targeted drug therapy, radiotherapy, or other ICI agents: 7 studies) to treat LAHNSCC (Table 1).

Figure 1.

Flowchart of the literature selection process.

Table 1.

Main characteristics of all included studies.

First author	Published year	Country	ClinicalTrials.gov ID	Intervention model	Masking	Study type	Study phase	Main inclusion criteria	ICIs	Dose of ICI	Other neoadjuvant Therapy	Sample size	Surgical resection	Outcome reported
Ferris RL et al	2021	USA/Canada/Europe	NCT02488759	Parallel assignment	Open label	Interventional	I/II	Stage III–IV resectable HNSCC	Nivolumab	240 mg iv d1,15	N/A	52	45	TRAEs (grade 3–4), MPR, pCR, pPR
Wise-Draper TM et al	2022	USA	NCT02641093	Single group assignment	Open label	Interventional	II	Local-regionally advanced stage III/IV HNSCC	Pembrolizumab	200 mg iv (7–21 d prior to surgery)	N/A	92	88	1-yr DFS PR
Zhang Z et al	2022	China	ChiCTR1900025303	Single group assignment	Open label	Observational	II	Resectable stage III–IVB HNSCC	Camrelizumab	200mg, iv, q3w	Paclitaxel 260 mg/m2 (or docetaxel 75 mg/m2) +cisplatin 75 mg/m2	30	27	pCR, MPR Incidence of TRAEs (grade 3–4)
Vos JL et al	2021	Netherlands	NCT03003637	Single group assignment	Open label	Interventional	IB/IIA	T3-4N0-3M0 HNSCC	Nivolumab(6)、 Nivolumab+ Ipilimumab (26)	200 mg iv (w 1 + w3)、 (240 mg iv+ 1 mg/kg) w1 +240 mg w3	N/A	32	29	MPR, PPR, NPR Incidence of TRAEs (grade 3–4)
Ju WT et al	2022	China	NCT04393506	Sequential assignment	Open label	Interventional	I	Locally advanced resectable OSCC	Camrelizumab	200mg, iv, qd d1, 15, 29.	Apatinib (250 mg, once daily	21	20	MPR Incidence of TRAEs (grade 3–4)
Darragh LB et al	2022	USA	NCT03635164	Single group Assignment	Open label	Interventional	I/IB	Locally advanced HPV-unrelated oral cavity or larynx HNSCC	Durvalumab	1500 mg, iv, 3–6 wk before standard-of-care surgery	SBRT (24Gy)	21	19	Safety, radiographic, pathologic and objective response; locoregional control; PFS, OS
Leidner R et al	2021	USA	NCT03247712	Single group Assignment	Open label	Interventional	IB	Locally advanced HPV (+) and HPV (−) HNSCC	Nivolumab	240mg, iv, w0, w2, w4	SBRT 40 Gy in 5 fractions 24 Gy in 3 fractions	21	21	pCR MPR
Uppaluri R et al	2020	USA	NCT02296684	Sequential Assignment	Open label	Interventional	II	Locally advanced, HPV unrelated HNSCC	Pembrolizumab	200mg, iv, 2–3 wk before surgery	N/A	36	36	pTR-2 1-yr relapse rate
Knochelmann HM et al	2021	USA	NCT03021993	Single group assignment	Open label	Interventional	II	Stages II–IVA OCSCC	Nivolumab	3 mg/kg iv, q3w	N/A	14	12	ORR
Huang X et al	2022	China	NCT04947241	Single group assignment	Open label	Interventional	IB	III–IVB locally advanced HNSCC	Toripalimab	240 mg iv, q3w	Gemcitabine (1000 mg/m2, d1, 8), and cisplatin (80 mg/m2, d1)	23	18	Safety, TRAEs pCR, MPR, ORR R0 resection rate
Schoenfeld JD et al	2020	USA	NCT02919683	Parallel assignment	Open label	Interventional	II	Locally advanced OCSCC	Nivolumab(14)、Nivolumab + (ipilimumab) (15)	3 mg/kg w1 + 3、 3 mg/kg w 1 + 3 + (1 mg/kg w1)	N/A	29	28	Safety PFS OS
Luginbuhl AJ et al	2022	USA	NCT03238365	Parallel assignment	Open label	Interventional	I	Any-stage resectable HNSCC	Nivolumab (20)	240 mg iv d1, 15	Tadalafil 10 mg orally once daily 4w	50	45	Safety
Redman JM et al	2022	USA	NCT04247282	Sequential assignment	Open label	Interventional	I/II	HPV-unrelated T2-4b oral cavity or laryngeal squamous cell carcinoma	Bintrafusp alfa	1200 mg, iv, d1,15	N/A	14	14	Feasibility Safety Efficacy
Ferrarotto R et al	2021	USA	NCT03565783	Single group assignment	Open label	Interventional	II	II–IV HNSCC	Cemiplimab	350 mg, iv, 3 wk、6 wk before surgery	N/A	20	20	ORR pathologic response safety and tolerability DFS, OS
Oliva M et al	2021	Canada	NCT03575598	Single group assignment	Open label	Interventional	I	T2-4a, N0-2 or T1 > 1 cm-N2 oral cavity carcinomas	Nivolumab	240mg, d15	Sitravatinib, 120 mg once daily until 48 h	12	10	Pharmacodynamic and immune effects safety and tolerability
Huang Y et al	2023	China	NCT04473716	Sequential assignment	Open label	Interventional	I	III/IVA oral squamous cell carcinoma	Toripalimab	240mg, iv, qd, on d 1, 22	Paclitaxcel of 260mg/m2, iv, qd, on d 1, 22 Cisplatin of 75 mg/m2, iv, qd, on d1, 22	20	20	MPR 2-yr OS 2-yr tumor recurrence rate
Ma TM et al	2023	USA	NCT03618134	Parallel assignment	Open label	Interventional	Ib/II	Locally advanced HPV + OPSCC	Durvalumab (3)、 tremelimumab + durvalumab (16)	1500 mg q4 wk × 2 cycles, 75 mg + 1500 mg q4 wk × 2 cycles	SBRT day 0	19	19	Safety and tolerability efficacy TRAEs
Patel SA et al	2023	USA	NCT03174275	Parallel assignment	Open label	Interventional	II	Locally advanced HNSCC	Durvalumab	750 mg, iv w1, 3, 5, 7, and 9.	Carboplatin (AUC 2) nab‐paclitaxel (100 mg/m2) 6 wkly	39	35	pCRR, cCRR,cRRPFS, OS, toxicity
Wang K et al	2023	China	ChiCTR2200055719	Single group assignment	Open Label	Interventional	II	III–IV HNSCC	Pembrolizumab	200mg, iv, d1, q21d × 2 cycles	Cisplatin 75 mg/m2 D1 paclitaxel 175 mg/m2 D1, q21 d × 2 cycles	28	22	PCR Safety DFS OS
Kim CG et al	2024	Korea	NCT03737968	Single group Assignment	Open label	Interventional	II	II–IVA HNSCC	Durvalumab (24), durvalumab+ tremelimumab (24)	1500 mg q4 wk, 1500 mg q4 wk + 75 mg IV q4 wk	N/A	48	45	Locoregional relapse rate Distant metastatic rate PFS
Gong H et al	2024	China	NCT04156698	Single group assignment	Open label	Interventional	II	Locally advanced HNSCC	Camrelizumab	200mg i.v. d1, q3w, 3 cycles	Docetaxel (domestic) 75 mg/m2 i.v. d1, Cisplatin 25 mg/m2 i.v. d1–3, Capecitabine 800 mg/m2 po bid, d1–14, q3w, 3 cycles	51	6	Overall Response Rate LPR PFS MFS, OS
Li, X. et al	2021	China	No. 201356HN	NA	NA	NA	NA	Locally advanced HNSCC	Sintilimab	200mg d 1 of each cycle, q3w	Docetaxel (75 mg/m2), platinum (75 mg/m2), and fluorouracil (750mg/m2/d for 5 d	163	52	ORR, 2-yr PFS, 2-yr OS
Ou X et al	2024	China	NCT04995120	Single group Assignment	Open label	Interventional	II	Locally advanced laryngeal and hypopharyngeal squamous cell carcinoma	Toripalimab	240mg, iv, d1, q3w for 3 cycles	Paclitaxel 175mg/m2 d2 or Nab-Paclitaxel260mg/m2 d2, cisplatin 25 mg/m2 d2–4 q3w	27	6	LPR, ORR pCR, MPR
Wu D et al	2024	China	NCT04826679	Single group assignment	Open label	Interventional	II	III–IV HNSCC	Camrelizumab	200mg, d1, q3w for 3 cycles	Nab-paclitaxel (260 mg/m2) and cisplatin (60 mg/m2) d1, q3w for 3 cycles	48	27	ORR, pCR, MPR, DCR, PFS, OS Adverse events
Ferrarotto.R..et al	2020	USA	NCT03144778	Parallel assignment	Open label	Interventional	I	Oropharyngeal squamous cell carcinoma (OPSCC), stage II, III, or IVA	Durvalumab, durvalumab+ tremelimumab	1500 mg, 1500 mg+75 mg, 1, 29	NA	29	24	Change of CD8+ tumor-infiltrating lymphocytes,ORR
Wang, H. L.et al	2024	China	NCT 05522985	Parallel assignment	Open label	Interventional	II	Locally advanced resectable oral squamous cell carcinoma	Toripalimab	240mg, d1, q3w	Paclitaxel (albumin-bound) 260 mg/m2, intravenous drip; cisplatin 75 mg/m2, d1, q3w	41	23	PCR, MPR, OR, OS, FPS

cCRR = clinical complete response rate, cRR = clinical response rate, DCR = disease control rate, DFS = disease-free survival, HNSCC = head and neck squamous cell carcinoma, HPV = human papillomavirus, ICIs = immune checkpoint inhibitors, iv = intravenous injection, LA = locally advanced, LPR = larynx preservation rate, MFS = metastasis-free survival, MPR = major pathological response, N/A = not available, NPR = no assumed pathological response, NR = non-randomized, ORR = objective response rate, OS = overall survival, OSCC = oral squamous cell carcinoma, OSPCC = oropharyngeal squamous cell carcinoma, pCR = pathological complete response, pCRR = pathological complete response rate, PFS = progression-free survival, PFS = progression-free survival, pPR = pathological partial response, PR = pathological response, pTR = pathological tumor response, q3w = 3 weeks using a dose, SBRT = stereotactic body radiation therapy, TRAEs = treatment-related adverse events.

3.2. Pathological responses after neoadjuvant immunotherapy

After neoadjuvant immunotherapy, the MPR rate was 16.5% (95% CI, 0.097–0.245, P < .001), and the pCR rate was 19.8% (95% CI, 0.107–0.313, P < .001) (Fig. 2). The overall pathological response rate, pooled for the cases in which the pathological response included MPR and pCR, was 39.7% (95% CI, 0.259–0.542, P < .001) (Fig. 2). The results of the analysis of variance showed that after neoadjuvant immunotherapy, there was a difference in the MPR rate between the neoadjuvant immune combination therapy group and the neoadjuvant immune monotherapy group (coefficient = 0.1865006, P = .003); there was no difference in the pCR rate between the neoadjuvant immune combination therapy group and the neoadjuvant immune monotherapy group (coefficient = 0.2098933, P = .131); and there was a difference in the total pathological response rate between the neoadjuvant immune combination therapy group and the neoadjuvant immune monotherapy group (coefficient = 0.3476412, P = .008).

Figure 2.

Pathological responses after neoadjuvant immunotherapy. (A) MPR rate; (B) pCR rate; and (C) overall pathological response rate. CI = confidence interval, MPR = major pathological response, qCR = pathological complete response.

3.3. CPS and pathological responses after neoadjuvant immunotherapy

Ten studies reported both pretreatment CPS and posttreatment pathological response. The correlation between pretreatment CPS values and posttreatment pathological responses was analyzed, and the results showed a significant correlation between pretreatment CPS values and posttreatment pathological responses (Z = 0.289, P < .001). The patients were then divided according to their pretreatment CPS (CPS cutoff = 1, 5, 10, 20, 30, 40, and 50). There were no significant differences in the posttreatment overall pathological response rate between the patients with CPS ≥ 1 and those with CPS < 1 (relative risk [RR] = 1.5, P = .155) (Table 2). However, there were differences in the posttreatment overall pathological response rate between the patients with CPS ≥ 5 and those with CPS < 5 (RR = 1.852, P = .012); the patients with CPS ≥ 10 and those with CPS < 10 (RR = 1.698, P = .015); the patients with CPS ≥ 20 and those with CPS < 20 (RR = 1.488, P = .035); the patients with CPS ≥ 30 and those with CPS < 30 (RR = 1.679, P = .028); the patients with CPS ≥ 40 and those with CPS < 40 (RR = 1.783, P = .02); and the patients with CPS ≥ 50 and those with CPS < 50 (RR = 1.819, P = .027) (Fig. 3).

Table 2.

Differences in pathological responses between above and below different CPS cutoff values.

CPS cutoff values	RR	95% CI	P value
1	1.500	0.858–2.623	.155
5	1.852	1.142–3.004	.012
10	1.698	1.110–2.598	.015
20	1.488	1.028–2.155	.035
30	1.679	1.056–2.668	.028
40	1.783	1.097–2.897	.020
50	1.819	1.069–3.096	.027

CI = confidence interval, CPS = combined positive score, RR = relative risk.

Figure 3.

Overall pathological response rate at different combined positive score (CPS) cutoff values. (A) Cutoff value = 5; (B) cutoff value = 10; (C) cutoff value = 20; (D) cutoff value = 30; (E) cutoff value = 40; and (F) cutoff value = 50. CI = confidence interval, RR = relative risk.

In the neoadjuvant immune monotherapy group, there were no significant differences in the posttreatment overall pathological response rate between the patients with CPS ≥ 1 and those with CPS < 1 (RR = 0.838, P = .841), the patients with CPS ≥ 5 and those with CPS < 5 (RR = 1.930, P = .457), and the patients with CPS ≥ 10 and those with CPS < 10 (RR = 3.255, P = .186) (Table 3). However, there were differences in the posttreatment overall pathological response rate between the patients with CPS ≥ 20 and those with CPS < 20 (RR = 6.635, P = .033); patients with CPS ≥ 30 and those with CPS < 30 (RR = 6.635, P = .033); patients with CPS ≥ 40 and those with CPS < 40 (RR = 7.565, P = .022); and patients with CPS ≥ 50 and those with CPS < 50 (RR = 8.585, P = .015) (Fig. 4).

Table 3.

Differences in pathological responses between above and below different CPS cutoff values in the monotherapy group.

CPS cutoff values	RR	95% CI	P value
1	0.838	0.150–4.699	.841
5	1.930	0.341–10.914	.457
10	3.255	0.566–18.735	.186
20	6.635	1.168–37.679	.033
30	6.635	1.168–37.679	.033
40	7.565	1.341–42.664	.022
50	8.585	1.527–48.264	.015

CI = confidence interval, CPS = combined positive score, RR = relative risk.

Figure 4.

Overall pathological response rate at different combined positive score (CPS) cutoff values in the neoadjuvant immune monotherapy group. (A) Cutoff value = 20; (B) cutoff value = 30; (C) cutoff value = 40; and (D) cutoff value = 50. CI = confidence interval, RR = relative risk.

3.4. HPV and pathological responses after neoadjuvant immunotherapy

Nine studies reported both pretreatment HPV status and posttreatment pathological response. We focused on the relationship between HPV status and pathological responses after neoadjuvant immunotherapy. Analysis of the correlation between pretreatment HPV status and pathological response after neoadjuvant immunotherapy showed a significant correlation between pretreatment HPV status and posttreatment pathological response (r = 0.69, P < .001). Analyzing the differences in pathological responses after treatment according to HPV status, the results showed that patients with HPV positive (+) status had significantly higher posttreatment pCR rates than those with HPV negative (−) status (RR = 2.15, P = .01) (Fig. 5). Similarly, in the neoadjuvant immune combination therapy group, patients with HPV (+) had significantly higher posttreatment pCR rates than those with HPV (−) (RR = 2.446, P = .005) (Fig. 5).

Figure 5.

HPV and pathological responses after neoadjuvant immunotherapy: (A) HPV (+) versus HPV (−) and (B) combination therapy group: HPV (+) versus HPV (−).CI = confidence interval, HPV = human papilloma virus, RR = relative risk.

We further analyzed the relationship between HPV status and overall pathological response at different CPS cutoff values. When the CPS was ≥ 1, there was a significant difference in the overall pathological responses between HPV (+) patients and HPV (−) patients after treatment (RR = 1.974, P = .024). However, when the CPS was ≥20, there was no significant difference in the posttreatment pathological responses between HPV (+) patients and HPV (−) patients (RR = 1.993, P = .09).

3.5. TMB and pathological response after neoadjuvant immunotherapy

Three studies reported both pretreatment TMB and posttreatment pathological response. Due to the limited data collected, we only analyzed the correlation between TMB and pathological response after neoadjuvant immunotherapy. The pooled correlation results were subjected to meta-analysis, which showed a correlation between the TMB values and pathological response after treatment (Z = 0.250, P = .047).

3.6. CD4+ TIL, CD8+ TIL, and pathological response after neoadjuvant immunotherapy

Three studies reported pretreatment CD4+ TIL, CD8+ TIL, and posttreatment pathological response. By analyzing the relationship between CD4+ TIL values and pathological responses after neoadjuvant immunotherapy, we found a significant correlation between CD4+ TIL values and pathological responses (Z = 0.511, P = .01). In contrast, there was no significant correlation between CD8+ TIL values and pathological response after neoadjuvant immunotherapy (Z = 0.135, P = .271).

3.7. Mutated genes and pathological response after neoadjuvant immunotherapy

Four studies reported both pretreatment next generation sequencing and posttreatment pathological response. In the included studies, the mutation rate was 75.9% for TP53, 16.3% for FAT1, and 24% for CDKN2A (Fig. 6). We analyzed the relationship between several mutated genes and pathological responses after neoadjuvant immunotherapy. The results showed that the MPR rate was 45.9% in individuals with TP53 mutations, 60.1% in individuals with FAT1 mutations, and 42.9% in individuals with CDKN2A mutations (Fig. 6).

Figure 6.

Mutated genes and pathological response after neoadjuvant immunotherapy: (A) TP53 mutation rate; (B) MPR rate in mutant TP53; (C) FAT1 mutation rate; (D) MPR rate in mutant FAT1; (E) CDKN2A mutation rate; and (F) MPR rate in mutant CDKN2A. CI = confidence interval, ES = effect size, MPR = major pathological response.

3.8. Sensitivity analysis and publication bias

After the sequential exclusion of single studies, the pooled findings did not show significant changes, indicating that the results were stable (Fig. S1, Supplemental Digital Content 3). In the meta-analysis, the funnel plot showed that most of the points were concentrated in the upper center and seemed to be relatively symmetrically distributed around the dotted line, indicating no clear evidence of significant publication bias (Fig. S2, Supplemental Digital Content 4).

4. Discussion

In this meta-analysis, we primarily explored the relationship between biomarkers and pathological responses after neoadjuvant immunotherapy in patients with LAHNSCC. Our results showed that there was a correlation between CPS values and posttreatment pathological response; the overall incidence of MPR and pCR after treatment rose with an increase in CPS, and HPV (+) led to a better pathological response; and the correlation was more pronounced in the patients with CPS of 5 or higher, suggesting that CPS and HPV status may be predictors of the efficacy of neoadjuvant immunotherapy in LAHNSCC patients.

Despite the increasing interest in the use of neoadjuvant immunotherapy in patients with LAHNSCC, there is still a lack of valuable biomarkers for predicting the effectiveness of neoadjuvant immunotherapy. There is a correlation between CPS and the efficacy of immunotherapy in individuals with recurrent or metastatic HNSCC. The results of the current phase III trial demonstrated that the use of pembrolizumab alone or the combination of pembrolizumab and chemotherapy significantly improved the OS of patients with CPS ≥ 1 and CPS ≥ 20. To further evaluate the importance of PD-L1 expression as an indicator of treatment prognosis, Burtness et al^[35] compared the therapeutic efficacy between pembrolizumab alone or pembrolizumab in combination with chemotherapy and cetuximab in combination with pembrolizumab. Their results showed that the efficacy of pembrolizumab alone or pembrolizumab combination chemotherapy rose with an increase in PD-L1 expression. However, there have been no reports on the correlation between CPS and pathological response in neoadjuvant immunotherapy for patients with LAHNSCC.

To the best of our knowledge, this is the 1st study on the correlation between CPS and posttreatment pathological response among studies related to neoadjuvant immunotherapy in patients with LAHNSCC. In our study, when patients with LAHNSCC were treated with neoadjuvant immunotherapy regimens, there was a significant difference in the posttreatment overall pathological response rates between patients with CPS ≥ 5 and those with CPS < 5. Our results suggest that CPS may be a useful biomarker of therapeutic efficacy in neoadjuvant immunotherapy for LAHNSCC. The correlation between CPS and the efficacy of neoadjuvant immunotherapy in other tumor types has been reported in some studies, but the results were not consistent. In neoadjuvant immunotherapy for breast cancer, the interim analysis of KEYNOTE-355, a phase III trial, showed that the therapeutic efficacy of neoadjuvant immunotherapy increased with increasing CPS; that is, there were significant differences in prognosis between patients with CPS ≥ 10 and those with CPS < 10 (23.0 vs 14.7) and between patients with CPS ≥ 20 and those with CPS < 20 (24.0 vs 15.9).^[36] However, in some studies, there was no significant correlation between CPS and immunotherapy efficacy. In the PERFECT study, there was also no statistically significant difference in the proportion of responders and nonresponders with CPS ≥ 10 (62% vs 30%, P = .069).^[37]

In our study, without considering other factors, the pCR rate after neoadjuvant immunotherapy differed depending on HPV status (RR = 2.15, P = .01). Due to the limited number of studies that included both HPV status and CPS, the different combinations of HPV and PD-L1 expression status did not show significant correlation with pathological responses after neoadjuvant immunotherapy. In a retrospective study analyzed by Zhang S et al,^[38] the addition of cetuximab, when compared with PD-1 inhibitor monotherapy, improved the objective response rate (ORR) in HPV (−) disease (pooled ORR in monotherapy vs combination therapy: 15% vs 46%, P < .001), but not in HPV (+) disease (17% vs 18%, P = .686). In the CheckMate 358 trial, the radiographic response rates in 49 evaluable patients were 12.0% and 8.3% in the HPV (+) and HPV (−) cohorts, respectively.^[5] Future studies with larger sample sizes are needed to further evaluate the effect of HPV status on neoadjuvant immunotherapy for patients with LAHNSCC.

TMB denotes the total number of somatic non-synonymous mutations (including single-nucleotide variants and small insertion/deletion mutations) detected within a specific tumor genomic region. In our study, the results of correlation analyses showed a significant correlation between TMB values and posttreatment pathological response (Z = 0.250, P = .047). TMB significantly and independently predicted response to pembrolizumab in HNSCC patients in the KEYNOTE-012 and KEYNOTE-055 trials.^[39] Hence, TMB shows some potential value in predicting treatment effects.

CD4+ TIL refers to CD4+ T lymphocytes that infiltrate the tumor tissue. Our results showed a significant correlation between CD4+ TIL values and pathological response (Z = 0.511, P = .01). CD8+ TIL refers to CD8+ T lymphocytes present in the tumor tissue. We found no significant correlation between CD8+ TIL and pathological response after neoadjuvant immunotherapy (Z = 0.135, P = .271). In a study of oral squamous carcinoma, the expression levels of the immunosuppressive marker CD39 in CD4+, regulatory T-cell (Treg), and CD8+ T-cell subsets differed significantly between responders and nonresponders, suggesting that CD4+ TIL could potentially serve as a predictor of the efficacy of neoadjuvant immunotherapy.^[40] Moreover, the peripheral blood immunophenotyping analysis revealed that the patients who reached MPR had higher baseline CD8+ T-cell activation levels, which manifested by higher expression levels of PD-1, TIGIT, and IFN-γ, and lower expression levels of PD-L1. This suggests that the activation status of CD8+ TIL correlates with the response to neoadjuvant PD-1 blockade therapy.^[40] Considering the limited number of the included studies, the predictive value of CD4+ TIL vs CD8+ TIL needs to be further explored.

Mutant genes are gradually becoming important predictive markers of the effectiveness of head and neck immunotherapy due to their ability to generate neoantigens and thus potentially trigger an immune response in the body. Among the multiple mutated genes analyzed in this study, the highest mutation rate was found in the TP53 gene, whereas the highest MPR rate was found in the FAT1 mutated gene, with differences between the different mutated genes. However, specific correlation analyses between the mutated genes and the pathological response after neoadjuvant immunotherapy did not show meaningful results. Considering that multiple genes may be involved in the process of tumor development and progression, the combination of multiple mutant genes may bring better predictive value.

In our study, single biomarkers were of limited predictive value for pathological responses after neoadjuvant immunotherapy in LAHNSCC patients. Given the limitations of single biomarkers in predicting the efficacy of immunotherapy, the combination of multiple biomarkers has become a hot topic of research. In 1 study, the ORR of the subgroup with TMB ≥ 175 mut/exome and PD-L1 CPS ≥ 1 was as high as 34% (17/50), and the ORR of the subgroup with TMB ≥ 175 mut/exome and T-cell-inflamed gene expression profile was also 34% (13/38), which was higher than the ORR of the subgroups with high expression of each biomarker alone, indicating that TMB, in combination with PD-L1 or T-cell-inflamed gene expression profile, could more accurately screen out the patients with a good response to pembrolizumab treatment. The predictive effect of the combined assay was superior to that of individual biomarkers.^[39] This suggests that the combination of multiple biomarkers has great potential in terms of predictive value.

There are some limitations to this study. First, the trials involved in the study were all phase I/II prospective trials, which were nonrandomized controlled trials with insufficient follow-up time. Second, the CPSs of several different studies were based on different measurement methods, which may have introduced bias. Third, when neoadjuvant immunotherapy was performed, different studies used ICIs at different time points, which may have led to bias in the results because of the difference in the timing of drug use. Fourth, the small number of clinically relevant studies and the limited data from trials that included HPV status and CPS may introduce statistical bias. Finally, the study lacked relevant indicators other than CPS that could be analyzed, and other indicators could be improved for comprehensive analysis in the future.

5. Conclusion

After summarizing the data from the included clinical trials, we can conclude that there is a correlation between CPS values and posttreatment pathological response when neoadjuvant immunotherapy is used in patients with LAHNSCC. A higher pretreatment CPS or HPV (+) results in a better therapeutic effect. There is a significant correlation between CD4+ TIL values and pathological responses, and the therapeutic effect is more pronounced when CPS is 5 or higher, suggesting that CPS, HPV status, and CD4+ TIL values may be predictive biomarkers of the effect of neoadjuvant immunotherapy in patients with LAHNSCC. A large-scale phase III clinical trial of preoperative neoadjuvant immunotherapy for LAHNSCC patients should be conducted to further validate the above results.

Author contributions

Conceptualization: Shicai Chen.

Data curation: Jianqiao He, Yi Ma, Guoning Yu.

Formal analysis: Jianqiao He, Xiaoqiong Shi.

Funding acquisition: Caiyun Zhang.

Investigation: Wei Wang.

Methodology: Hongliang Zheng.

Project administration: Minhui Zhu.

Resources: Caiyun Zhang.

Software: Guoning Yu.

Supervision: Minhui Zhu.

Writing – original draft: Yi Ma.

Writing – review & editing: Caiyun Zhang.