Neoadjuvant immunochemotherapy for locally advanced resectable oral squamous cell carcinoma: a prospective single-arm trial (Illuminate Trial)

Background: Locally advanced oral squamous cell carcinoma (LAOSCC) is associated with a high rate of recurrence and poor survival. Given the recent successes of neoadjuvant immunochemotherapy (NAICT) in solid tumors, it is promising to use this treatment modality to achieve a better pathological response and improve the survival of LAOSCC, and clinical evidence is needed to assess its safety and efficacy. Patients and Methods: A prospective trial of NAICT with toripalimab (PD-1 inhibitor) and albumin paclitaxel/cisplatin (TTP) was conducted in patients with clinical stage III and IVA OSCC. Intravenous albumin paclitaxel (260 mg/m2), cisplatin (75 mg/m2), and toripalimab (240 mg) were given in sequence on day 1 of each 21 day cycle for two cycles, followed by radical surgery and risk-adapted adjuvant (chemo)radiotherapy. The primary endpoints were safety and major pathological response (MPR). Targeted next generation sequencing and multiplex immunofluorescence were performed to assess clinical molecular characteristics and the tumor immune microenvironment in the pre-NAICT and post-NAICT tumor samples. Results: Twenty patients were enrolled. NAICT was well-tolerated with a low incidence of grades 3–4 adverse events in three patients. The completion rates of NAICT and subsequent R0 resection were 100%. The MPR rate was 60%, including a 30% pathological complete response. MPR was achieved in all four patients with a combined positive score of PD-L1>10. The density of tertiary lymphatic structure in post-NAICT tumor samples predicted the pathological response to NAICT. During the median 23-month follow-up, the disease-free survival was 90%, and the overall survival was 95%. Conclusions: NAICT with the TTP protocol in LAOSCC is feasible and well tolerated, with a promising MPR and no obstruction on subsequent surgery. This trial is supportive of further randomized trials using NAICT in LAOSCC.


Introduction
Oral squamous cell carcinoma (OSCC) is one of the most common cancers of the head and neck region. More than 60% of OSCC patients are at late clinical stages at the first clinic visit, and their 5-year survival rates are less than 50% [1] . Despite the current availability of multiple traditional therapeutic strategies and immunotherapy,~50% of patients suffer local recurrence or distant metastasis [2,3] . The lack of significant improvement in survival and quality of life highlights the urgent need for novel treatment strategies for OSCC.
Neoadjuvant therapy with clinically validated antitumor activity has driven many clinical trials in solid tumors [4][5][6][7] . The theoretical purposes of neoadjuvant therapy include reducing the burden of locoregional disease, thus facilitating surgical resection and the elimination of minimal residual tumor, which reduce the risk of distant metastases and allow symptomatic improvements that can result in increased performance [4] . Major pathological response (MPR) to neoadjuvant therapy, which is defined as less than or equal to 10% residual viable tumor cells in the primary tumor bed after neoadjuvant therapy, has been validated to predict longer survival [8] , so it has been recommended as an alternative endpoint for neoadjuvant clinical trials in many solid tumors, including OSCC [8][9][10][11] .
Immune checkpoint inhibitors have been demonstrated to improve survival in the first-line and second-line settings of recurrent/metastatic head and neck squamous cell carcinoma (HNSCC) [12][13][14] . These successes have led to immune checkpoint inhibitors being tested in the neoadjuvant setting for the treatment of locally advanced HNSCC, including OSCC. However, neoadjuvant antiprogrammed cell death 1(PD-1) monotherapy has been reported to elicit a relatively low MPR rate (6% for pembrolizumab and 8% for nivolumab) in OSCC [15][16][17] . Preclinical data suggest that adding chemotherapy to PD-1 inhibitor therapy can increase antigen release due to tumor cell death, which stimulates T-cell-mediated immunity; furthermore, several clinical trials have shown that neoadjuvant immunochemotherapy (NAICT) results in significantly longer survival and a higher percentage of patients with a pathological complete response (pCR) in solid tumor [18][19][20][21] . However, there is insufficient clinical evidence in support of an improved pathological response rate to NAICT in OSCC.
In this study, we conducted a prospective, single-arm trial to assess the safety and pathological efficacy of NAICT with toripalimab (PD-1 inhibitor) and albumin paclitaxel/cisplatin (TTP) in patients with locally advanced resectable OSCC (LAROSCC); the completion of NAICT, R0 resection of the primary tumor after NAICT and predictive biomarkers of pathological response were also investigated.

Study design and participants
This trial was a prospective, single-arm, open-label trial. This work has been reported in line with the strengthening the reporting of cohort, cross-sectional and case-control studies in surgery (STROCSS) criteria [22] , Supplemental Digital Content 1, http://links.lww.com/JS9/A569. The key inclusion criteria were patients aged 18 to 75 years with a histopathological diagnosis of OSCC of clinical stages III or IVA (AJCC, 8 th edition), an ECOG PS of 0 to 1, and adequate organ and bone marrow function. The key exclusion criteria included prior systemic cancer therapy, known immunosuppression, active autoimmune disease, or any condition that could interfere with trial participation. The full protocol is provided in Supplemental Digital Content (SDC), Supplemental Digital Content 2, http://links.lww.com/JS9/A570. Each patient signed informed consent before participating in this trial.

Treatment
As shown in the trial flowchart (Fig. 1), the enrolled patients received two cycles of NAICT (21 day each cycle), including intravenous albumin paclitaxel (260 mg/m 2 ), cisplatin (75 mg/m 2 ), and toripalimab (240 mg) in sequence on day 1 of each cycle. Dose discontinuations, interruptions, and modifications of NAICT were permitted when hyper progressive disease (PD) or unacceptable toxic reactions occurred. Radical surgery was performed within 2 weeks after two cycles of NAICT. The surgical scope was determined by the size of lesions at baseline according to previous neoadjuvant trials [23,24] . The pre-NAICT tumor bed was recorded by photographing and radiographic examination. The initial primary lesion margins (both the longest and shortest axes) were marked by at least four points with tattoo ink, which were 0.5 cm away from the palpable lesion margins. Even if the post-NAICT tumor shrank, the resection range was determined by the pre-NAICT tumor bed, so surgical margins were 0.5-1.0 cm away from the markers; initial intraoral photographs and imaging findings

HIGHLIGHTS
• Neoadjuvant immunochemotherapy (NAICT) in patients with locally advanced resectable oral cancer was feasible and well tolerated, with no new safety signals. • The completion rates of NAICT and R0 resection were 100%. • The major pathological response rate was 60% and 2-year overall survival was 95%. • The density of tertiary lymphatic structure in post-NAICT tumor samples predicted the pathological response to NAICT.

Assessments
Pathological response to NAICT of primary tumors were assessed on hematoxylin and eosin (HE) staining, and it was showed with the proportion of patients with a MPR, including pCR, which was defined as the absence of any tumor cells [8] . Post-NAICT pathological response pattern was described. Taking breast cancer as a reference, pathological response patterns to neoadjuvant therapy were classified as unifocal or multifocal regression [25] . The detailed was shown in SDC, Additional Study Methods. Radiographical response to NAICT was performed according to the RECIST v1.1 [26] . Neoadjuvant and adjuvant therapy-related adverse events (AEs) were recorded throughout the study and graded per the NCI-CTCAE v5.0 [27] . Surgeryrelated complications were graded per the Clavien-Dindo Classification [28] .

Study endpoints
The primary endpoints were safety and MPR. The secondary endpoints were the 2-year overall survival (OS; time from enrollment to death from any cause) and disease-free survival [(DFS); time from enrollment to pathologically confirmed disease recurrence/metastasis or death from any cause].

Biomarker analyses
Biomarker analyses were performed to evaluate pathological response according to tumor PD-L1 expression, tumor mutational burden (TMB), and tumor-infiltrating lymphocytes (TILs) in the tumor immune microenvironment (TIME). PD-L1 expression in tumor biopsy samples was detected and evaluated by certified pathologists using IHC staining with the 22C3 pharmDx assay. The combined positive score (CPS) was defined as the total number of PD-L1-stained cells (including tumor cells, tumor-associated lymphocytes, and macrophages) divided by the total viable tumor cells multiplied by 100. TMB and genetic analysis were evaluated by targeted next generation sequencing. Multiplex immunofluorescence (MIF) was performed to evaluate TILs. Further details on the biomarker assays and analysis are outlined in SDC, Additional Study Methods.

Statistical analysis
A sample size of 20 evaluable patients was required to achieve 90% power to detect an increase in the MPR rate from 7% (anti-PD-1 monotherapy, on the basis of data from pembrolizumab and nivolumab monotherapy) to 30% [15,17] , with a one-sided exact test and a significance level of 0.05. For descriptive analyses, categorical data are expressed as numbers and percentages. Time to-event analyses were summarized using Kaplan-Meier methodology. Comparisons of biomarker subgroups were performed using a χ 2 , Fisher exact, or unpaired Mann-Whitney/Wilcoxon rank-sum tests.

Patients and treatment delivery
Between November 2020 and April 2021, 25 patients were screened, among whom 20 were enrolled in this trial. There were 16 patients at clinical stage III and four patients at stage IVA ( Table 1). The median age of patients was 54 years. The tongue was the most common primary site (detailed in SDC, Supplementary Table 1, Supplemental Digital Content 2, http:// links.lww.com/JS9/A570).
All patients received two cycles of NAICT and subsequent radical surgery. The completion rate for NAICT and surgery was 100%. Three patients postponed the second cycle for no more than 3 days because of COVID-19 epidemic control. During NAICT, no hyper PD or unacceptable toxic reactions occurred. Eighteen patients received radical surgery on time, while one patient delayed surgery for one day because of insurance, and one patient delayed surgery for 6 days because of personal preference. The median time from initiating NAICT to surgery was

Efficacy of NAICT
All 20 patients were evaluable for pathological and radiographical response to NAICT (patient No. 10 was taken as an example in Fig. 2A-C) Figure 2D. Pathological and radiographical responses were significantly consistent in linear regression analysis (P < 0.001, Fig. 2E).
Regarding regional lymph nodes (LNs), pathological regression was found in four of six patients (11 metastatic nodes in total) and was characterized by the presence of necrosis, multinucleated giant cells, and calcification. Only two nodes achieved MPR (including one at pCR level), and these occurred in one patient (No. 16 patient) who achieved pCR in the primary tumor (SDC, Supplementary  In the MPR group, pathological tumor bed regression led to large areas without residual tumor cells, and microscopic residual foci could be found deep inside the lesion and even close to the margin.

Biomarker analysis
NGS was performed in 19 patients. The most frequently mutated gene was TP53 (16/19, 84%). Copy number gain of CRKL, CCND1, and FGF19/3/4 and copy number loss of FAT1 only occurred in the non-MPR group. Copy number loss of CDKN2B only occurred in the MPR group (Fig. 3A). Patients with higher TMB tended to achieve MPR, but the difference in TMB between the MPR and non-MPR groups was not significant (P = 0.259, Fig. 3B). The relationship between CPS and the pathological response to NAICT was analyzed. Based on the different CPS cutoffs (1,10), all four patients with CPS greater than 10 achieved MPR, and patients in the CPS greater than or equal to 1 group had a higher MPR rate (6/8, 75%) than those in the CPS less than 1 group (6/12, 50%), but the difference between the two groups was not statistically significant (P = 0.423, Fig. 3C).
According to MIF for TILs staining, the density and spatial distribution of TILs in the TIME were analyzed. More tertiary lymphatic structure (TLS) was found in post-NAICT samples than in pre-NAICT samples (P = 0.0028), and the difference was significant in the MPR group (P = 0.009) but not in the non-MPR group (P = 0.2) (Fig. 3D, E). In pre-NAICT samples, compared with the non-MPR group, tissues from the MPR group were infiltrated with more CD8 + (P = 0.054), CD68 + CD163 − (P = 0.09), cells in the tumor region and CD20 + (P = 0.059) and CD3+ (P = 0.098) cells in the stroma region. (SDC, Supplementary Figure 6, Supplemental Digital Content 2, http:// links.lww.com/JS9/A570).

Outcomes
As of February 2023, all patients had been followed up for between 23 and 27 months, with a mean follow-up time of 24 months. One patient died of acute hemorrhage of the upper digestive tract 4 months after treatment; the death was assessed as unrelated to OSCC. Lymph node metastasis occurred in one patient 8 months after treatment. One patient experienced esophageal squamous cell cancer 1-year after treatment and was diagnosed with a second primary cancer. To date, the estimated 2-year OS rate is 95%, and the DFS rate is 90%. The OS curve and DFS curve are shown in SDC, Supplementary Figure 7, Supplemental Digital Content 2, http://links.lww.com/JS9/A570.

Discussion
The results in our trial demonstrate that NAICT with the TTP protocol is well tolerated with no unexpected treatment-related toxicities in LAROSCC. The completion rate of NAICT and subsequent surgery was 100%, and the pathological response to NAICT was satisfactory with a 60% MPR.
The safety of neoadjuvant therapy and whether it will delay the subsequent surgery have always been a concern for clinicians. The incidence of grade 3/4 AEs in neoadjuvant chemotherapy with the TPF regimen ranges from 9 to 38% [22,28] , and these AEs sometimes delay surgical treatment. In our trial, only three patients (15%) suffered grade 3/4 neoadjuvant therapy-related AEs, and no AEs led to discontinuation of the whole study, dose reduction, surgery delay, or death. In contrast, the all-cause grade 3 or higher toxicity rate was higher in the KEYNOTE-048 trial with immunochemotherapy in R/M HNSCC [13] . We propose that the patient's physical condition is more tolerant to treatment in the neoadjuvant setting than in the adjuvant setting. Although surgery-related AEs were observed in this study, these were considered unrelated to neoadjuvant therapy; however, trials with larger sample sizes are necessary to definitively indicate the safety of NAICT and its effect on surgery.
As a better pathological response to neoadjuvant therapy predicts a better prognosis, the rate of MPR has been proposed to be a crucial criterion for clinical trials of neoadjuvant therapy in OSCC or HNSCC [11] . The neoadjuvant TTP protocol has shown promising MPR rates (60%), compared with neoadjuvant chemotherapy with PF (33%) or TPF (27.7%) in OSCC [23,29] , and compared with pembrolizumab monotherapy (6%) and nivolumab monotherapy (8%) in HNSCC [15][16][17] . In our previous study of neoadjuvant therapy using a chemo-free combination of camrelizumab and apatinib in OSCC, the MPR rate was 40% [24] . In contrast, the pathological response to NAICT in LAROSCC per our trial was satisfactory, although cross-study comparisons should be made with caution.
The inconsistent pathological response to NAICT between the primary sites and metastatic LNs is observed in our trial. The MPR rate of the primary sites is as high as 60%; however, that of the LNs is much lower (one in six patients; two in 11 nodes). In a pooled analysis of lung cancer with NAICT, the pCR rate is 40.1% in the primary sites and 62.5% in the metastatic LNs [30] . A neoadjuvant anti-PD-1 therapy in HNSCC also reported a better pathological response in the LNs than in the primary sites [31] . In comparison, the small sample size prevents us from obtaining a definitive conclusion. Besides, the assessment procedures of immune-related pathological response in LNs have not been well defined, and the evaluation of the exact percentage of viable tumor cells in LNs is scarcely reported. Therefore, more attention should be paid on the pathological response in LNs for neoadjuvant treatment studies. Different degrees of primary tumor regression have been reported in LAROSCC treated with neoadjuvant therapy, but recommendations for surgical margins after neoadjuvant therapy remain ambiguous and lack prospective evidence. In breast cancer with neoadjuvant therapy, breast conservation is realized among those who are initially breast-conserving surgeryineligible and downstaged to be breast-conserving surgeryeligible; however, clinical evidence shows that a multifocal pattern of residual disease after neoadjuvant therapy predicts a higher rate of local-regional recurrence [25,32] . The pathological regression pattern of neoadjuvant therapy should be described in this regard. In our trial, the main pathological regression pattern was multifocal regression, in 92.9% (13/14) of patients; meanwhile, unit residual foci were found only in one non-MPR patient. All patients in the MPR group (except those with pCR) showed a multifocal regression pattern, and in some patients, residual tumor microfoci were found away from the main tumor epicenter, even near the tumor margins. The primary tumor looked melted in situ after NAICT. In this condition, the narrowing scope of tumor resection carries the risk of residual tumor cells after NAICT in OSCC. And more clinical evidence is urgently needed to determine the scope of tumor resection after neoadjuvant therapy in OSCC.
Identifying response biomarkers is important for selecting the appropriate candidates for neoadjuvant therapy and avoiding unnecessary treatment-related AEs in patients who cannot benefit. By comparing pre-NAICT and post-NAICT TIME characteristics, we found that post-NAICT tumor samples showed more TLS than pre-NAICT samples, and the difference was more significant in the MPR group. Previous reports have described the correlation between the presence of TLS and the clinical benefit of immunotherapy [33,34] , and we preliminarily explored whether TLS formation may predict the pathological response to NAICT in LAOSCC. From previous anti-PD-1 trials, there is evidence that PD-L1 expression levels in the primary tumor may indicate clinical benefit. This led to the 2019 approval of pembrolizumab as first-line therapy in patients with HNSCC whose tumors had CPS greater than or equal to 1 [13,35] . In our trial, all the four patients with CPS greater than 10 achieved MPR. Although patients with CPS greater than or equal to 1 had a higher MPR rate than those with CPS less than 1, the difference between them was not statistically significant. Other biomarkers validated in solid tumor, including TMB and TIME [36,37] , were also tested in our trial, but we found insufficient evidence.
In conclusion, NAICT with the TTP protocol in LAROSCC is well tolerated with no obstruction on subsequent surgery, and the MPR rate is promising in terms of pathological response. Although the limited sample size precludes exploring predictive biomarkers, the satisfying safety and pathological response results suggest that further randomized trials with NAICT in LAROSCC are well recommended.
Ethical approval